Chinese factories are making cheap voice regocnition chips, different models and types, for phones and other electronic equipment. These are very cheap, for bulk orders price is less than one dollar in most cases (or as little as 0,1 dollar for large order). These cheap voice recognition chips are suitable for cheap phones in development countries for people who cannot read or write. If cheap voice synth chips that imitate human voice also exists those chips can also be included in cheap phones. But only Yamaha NSX-39 is known to me so possible softsynth that uses phone
s memory and central processor is best for human voice synth. And using phone as musical instrument: earlier in the 2000s very small memory/ processor requirement music softsynts for ringtones etc. were offered for phones (few dozen kilobytes of memory was needed for music softsynth). Now these (dozens different existed about 10 - 15 years ago) are obsolete and not used anymore. These mobile phone music synths with low processor and memory footprint can be used as musical instruments in cheap phones. Nowadays voice recognition chips for phones (Cyberworkshop Hong Kong), video processors accelerators for phones (FTDI) and screen control chips for phones (Sensory.com) all have their own music synth included in their chips (microcontrollers inside the phone). The special music synth chips for phones used 10 years ago have become obsolete and not used anymore, only Yamaha SMAF and Dream (before Atmel SAM) exist still, others are not in production anymore (there were about dozen manufactures, including NEC, and others were mainly chinese factories like Vimicro, and Oki, which is japenese factory, etc.). These obsolete synth sound chips can be used still in very cheap mobile phones, not for ringtones but as general musical instrument. And for more expensive phones, if all three microcontrollers that phone have (screen control, video accelerator and touchscreen control) have own music synth, and if Yamaha or Dream chip is also included, that makes maximum of four different hardware synths in one mobile phone, (not counting possible softsynths in phone´s memory). If the MP3 player of the phone is constructed like in the paper “Design for a scalable polyphony - MIDI synthesizer for a low cost DSP” by Antti Huovilainen in 2010, so that MP3 player is also a wavetable music synth, that rises hardware synths inside the phone to total of five. These maximum of five hardware synths will propably turn expensive mobile phone a powerful music making machine, if all five are being used simultaneusly. And about cheap musical instrument: chinese factories make cheap toy keyboards that are priced from 2-3 dollars (cheapest models), about 37 key- instrument, to about 5-6 to 10 dollars and little more (49 to 54 and more keys keyboard). These toy keyboards look like professional synths of the 80
s, but inside they have electronics of toy musical instrument. Even touch response keyboards (polyphonic aftertouch) are for sale beginning about 20 - 30 dollars. Most of those cheap chinese keyboards (there are dozens or hundreads manufacturers and about one thousand or more different models) have USB MIDI connection. These prices are in alibaba.com and other chinese netshops for factory order of 1 - 100 pieces, or several hundread pieces factory order. At same time in china is for sale old soundcards (new factory production of obsolete electronics) about 20 year old technology, Sound Blaster etc., with their synth sound chips that are old Ensoniq, EMU, Creative, Creamware, Crystal Semiconductor CS9236, Yamaha, and Roland LA32, even old Casio (NEC) sound chips from 20 year ago or more (in Casio synths of the era) are still in production in China, or production will began if they are ordered. Mainly PC sound card chips and old style sound cards are found but also old synth keyboard chips (20 years or more old models) are offered. Old sound cards with their music synth chips have price range starting from 1- 2 dolllars upward for large factory order, and separate old synth chips are in the same 1-5 dollar price region. Why not using old soundcards / chips inside those professional synth looking toy electric keybooards, and turn those keyboards to real pro musical instruments, that has only 5- 10 dollar factory price at cheapest. Even models with working microphone are offered at 5 - 10 dollars so sound source sampling through mike is possible and keyboards with USB MIDI connection have price range of little over 10 dollars to starting at 20 dollars prices. So USB MIDI keyboard with working microphone, or polyphonic aftertouch is possible at about 10 - 20 dollar factory price. Inside those cheap keyboards would be old PC sound card, or old synth chip with associate DAC chip etc, or synth- on - a chip (all in one chip) modern Yamaha SMAF or Dream chip, whose prices are also low. Also microcontreller based solutions like Arduino, Raspberry Pi and PIC have gainned popularity, but when these designs are sold commercially their prices are over 100 dollars per piece for music synth application. Altough they are based on very low price microcontrollers they are manufactured small batches handmade, not in mass production. Instead of small batches of handmade pieces if they would be made mass production techniques and sold to thousands or million pieces in some bazaar in poor country their price would be only few dollars that is close to their bill of material costs. So if the open source community manufacturers would concentrate on mass production (using some chinese OEM factory) instead of handmade small batches these otherwise cheap synths, they could sell millions of them in development countries. Some solutions that are primitive Pico microcontreller based could be grouped to one PCB board (16 - 32 microcontrollers) and associate electronics and sold them as one piece of equipment music synth with 16 - 32 synth engines. Same perhaps is possible Atmega 2560, Raspberry Pi 3 and more powerful PIC microcontrollers. The C.H.I.P microcomputer also has some music applications. Also printed electronics is coming, one raport states that all investments to printed electronics China has 12% and India 6% stake. In india are firms like Keetronics that already offer printed batteries and displays. In Youtube there is video clips of "midi paper", midi controller made of paper, and "piano paper box" by Catarina Mota , a keyboard controller. In Netherlands Macdonalds offers free midi controller with meal, this "MacTrax midi controller" is manufactured using conductive ink, and is connected to smartphone (or feature phone). This is offered in industrial country (Netherlands), not in some poor country where it would be instant hit product, altough it would have be sold at some small price (and not given away free). If printed electronics is used in something like telephone, using simple solutions, perhaps ARM0 processor cores (less than 12 000 logic gates), and then grouping them to 100 core (EZchips) or more clusters like ARM MPcore, if working processor is built using printed electronics. Micro SD cards are so cheap that they can be used as mass memory, batteries and display is printed electronics too and radio frequency communacation using Internet of things low data rate protocols (with substantial data compression) if telephone is to be made using printed electronics (roll type "paper printing manufacturing" plastic electronics). If printed electronics is used for musical instrument manufacturing, old digital synth chips like Yamaha Dx7 and Casio/NEC, old Yamaha PC sound card synth chips (1980s) etc. are candidates for first products. If analogue electronics is made using roll printing then old Curtis and SSM sound chips in analogue synths are candidates for first products.
Chinese factories are making cheap voice regocnition chips, different models and types, for phones and other electronic equipment. These are very cheap, for bulk orders price is less than one dollar in most cases (or as little as 0,1 dollar for large order). These cheap voice recognition chips are suitable for cheap phones in development countries for people who cannot read or write. If cheap voice synth chips that imitate human voice also exists those chips can also be included in cheap phones. But only Yamaha NSX-39 is known to me so possible softsynth that uses phone
An example of useful technologies that toll free internet might use is Inetsat and Clinixon, both from Uruguay. Inetsat transfers TV content at price of only fraction of satellite transmisson worldwide and Clinixon is health monitoring app for medical patients because Uruguay have in countryside hospitals far in between. These are examples of locally developed third world solutions developed at third world and designed for their own local needs, not outside developed and then imported to some underdeveloped country internet or technology solutions that are around the world today. Also in the USA is a new internet app called “puggle.me”, it is in Github so it is perhaps in design stage. Its purpose is to minimize internet costs when using internet browsing by choosing cheapest option for search results. In most minimal form it send search results to mobile phone using SMS text messages to minimize cost. It uses Bing as as a search engine and Phantom JS as web page scanner and special database cache to improve efficiency. That puggle.me technology is most suitable for development countries and their internet, because cheapest technological solutions for end user are needed there, and in internet technology paricularly.
And about musical instruments: Yamaha sound chips in their modern LSI sound generator line have digital single chip amplifiers that have maximum output of 2x30W, and some models like YDA 156 and YDA 174 have inbuilt synth engine and DSP too with equalizer and mixer. These models can be used as single chip keyboards and amplifiers, keybord could have equalizer and mixer in the keyboard layout together with normal keyboard functions. If the same keybord is fitted with loudspeakers, either electrostatic thin type speaker (or two speakers for stereo output) or 1-2 normal box- like loudspeakers, with 2x10W - 2x30W max power or more (so the speaker won
t get sound distortion if the amp is in its maximum setting when the speaker has power requirement higher than amp) the keyboard can be used not only keyboard synth, but also as a live mixing console and loudspeaker combo, so performing band doesnt have to buy separate live mixer console and loudspeakers. The keyboard can be then used as guitar amp etc., and as guitar synthesizer etc. All those functions are in one chip, whose price is low. When electrostatic thin speakers are used the keyboard look like a laptop PC, the cover of the instrument is the speaker and turned towards audience when played. There are other Yamaha chips with built-in speakers, including the whole mobile phone sound synth chip product line (YMW line, and YMW 820 etc). These low power speakers inside synth chip could be used in super cheap keyboards like chinese toy keaboards with 37 key layout etc. They have inbuilt speaker and opamp in the sound chip itself so no speaker is needed in the keyboard. Due to low sound volume of these in-chip loudspeakers sound chips sould be placed near the surface of the keyboard and “horn” acoustic structure that is common in discant section of ordinary “box” loudspeakers installed over soundchip to amplife sound. Price can be in some third world bazaar be as low as 5 - 7 dollars per one keyboard with all-in-one sound chip. If there is no chance for midi in or out connection on the keyboard, perhaps CV/gate control is possible. Model Yamaha YMF 825 has inbuilt 0,9W speaker. Also grouping these cheap synth chips more than one in one PCB board, for example if 8 or 10 chips would be inside one keyboard every finger would have its own synth engine, so pressing the separate key of the keyboard triggers separate synth engine and combined polyphony could be over 1000 voices, if Yamaha or Dream chips are used. Price would still be very low, about a couple dozen dollars for cheap chinese OEM product. For non-digital devices techniques like Portable Oscillator (PortOSC) by Petri Huhtala, Olegtron 14-bit chips driving old 12-14 bit analog/digital synth patches etc. can be used. Using old Yamaha YMF 292 chip with its flexible sound routing together with modern amp/DSP synth chip etc. And using other old chips like Roland 32, NEC uPD9990 mobile phone chips and earlier NEC series (NEC chips were used at Casio keyboards 1980s/90s), Ensoniq OTTO, EMU, Crystal Semiconductor CS9236 etc. In fact it is possible to build a"vintage synth collection" in one PCB board using those old chips together in one board, and its price will still be very low, about few dozen dollars when mass produced in large quantities, including associate DAC chips etc. required on the board. Those old chips 20 ears ago are still in production in China (or Mexico or Indonesia), as are old PC souncards that include those chips. Almost every integrated circuit that was in production 20 years ago is still in production in those three countries.Prices are very low. Most succesful keyboard in history is yamaha DX7 with over 100 000 produced. But using nowadays cheap technology is possible to build cheap keyboard with prices starting from 5 dollars upwards that give same capabilities, and sell it millions of pieces, perhaps hundreads of millions if its price is low. If Yamaha DX was success with 100 000 sold, if some keyboard is sold 100 million units, and brings to its maker one dollar per unit profit, in the end keybard maker makes exponentially more money than selling elitistic keyboards at high prices to customers (Yamaha Reface series is example of such elitism). However, selling low-end cheap products straight to bazaars of development countries in volume of millions would bring much more profit (chinese keyboard manufactures have noticed that), and selling microcontroller based synths at hundreads of euros built one piece at time handmade (that commercial Arduino and Raspberry Pi synth makers are doing) is strange when they could sell millions of the same devices at low price/ large volume in development countries and earn much more money than they are doing now. There are FPGA based designs that open source community has been doing for past 20 years, some are academic projects, other are hobby projects, and now over 100 FPGA synth designs are available. None of these over 100 different FPGA synth designs is in actual mass production. Those few that are in “production” by commercialized open source firms are handbuild one piece at time, and price is several hundread dollars per piece, altough using chinese OEM factories and mass production price would be only few dollars when using old FPGA designs 20 years ago. FPGA offers powerful sound making tool (Juergen Schumachers
s over 8000 voice synth in 96khz.org). Xilinx designs such as XC4000, XC5000 and XC6200 series are over 20 years old, so cheap chinese OEM production of these old designs is propably available, and Virtex 1 and Spartan 1 dates from november 1997 about. Most of FPGA synths are Spartan 3 based, and Spartan is older Xilinx FPGA based from the 1990s Xilinx designs. So super cheap FPGA synth that uses chinese OEM production is possible, price would be only few dollars in mass production (whole synth board PCB) when old model FPGA chips are used. Several old model FPGA chips can be grouped to one board if that is cheaper than one modern FPGA, or several old and simple FPGA “kernels” grouped to one silicon chip. There are over 100 FPGA synth designs/ concepts to choose if someone would just start making them in large scale and low price. Or using Papilio board open source FPGA, or Archipelago which is open source academic FPGA project, or Atmel FPSLIC (like Arduino) or other modern but cheap solution. Other open source FPGAs are XESS XULA-200, Beagle Bone SDR, and Numato Opsis, all three based on Xilinx and last one is from India. CubicBoard is Altera design, Ztex FPGAs are open source also, and IceStorm is Lattice design. Based on OpenRISC 1000 is Dyna Lith System INCITE FPGA, and OR32-1208 ORSoC is ASIC version of OpenRISC 1000. Opencores.org is full of different DSP designs. Most open source FPGA synths are Xilinx or Altera based, but perhaps Achronix Speedster offers more capability, such as running over 800 processor cores simultaneusly and then running 800 softsynths or synth engines simultaneusly in one chip. Kalray CPU is capable running 256 cores simultaneysly and it is cheap and power efficient chip, but very high computational power (for its cheap price). The new Mill CPU is based on ideas of old Philips Trimedia processor, and Trimedia has Linux ported open source OS (called TMlinux), and Trimedia is old processor family beginning in 1988. Hitachi SuperH CPUs are coming to complete open source public domain (no license fees anymore), SuperH4 / J Core 4 in 2016. These have super efficient code density, the same code is used on ARM microcontrollers, and SuperH was used in Korg, Yamaha and Roland synths in 1990s but as keyboard scanning etc.duties. Renesas MCU16 was used in the synths also. There is Parallax Propeller multicore open source microcontroller, and PULP (Parallel Ultra - Low Processor Platform) and its Pulpino version. STMicrolectronics has its microcontroller as open source, and its another microcontroller ST-200 VLIW has MPE VLIW processor in opencores.org and at Delft university p-VEX and r-VEX open source versions. At opencores.org is starnge “256 bit CPU” that has 32 bit instruction set but 256 bit Arithmetic Logic Unit. In opencores.org is also free implementation of PIC 5-series microcontroller, so it possible to to have it in FPGA version (many microcontrollers on one cheap chinese Xilinx clone FPGA) like ARM and Atmel microcontroller cores have softcores also. Grouping many of these softcores to one 1 dollars in one FPGA will make super cheap Arduino / Raspberry PI / PIC project “collection” in one chip. Old Texas Instruments TMS320 chips are cheap now when chinese OEM production is used. Year 2000 Korg OASYS Pc card cost 2000 dollars and OASYS keyboard 8500 dollars. Now it is possible to build keyboard that outward look like Korg OASYS using chinese OEM keyboard and inside put those 5 TMS320 DSPs that PC card version had, and sell this keyboard about the price of 50 dollars, and functionality is like OASYS PC card. Olimex also sells Rockchip and Allwinner based cheap open source development boards (ARM).Anadigm FPAA (Field Programmable Analog Array) is sold at the price of about 5 dollars a piece for 1000 piece order, and is used at Clavia Nord Modular “G3” in 2009 that never went into production together with Xilinx FPGA. Also Sonoric Instruments designed synth using Anadignm FPAAs in 2006 that went not into production either. These are real analogue (not digital) sound processings. Nord Modular prototype used them in 32 voice reroutable synth voices per FPAA and FPGA as digital oscillators. Modern open source DSP examples are freedsp.cc, OWL open source DSP, Mako DSP, FreeDSP (Merhec 2014) and “ePUMA: Embedded parallel DSP processor architechture with unique memory access”, together with its eSORT algorithm. For optimizing DSP can Goertzel algorithm be useful etc. There are multicore and microprocessor projects like OpenRISC2000, OpenSPARC, RotorCPU, Mozilla Servo / Rust, OpenAMP, OpenPITON, Manycore research framework ALMA, Skeleton programming framework SKEPU, commercial solutions like Cavium Octeon, Toppers/FMP, and Adapteva Parallela board. For DSP there is Magnolia DSP 2015, and “Implementing and optimizing of the entire system toolkit of VLIW DSP” .But question is not are they open source or not but are they cheap or expensive, and only cheapest technology is needed to cheap instruments that are to be sold in development countries. And old DSP chips from synths like Creamware / SCOPE audio Pulsar / Noah, Soundart Chameleon and Creamware/ SCOPE Audio Supercomputer that used old SHARC chips etc that are still in production in china at cheap price, modern versions of these 20 year old high end designs (Audio Supercomputer cost about 40 000 dollars and none were sold) can be built in just few dozen dollars / few hundred dollars when faithfully duplicating their original layout (old SHARC / processor chips 20 year old models etc.), because their microchips are outdated and price of those chips 20 years later is so low, still found in production in some chinese factories. And about midi controllers: 20 years ago there were concepts like Buchla Thunder/ Lightning/ Wind, and Tactex Control MTC Express, Kurzweill Expressionmate, multi - voice theremin by John Rigg (12-voice theremin), STEIM Sensorlab, EMS Soundbeam, Wave Idea Bitstream Pro, Interactive Light Body Harp and Dimension Beam (now Roland), Optivideotone by prof. Scott F. Hall, Matthews Radio Baton (Radiodrum), Interactive Entertainment Synth - a- Beam, Vivid Group Mandala VR System, Infusion i-cube (now called audiocubes), MMB Music MidiCreator, GestureCreator and SensorCreator (University of York). Now 20 years later only Audiocubes, and things like Haken Continuum are still in production, EMS Soundbeam is not used as music making anymore, and others have been forgotten. If some chinese factory makes cheap OEM versions of those old concepts, expensive 20 years ago but cheap when using todays cheap effective microprocessors, and for a price of only few dozen dollars (that is possible because those concepts are 20 year old now) that would offer cheap MIDI interfaces that are also highly developed. To poor people for musicmaking. And using cheap analogue synths: if they are being build using cheap components, they do not stay in tune, like Mexico- build Hard- Mod synths made with cheap components. In order to improve this digital oscillators are needed in cheap analogue gear. If it is possible to use only oscillators in cheap synth-on a -chip, like Dream synth chip or EMU 8030, and have oscillator Midi out connection (it is possible perhaps use these chips norman Midi out connections to only oscillators) they would provide cheap digital oscillators to cheap analogue synths, Dream chips have max 256 voices and Emu 8030 128, so they have over hundread of oscillators. If for example modular synth have maximum of 8 voices, over 100 or several hundread oscillators is huge improvement, and digital oscillators are more stable than analogue. If not PortOSC (portable oscillator) by Petri Huhtala, simplified digital oscillator / analogue synth solution, is used.
Might I suggest posting messages that are a little more concise. It seems that you have a lot you wish to talk about but when other people encounter a wall of text it can be intimidating to either read the entire post and/or respond to all the points mentioned. A blog would lend itself better to longform content.
Keep up the passion for sharing though!
If it is possible to build Korg OASYS clone for 50 - 70 dollar retail price using cheap chinese 4 x TMS320 OEM production (old model chips, about 20 year ago, still in production in China) and 1 x microcontroller CPU and cheap chinese toy keyboard / electronic organ which has touch response keyboard, then even more cheaper would be clone of Clavia Nord Modular / Micro Modular using same TMS320 chips and Olimex TMS430 board as dev board (with TMS320 chips). If Korg or Clavia themselves just would make cheap keyboards for third world markets using these principles and cheap circuits. Atmel (Arduino) has many softcore versions that can be implemented on FPGA (Xilinx etc.) Old Xilinx FPGAs are sold in china about one dollar per chip (4000 / 5000 / 6200 series) so many Arduino projects (music synths etc.) can be implemented together in one 1 dollar FPGA chip making super cheap synths and other electronic Arduino projects possible, all different projects in one chip, at least old Arduino projects about 5 - 10 year ago that require less resources (computation time, memory etc.) Earlier simple PIC microcontrollers perhaps can be in softcore too. For effective hybrid synth (digital oscillator analog filter) perhaps mutant vactrol filter will do. And synths in the midi cable like DSP plug G1 (Jan Ostman), Mixtela and Midivamp, and Roman Sowa designs that are super cheap to manufacture are suitable for third world market and are cheap microcontroller based. For synths sound frequency over 7 - 10 khz is not needed because High Frequency Huey coding and Sub Band Replication (SBR) can reproduce high frequencies in the end 2 channel stereo mix or 4 -6 channel surround mix. Nowadays more efficient than SBR or huey coding audio High Frequency Reproduction (HFR) or audio High Frequency Replication (HFR) techniques are available, several different ones, that are either extensions of previous (SBR) formats or completely new ones. The patents of Fredrik Henn and writings of Tomasz Zernicki are perhaps latest in this state -of the-art development. all those HFR techniques require minimum of 7khz frequency (or 4khz is the minimum, and 10 khz the maximum needed), so sampling frequency can be limited to 14 Khz (or 16 Khz if that is more suitable), so synths can have more voices / processing power but effect in actual sound that is heared is minimal because used sounds that synths generate are electronic and “unnatural” anyway, and all high frequencies are generated using high frequency reproduction in the final 2 - 6 channel mix. Also cheap digital synths (Arduino, Raspberry Pi) need same kind of system that is in use analogue Eurorack world, one unified standard for modular cheap synths. Eurorack synths have “synth voices”, non-modular single unit synths together with modular versions, and desktop synths have Eurorack versions for modular use. Same principle can be used in Arduino / Raspberry Pi DIY synth world, together with “desktop” versions of nowadays microcontroller based synths can be universal standard for modular rack form (or form like emSynth Miniature Synth, Patchblocks, Bastl Instruments Trinity, etc.). There are Axoloti, Arduino Uno, RCduino, Lush One, Niklas Roy Micro Monster Modular, and analogue versions like Caspar Electronics Mini Modular, but common form factor for semimodular, modular and single synth voice (like Eurorack) is missing in DIY microcontroller synth world. Like desktop synths are turned into Eurorack form, microcontreller synths can be sold as desktop form and “microrack” like Erica Synths Picorack. The form factor of Eurorack is perhaps too big for microcontroller synths, emSynth, Patchblocs etc. can be more cost effective. One attemp for digital modular is FluorumLabs SonicTrain (Artem Godin) that combine both analogue and digital signal paths. Also net- based synths or web- based synths, that are mostly free are offered in rich countries, but in poor countries, perhaps with limited graphical user interface they can be useful, less graphic keep data rates low and because they are web based music making tools they don t consume end user s resources like memory that free soft synths do. Different microcontrellers (Arduino is Atmega, Raspberry Pi is ARM, there is Intel Edison that has x86 inside etc.) propably has intreoperability problems so perhaps different microcontroller architechtures are not interoperable, but still they can be all grouped in one “rack form” if they are gonna be rack mounted, or otherwise chained like Patchblocs etc, then the chain form and standard for chaining would be universal. Communication would be digital between devices (Internet of Things communication standard, or Ethernet, or “raw digital data” etc.), using CopperLan or other cable (if cable connections are used, and CopperLan uses standard telephone cable, because distance is short, only few centimetres, perhaps 100 megabits / second is data rate). The multi- channel music (perhaps several hundread channels of synth sound) and midi data is flowing through the chained synths, or only midi data itself. If connections are direct without cable then emSynth or something like that. Even radio frequency communication (Wifi, Bluetooth, NFC, Internet of Things communication protocols etc.) can be possible between synth devices, and music itself (several hundread channels of synth sound) can be transferred between devices using radio waves. Acrobotic Sound Synthesizer Kit is using Direct Digital Synthesis for sound and microcontroller so that is the most cost effective (but not perhaps best sound quality). And using analogue electronics combined microcontroller oscillator and analogue filter is an option so perhaps room for analog circuits (and analog signal path between devices?) in the “rack mount” is perhaps suitable. There are tons of analogue electronic waste discarded each year, collecting used electronic parts and recycling to new devices both helps waste problem and brings cost of electronics down. One example is Verde Audio, making “high tech” from scrapyard parts. If someone would just began manufacturing large scale mass production using old used analogue electronic components that are otherwise electronic waste, price of devices would be very low and suitable for development countries market. Of course guarantee is impossible because old electronics can broke down anytime, but perhaps changing potentiometers for new ones (old electronics manufactured before mid 1980s broke down because of weak potentiometers) improve quality. Analogue synths, radios etc. (and digital equipment using digital old components that are otherwise going to scrap) using scrapyard components is possible to manufacture. If factory is in development country near electronic waste depot production costs would be very low. Nowadays cheap printed electronics are in use, for example Macdonalds Mactrax midi controller, EJTech (Hungary), Novalia (England), Touch Board, O.System (RCA), SketchSynth (Billy Keyes). But those printed electronic products are not designed for, nor they are sold in development countries, where they could sell them millions of pieces. In rich countries people don
t need few dollar price primitive midicontroller, so they are low interest niche products, but in poor countries cheap tech like printed electronics midi controller would be a hit product. And ideas like microtonal keyboards with almost 1000 keys manufactured using printed electronics etc. 20 years ago there was Sony "Super Bit Mapping" for increasing 16 bit precision audio to 22 bits (theoretcally) per herz, and that does not require special decoder, without decoder SBM audio sound like normal 16 bit audio. Lowest quality audio that is listenable is 6 bit, old drummachines and samplers used 6 bit audio. If AMR telephone audio codec would have 6 bit rate (now it has only 13, 14 and 16 bits), and audio is coded at SBM, quality is 12 bit dynamic range which is better than available at MP3. And it is listenable without SBM codec at 6 bit quality. AMR codec data rates drop signifigantly (from 16 to 6 bits) and more information per bandwith could be sent. It is even possible go down to "1 bit PCM" (actually it is ADPCM because only 1 bit is used, in opencores.org is 1 bit ADPCM project, and perhaps using bitsnbites.eu -netpages DDPCM Dynamic Differential PCM principles, or using Recursevely Indexed Quantizer RIQ-ADPCM, or other effective method ). Only 1 bit per herz if dithering is used in audible volume, 8 bit is maximum dither improvement in audio quality and together with SBM it is about 10 bits or 60 decibels for dynamic range and noise distortion range, still near MP3 maximum quality. Using only 1 bit ADPCM for each frequency herz. But now only decoder with SBM will make audio listenable. SBM patents have already expired. Perhaps even lower frequencies (1,2 Khz to 7 Khz maximum) is possible, high frequency replication is needed everything over 1,2 - 7 khz, and perhaps high pass filter for quantization noise even in 1,2 - 7 khz frequencies. And using high frequency replication: Some methods start at 4 khz, so only 8 khz sampling rate / highest frequency 4 khz is needed, over that everyhing is reconstructed with high frequency replication. Because of quantization noise, only 3,5 - 3,7 khz is usable, not 4 khz.Sound quality is lower in 4khz than 7- 8 khz frequency limit, but acceptable because minimal data rates are needed in toll free internet. Synths make electronic sounds so perhaps only 3,6 - 4Khz khz is needed for their sound generating (synth engines) and everything over 4 khz is reconstructed using high frequency replication in the final mix (2 - 6 channels). Even 2Khz is possible in some high frequently replication schemes, it makes 4 khz of the highest frequency replicated, that is enough for most musical notes, so cheap synth on a chip solutions could be more simple. Also bitrate could be dropped from 16 bit to 6 bit in the synth circuit, the final DSP that makes the 2 -6 channel final mix increases bitdepth like some high tech audio DSPs do, from 6 to 15 bits or from 6 to 24 bits, the increased dynamic range of these upsamplings is perhaps 12 bits in the final audio from 6 bits. Also using aural exciter or harmonizer in the final mix, those circuits are relatively simple and can add even more frequency on the top of the already replicated (doubled) frequency range of high frequency replication. Synths using only aural exciter circuits are Spatial Aural Exciter by Andy Kramer and EXCTER by Jari Suominen. And using EML Polybox (or RSF Blackbox or Ethenvar Patch Chord) to increase paraphonic polyphony in the sound so only 4- 32 voices are needed and and result is several hundread paraphonic synth voices. Not strictly coping them but using same principle for paraphonic voices. And for cheap and effective synthesis "direct digital synthesis" by Orino (writer). And using divide down oscillator circuit like old paraphonic keyboards (Polymoog) in single chip systems will improve sound of primitive synth on single chip solutions. They all can be grouped together ( high frequency replication, aural exciter, and Polybox type paraphony together with divide down oscillator type keyboard paraphonic voice, and finally from 6-8 bits voice to 15/16 - 24/32 bit voice DSP bit depth enchantment with increased header bits, sound quality is about 10 -12 bit dynamic range if 6 bit is used in the beginning. If synth chip uses ADPCM like some or most single chip synths can do, then using 1 bit ADPCM for sound generation and then perhaps in the DAC stage adding digital dither to 1 bit ADPCM and expanding precision to 4 - 8 bit ADPCM using "header" that is 3- 7 bits long) all these techniques only in the final 2-channel mix (or mono or 5-6 channel surround sound) . So actual synth voice can operate low bitdepth and low sample rate. These techniques can be used in speech voice compression and music compression via mobile phones also. Polyglot speech synthesis can be utilised in the speech synthesizer if the phone have one. So synth in the phone can now generate more voices or effects. For telephones methods such as ADPCM and logarithmic (4 bits logarithmic equals to 7 bits linear) are more efficient than ordinary PCM, but AMR codec operates only in PCM and only in high bitrates. RIQ-ADPCM or methods like ITU G.711 standard that uses logarithmic u- and a- law algorithms (compressing 14 bits to 8 bits, if 4 bits logarithmic is used that is not in the standard, 7 bits precision is possible) and then using G.727 type compression the final bitstream is, 3bits or 2 or 1 bits bits (if in the beginning is used 4 bit logarithmic not 8) using "dropping insignificant bit" scheme. Adding dither and other sophisticated noise shaping sound quality should be good but data rate very low. And all this using old simple telephone standard techiniques. For simple low bitrate voice / music codec for toll free internet and simple devices made with printed electronics etc. Most comprehensive compression is in the MPEG4 standard, it has inbuilt softsynths / soundfont synths/ hardware synths for sound and music, and uses TwinVQ compression which is best available slow bitrate audio compression. But it is complex, requires license and is used very little nowadays. However TwinVQ is best low bitrate comperession for audio, and is dated from about 1995 so it is license free technique, altough different codecs that use it are not yet license free (they are from 1997 - 1999 about). Some really cheap music synths are being sold to Apple (iOS) phones, however they require iPhone or iPad to work, that costs hundreads of dollars. Building hardware (keyboard and cheap ARM processor with memory chips) console to iOS apps like Peter Vogel Fairlight or Wolfgang Palm and Waldorf iOS apps, these cosoles can be made very cheaply and outward look of old Fairlight or PPG Wave, but price is only few dozen dollars when manufactured in china. Problem is that iOS is Apple product. If however all graphical user interface and software connections out are eliminated, so that only bare hardware (the music synth itself) is functional perhaps the problem with the license of Apples OS can be solved, and because these synths operate only synth itself and not operate any other software form, interface is only though hardware console and not throuh software, cheap hardware synth using cheap iOS app is perhaps possible without license payments for OS. Softsynths are cheap so making hardware consoles that mimic for example Moog System 15 outward look that cost 10 000 dollars, but in fact use Moog iOS app (30 dollar price) virtual analogue or Arturia Modular V instead, these cheap hardware consoles could be made for other soft synths also, like Native Instruments Reaktor etc. These minimal price hardware (virtual) modular synths and other would have market place in third world. DiscoDSP imitates Clavia Nord Lead and costs 150 dillars, real Nord Lead 2200 dollars. If cheap keyboard with DiscoDSP inside is made “poor man`s Nord Lead” is reality etc. The Plugiator tabletop synth was partly product from India, and in Pakistan there is “Pakistan syntheziser company” soft synth firm, so music electronic industry in third world countries exists. If Windows Phone 10 can operate Windows PC programs, and Windows Phone 10 OS is now free, very large tablet PC with large display can then run Windows PC softsynth programs theoretically using Windows Phone as OS that is free from payment. Earlier phones used synth sound chips from Vimicro, Rohm, Oki, Crystal Semiconductor CS9236, Atmel SAM or Macronix synth sound chip for phones, and Micronas 3151G, nowadays only Yamaha and Dream chips remain in production. If super cheap phone includes one of these discontinued wavetable chips that are cheaply made and old already, synth engine inside every cheap phone ispossible, and if several chips (for example Yamaha and Dream together) inside cheap tablet PC or laptop, cheap music making PC that not necessirely even needs soft synths when they have hardware synths included, is possible to build cheaply. And for hardware music synths: if it is possible to built cheap hadware controllers for soft synths then it is possible to build software controller hardware synths for analogue circuit mainly. Hardware analogue synths need much knobs and switches but if software program like Mota Volta or Expert Sleepers Silent Way handles all expect putting wire in the right place (modular analog synth that still has wire connections between modules but knobs and switches has been eliminated and operated by software on the display and keyboard controller).
Because neuromorphic chips like NeuroGrid are coming, and people wonder what to do with it, because computing with them is very complicated and totally different from von Neuman computing, one use for neuromorphic chips is perhaps muisc synth. there are examples such as Jomox Neuronium Resonator, Hartman Neuron and Dewanatron Neuronics (not even published yet). Using NeuroGrid as music synth, it is analogue circuit (or 11 circuits) manufacxtured old fashioned 180 nanometer technique, but it is analog and not digital. Due to its old fashioned fabrication, mass production of NeuroGrid would cost about 400 dollar per circuit board. And computing power is 9000 times higher than avarage PC and power consunption 2 watts. But programming enviroment is totally different than nowadys. Using neuroGrid as Hartman Neuron type music synth is one possible commercial solution for neuromorphic new chips. And about using fuzzy logic etc.: “High frequency range in low-bitrate audio coding using predictive pattern analysis”, “Video data compression using artificial neural network network” 1991, “A micropower learning vector quantization for parallel analog-to-digital data compression” A. Lubkin. This one is almost impossible to found in the internet: “Sample by sample adaptive differential vector quantization” (SADVQ, but this SADVQ is not the “serial adaptive differential vector quantization”). And how to use printed electronics: it is possible to use simple CMOS 4000 series (like 4060, one 4060 is enough for primitive “music synth”, or 4106, or XR2206 circuit). Manufacturing them using printed electronics is not difficult task, and then coupling them many in one circuit board for multivoice synth that cost almost nothing (can be given away for free in development countries if commercial sponsor is found, for example printing advertisements on the surface of the keyboards, advertisement money pays the musical instrument and it is free for end user). Also because analog electronics can be build using printed electronics (roll printing mass production) too, why not building some simple like Minimoog on printed electronics, or one chip integrated circuit version of Minimoog (like Anadigm FPAA). or Jen 1000X, or Unisono-83 (rare checkoslovakian synth that has simple circuits) in printed electronics form that cost almost nothing, or from the netpage karmetik.com (artist community using electronics) Jim Murphy
s Gemini modular synth that uses minimized amount of circuits, or Castle Roctronics modular CMOS synth. All these in simple printed electronics, roll printing, cost is almost nothing. For compressing any audio sound, such as music codecs like MP3: efficient methods for roll printed circuits and other are direct digital synthesis by Orino 2005, patents by Ossi Kalevo using "ODelta" encocoding and modulation method, and pyramid vector quantization, cubic vector quantization, spherical vector quantization etc. used with or without ADPCM, or RIQ-ADPCM (Recursively indexed quantizer). And multipre description coding,and multiple description delta-sigma modulation / coding. For low bitrate audio there is "Session P26 low bit-rate audio coding" conference papers (aes.org/events ,2008), "A complex envelope sinusoidal model for audio coding" (Bartkowiak), " A simple adaptive matrixing scheme for efficient coding of stereo sound" Bartkowiak, "Low complexity parametric stereo coding" Schuyjers 2004. if very low bitrate ADPCM or delta-sigma modulation is used (only 1 bit for each herz, no oversampling ratio over 1 even in delta-sigma), then perhaps both low pass and high pass filters must be used on audio stream to bring down quantization noise. Limiting frequency to 2 / 4 khz and using most efficient High Freqiency Replication, together with aural exciter on high frequencies that are top of the 4 - 8 khz range available in HFR. Floating point numbers cant go below 12 -13 bits before they lose precision. 8 bits is available using logarithmic u-law and a-law scale, precision 13 - 14 bits. But minimal logarithmic scale that has some accuracy is 4bits log, that has precision of 7 bits or values between 1-100 in decimal system about. How can this accuracy is possible to maintain in 1 bit system? Perhaps using delta modulation together with this 4 bit logarithmic, delta modulation brings bitrate down to 1 bit (1/4 from original) without significal precision loss. There are old papers like “Adaptive rate delta modulation for non-synhronous, full duplex, digital communications” B. A. Harvey, " A new adaptive delta modulation system" H. J. Kim 1975 about, “Constant factor delta modulation - a new instantenously adpative delta modulation system” Kyaw Aret 1973 (C.F.D.M) , “Adaptive delta modulation for improved voice quality” 19.3.2008 CML Microcircuits at dataweek.co.nz netpage, " A 2-bit adaptive delta modulation system with improved performance", National Semiconductor and their “Digitalker” delta modulation for phones, " A control-theoretic approach for efficient design of filters in DAC and digital amplifiers" Tsakalis 2010,“A modulater hybrid filter bank for wide-band analog-to-digital converters” 2014 Chunyan. Delta-sigma modulation can also in parallel processing version (“Single bit pseudo parallel processing low oversampling” Hatami) and its improved version (Johansson, Svensson 2014 " A novel speculative…") can be effective, but complicated. There are efficient noise shaping methods for delta-sigma modulation that have 50 decibel (8bit) surface to noise and dynamic range and 70 decibels range (12 bits) if dithering noise becames audible. So not even oversampling ratio over 1 is needed in delta sigma modulation to decent quality audio codec, using only 1 bits. And if distributed parallel model is used perhaps bitrates as low as 1/16 bits per hertz (or 1 bit for 16 herz) or 1 bit for 32 herz (if parallel distributed delta-sigma process is 32 to 1 ratio) in delta-sigma modulation. And that could even compressed even further perhaps using quadrature mirror filters and other methods that Sony ATRAC and SACD is using. ATRAC promises at 48 kilobits/ second “CD quality sound” using 16 bits PCM. If one bit delta-sigma is used and oversampling ratio of 1 compression is near to 3 kilobits/sec and distributed model (Hatami and others) perhaps only 0,1 kilobits / sec for CD quality sound. If 48 khz sampled / that has 24 khz audio range is using 60% (14,4 khz) for actual audio and 40% (9,6 khz, rest of available 24 khz audio spectrum) for quantization noise that is “almost oversampling ratio 2” audio codec for delta-sigma modulation. If frequency is only 0- 2000Hz, higher range is artificially manufactured using high frequency reconstruction, SBR that MPEG uses creates high frequencies sliding between 4 khz and 10 khz base, creating 16 khz sound. Lowest possible replication range in some models is 2khz, creating 4 khz sound. But that is sliding like SBR and only minimum, perhaps even 7khz tones is possible in only 2khz base sometimes. So using 2khz base (4,4 - 4,8 khz sampling rate, over 4/2khz is for quantization noise) 4 - 7 khz is achiveable, over that 4 - 7 khz frequency using aural exciter even more high frequency sound can be generated, altough they are just hazy harmonics. If now this 2 khz sound is using 1 bit ADPCM, 4 bit logarithmic scrutinized for 1 bit, or delta-sigma modulation, only 1 bit for herz is needed. That sound can be compressed even further using parametric stereo (2 channels to 1) and then all those compression methods that modern audio compression uses (quadrature mirror filters, MDCT, vector quantization, perceptual coding, adaptive bitrate, adaptive quantization step etc) and scrutinize that audio about 10 times smaller still (or even 15 - 30 times smaller, like most effiicent audio codecs can do with raw data stream). Result is 0,5 kilobits mono/parametric stereo and 1 kilobits stereo audio stream. perhaps 64 bits /sec is reserved in parametric stereo model for 2 channel coding “side information” and 128 bits in stereo stream for high frequency “side information” so no “blind source encoding” is needed. That 64 / 128 bits is propably packed like other audio so actual information density is 640 or 1280 bits /sec for side information. When that 1 bit audio enters in the DAC it is enchanted using additional header (ADPCM can be increased from 1 bit to 4 or 8 bit ADPCM, (1 bit actual audio,rest is header), and 1 bit log delta transferred first to 4 bit log, then 7 bit linear PCM (with dither etc. that coding uses) and finally to about 21 - 24 bit linear PCM (with everything over 7 bits is header) for improving audio quality. This is minimal data rate audio codec. In ADPCM and 4bit log 1bit delta using dither, Super Bit Mapping, extensive noise shaping etc. is used for improving sound quality, using dither in a way that it becomes audible if large dynamic range is needed. Delta-sigma already have very sophisticated and efficient noise shaping methods and low oversampling ratio versions, and parallel distributed versions for low data rate, that have theoretical background, but no real real world implementation because old Super Audio CD standard does not use them. There can be this “minimal data rate audio codec”. So first 13 bit PCM is transferred to 7 bit PCM using Super Bit Mapping or similar technology, saving 6 bits, then 7 bit PCM is transferred to 4 bit logarithmic (same accuracy), then 4 bit logarithmic to 1 bit delta (or Takis Zourntos
s "nonlinear control" model, but without oversampling) or DPCM, or other method 4 bits to one, accuracy is still almost 4 bit but data rate now only 1 bit for herz. And final accuracy is almost 13 bit, using all these methods for scrutinizing 13 bits to only 1 bit. Almost 13 bit accuracy in one bit can be improved still using additional header at DSP, 13 bit PCM + 19 bit header (1,5 times original data) = 32 bit, accuracy is now about 16 bit CD quality. All this using only 1 bit/herz. For total cost free internet, paid by advertisements and commercial sponsors, in development countries. It needs "minimal data rate video codec" too, with minimal acceptable quality (small pixel count, colour quantization logarithmic, minimal colour values per pixel, small frame rate, minimum number of keyframes, maximum compression for everything, new compression techniques for video coding has been invented lately for super-efficient coding of video). For free internet. TouchKeys by Andrew Macpherson is a method for music keyboards that is simple refit to keyboards without touch sensitivety to add it and plyphonic aftertouch. For hardware music synth using only 1 bit ADPCM and then heavy dither that is audible, 8 bits is reached improvement in accuracy, 1 bit ADPCM is acceptable sound quality now and listener propably does not even notice dither noise because electronic sounds have such humming quality anyway. Delta-sigma modulation without oversampling ratio bigger than 1 is achivied in nowadays noise shaping methods, 50 decibels signal to noise ratio and 70 decibel dynamic range but dither becames over 50 decibel audible. Mp3 maximum dynamic range is 64 decibels so this delta-sigma without oversampling is feasible. Or use Takis Zourntoss nonlinear control model.
There are even methods for go below Nyquist sampling rate, and it is possible to reconstruct the signal with just a fraction of bandwith. On paper " Sub-Nyquist sampling: bridging theory and practice" 2011 Mishali, is methods like folding Nyquist window etc. Successive Approximation DACs (SAR DACs) have promising capabilities for ultra low sampling rate. On paper “new approach based on compressive sampling for sample rate enhancememt in DASs for low-cost sensing nodes” 2014 Bonavolonta is information that only 2% of required sampling rate is needed to reconstruct the the whole frequency band, so only 200 samples from 10 000 samples is needed to reconstruct all 10 000 samples. Efficiency is enormous 50 to 1 ratio. 50 to 1 ratio below Nyquist rate is so enormously small sampling rate that even if quality of end product (audio or video) is bad, but if it is at least minimally audible or visible, it must be used in ultra low bit rate audio and video coding. And that is only for sampling rate, other compression methods can be used for on top of that and make data rate smaller still. If that sub-Nyquist sampling is possible (never mind the quality) it must be used on minimal data rate audio / video codec. Others on on the same matter: “Non-uniform sampling algorithm and architechtures” Luo 2012, “Jittered random sampling with successive approximation ADC” Luo 2014, “A bit-constrained SAR ADC for compressive acquisition of frequency sparse signals” Waters, “Calibration of sampling clock skew in high-speed, high resolution time interleaved ADCs” Kumar, “Compressive sampling with successive approximation ADC architechture” Luo, “Parallel-sampling ADC architechture for power-efficient multi-carrier system” Lind 2011, “An energy-efficient noise-shaping SAR ADC in 28Nm” 2015, “Compressive sense based reconstruction algorithm non-uniform sampling based data converter” US pat. 8547260 2013 Sestok, “A jittered-sampling correction analog-to-digital converter” Parsons 2014, “A compression sampling system based on sparse AR model” 2014 Ye, “Compressive sense based reconstruction algorithm for non-uniform sampling based data converter” patent 20130069807. If video signal is coded, using perhaps only 1 bit for grayscale per pixel with efficient dither (dither turns black and white pixels into grayscale), for colour minimal colour values, dithered logarithmic colour, many pixels share same YUV colour etc. Splitting video picture to very small subframes and coding them separately using those random sampling/ jittered sampling sub-Nyquist techniques. In netpage endlesscompression.com is examples of effective compression methods. For stereo sound is “Coding of stereo signals by a single digital delta sigma modulator” 2014 Callegari. For sound also coding 4 bit logarithmic (smallest logarithmic with significant precision, 7 bit) with delta coding, or DPCM, or ITU G.727 style “drop insignificant bit” scheme that turns 4 bit to 1 bit (using 4 to -3 - to 2 -to 1 bit bit dropping scheme), in order to maintain good precision of 4 bit logarithmic but having it in 1 bit form almost with same accuracy. Video game insdustry used for sound Xan DPCM, id RoC DPCM (RoC Multimedia) and Duck DK DPCM. For delta-sigma modulation: “A higher-order mismatch-shaping method for multi-bit sigma-delta modulators” Alexander Lavzin, “A thermometer-like mismatch shaping technique with minimumelement transition activity for multibit delta-sigma DACs” Sanyal 2014, “A new DAC mismatch shaping technique for sigma-delta modulators” Aboudina. If quantization noise is high in low sampling rates, perhaps using “multiband dynamic range compression” and “Linkway - Riley crossover filters” if they bring any help. For very low sampling rates 4,8 khz that have 2 khz for audible range and 1/6 or 0,4 khz for quantization noise if Nyquist rate is used, 1/10 of normal 48 khz sampling rate. Also 4416hz that is approximation of 44,1 khz rate 1/10, if that is divided by two and 1/8 is reserved for quantization noise, only 1932 hz is left. Human hearing cannot hear frequencies below 50 hz so perhaps 48hz - 1980hz is available, rest is reconstructured using high freqiúency replication, 4-7 khz range, and over it aural exciter brings even more high frequencies. Similarly 4,8 khz ( 2 khz) can be used using 48 - 2048 hz range. There is also “quadrature delta-sigma modulation” but its suitability for audio coding I dont know. Exotic coding/ modulation methods are “spherical fibonacci mapping”, “spherical hashing” (J. Heo), “Internal noise shaping with pulse width modulation” (Midya), “Quantization noise coupled delta-sigma ADC with delay”, “Time quantized frequency modulation with… codes” (Malcom Hawkesfrod), “Analytical… of VCO-ADC quantizer using pulse frequency modulation” Hernandez 2012, “Lossless compression for u-law and ADPCM on the basis of a fast algorithm” Ang, “ADPCM with adaptive post filtering delay tree audio coding”, “Low delay vector-quantization subband ADPCM coding” Fink 2015. On paper “32Mw 320Mhz continous-time-complex delta-sigma ADC for multimode delta-sigma wireless LAN” is something like Quadrature Amplitude Modulation coupled with delta sigma modulation. If information packing /compression capacity of QAM can be coupled with delta or delta-sigma, delta modulation efficiency can be in very high information packing level. On patent “Method and device for encoding and decoding data in unique number values” patent by Ipo Paul Willem van der Boom is old method from 1972 that compress text in number values, for text compression. Also besides video and audio compression cost free internet needs text compression, and ScreenPressor type web page graphics compression, if web browser itself does not support web graphics.compression. There is also something like “pyramid logarithmic quantization” altough it is not called by that name, different versions are “Perceptual vector quantization” in daala video codec (that one looks like quadrature amplitude modulaion at modern version, if that just could be coupled with delta coding etc. logarithmic like delta-sigma), “Piecewise-linear radial compression function for pyramid two-dimensional quantizer” 2014 Jovanovic, “Multirecognition vector quantization for video coding” Calvagno 1997, “A robust image watermarking based on gradient vector quantization and denoising using counterlet transform” Kullayamma 2014, including “logarithmic quantization index modulation LQIM”. “Harmonic vector quantization” Ericsson research, “Modelling geometric-temporal context with directional pyramid co-occurrene for recognition” Yukal, "The pyramid hierarchically contour processing " Meer 1990. There is also “Oversampling audio without delta sigma modulation based on nonlinear control” by Takis Zourntos, it is much more stable than delta-sigma modulation and therefore more effective. There are also texts by Elettra Venosa, mainly about “Time-interleaved ADCs” and other low data rate converters. There is “Simplified floating-point division and square root” by Viitanen 2011, for DSP processing.
Minimal operating system like Symbian has an advantage that because it is old, it has programs and apps that have very little memory requirement. Symbian has some 101 000 - 102 000 apps, and erliest are before year 2000 when Symbian was called EPOC. Those old small programs can be stored huge amount in one memory chip inside phone, those old programs are few dozen or few hundread kilobytes each. Games etc. and different computer programs. Because they are so small, and Symbian has lost its position as commercially succesfull product, those apps have small or nonexistent value. 10 or 100 or 1000 or 10 000 of those Symbian apps could in one memory chip for free and that memory chip is inside one cheap feature phone. The cheap phone comes preloaded with about 10 000 free apps inside. For free internet cheap phone with minimal operating system that works on low power and memory and transistor count and processor can be manufactured using printed electronics, and preloaded with some 1000 - 10 000 small memory requirement apps so phones owner does not have to scan internet to obtain them, would be perfect solution for cheap free / minimal price phone for cost free internet. Netpages like tinyapps.org that contain minimal computer programs are examples of development that is good for cost free internet. Programs are few kilobytes or few hundread kilobytes max. If processor is not manufactured using printed electronics, battery, display, keyboard etc. could be made cheaply using roll printing and “modular phone” concept that user can itself replace components on phone without discarding the whole phone could be useful, but not perhaps in the cheapest few dollar production costs basic models. Symbian fulfills those requirements for simplicity and low memory requirement apps, that Symbian has huge back catalogue with, and that huge catalogue or part of it, free apps, can be inside every Symbian phone that is offered for cost free internet. Simplest solution for ultra cheap phone is that phone has not specific operating system at all. Perhaps simple RTOS for most simple procedures but no any complicated OS. Perhaps still without operating system cheap and simple electronic games like simple electronic hand held game consoles of the 1980
s can be included on the phone, like old "worm" game of old Nokia phones etc, and simple applications and programs (mainly text-based) for the phone and for the phones user, but nothing very sophisticated is perhaps not needed anyway for ultra-cheap phones made with printed electronics etc. Ogg Opus is unified speech and audio (music) codec and it is license payment free, so AMR speech codecs and MP3 etc. music codecs can be eliminated (altough MP3, AAC, and Microsoft audio codecs are license payment free already for phones, but speech codec AMR is not). Ogg Opus already uses “high frequency huey coding”, equivalent to AAC Sub Band Replication, but other methods like “High Frequency Residue Replication” Ma 2009, are being studied. If quadrature filter banks like in AAC or TwinVQ quantization is used in Opus perhaps compression ratio becomes even better. For video compression there are license free alternatives like libde265, Google
s video compression and Daala (not ready yet), and some efficient but exotic like ScreamingHD. For coding stereo content into mono signal there are methods like "Directional Audio Coding" DirAC, and "Spatial Impulse Response Rendering" SIRR, both from Aalto Acoustics Labs ( by Ville Pulkki), these methods code in one loudspeaker stereo spatial sound using psychoacoustics, but how they work on headphone listening I dont know. And even more complicated methods like Karhunen-Loeve transform modified to multi channel sound coding exists, either 2 channels from 1 channel, or 6 channels from 2 channels coding. And in MPEG 4 standard there is Binary Format for Scene description BISF and its Advanced Audio (AABIFS) version, and MPEG 7 AABIFS also, that has virtual sound enviroment capabilities.
Previously I suggested that some 3G or 4G network phone
s data rate (2Mb to 100Mb/sec) could be split into smaller parts and using that same data rate that is now used for one one phone could be split between many users, if data rate is only 20 kilobits/sec this "virtual 2G" network can be installed for example 1000 users instead of only one user using 2Mb data rate. Because teleoperator costs drop now only 1/1000 of previous, and simultaneysly advertisers reach 1000 times more customers (1000 customers can be now reached at the same cost than one customer using 2Mbs/sec data rate, advertiser gets 1000 times more money per customer), total cost free internet for end user can now be true. If 4G/LTE network is used data rates are 20 - 100Mb/sec and splitting that in 20kilobits/second sections brings 100 - 20 000 "data users" in the place of only one customer. However it seems to me that splitting data between customers is not possible in nowadays technology. But there is solution for this: multiuser "phone cell", or multiuser content system. The radio "cell" that one customer uses in 3G or 4G cell phone network is now shared between 100 - 20 000 customers in multiuser content system. Now every message received in this radio "cell" targeted for only one phone is received to thousands of telephone / tablet PC owners. In order to keep messages private data encryption is needed: For example every receiving message has "adress code" similar to individual phone number, altough thousands of phones inside radio frequency cell receive that same message, phone hardware checks the address code in the beginning of the message, and if the address code is not the same as in the telephone, telephone does not react to that message and does not process or open the message. Information can also be encrypted, text (SMS messages and email), and voice can use "scrambler" that military radio phones use to hide its information. So altough thousands of phones receive the same message or information, only if address code is right the phone opens the information, and if encryption is used not even opening the message bring information usable to someone who is not in the same address code. The encryption key must be of course individual to every user. If encryption on the hardware cannot be used or radio communication already uses it in the 4G network, then using double cryptography that uses one code to open the message, but the opened information is encrypted using another coding method, that another code is decrypted in the software in each individual phone. I don t know how video or visual information can be encrypted. There are things like Mikogo desktop that is shared desktop / operating system. In Wikipedia is "comparison of remote desktop software". But this multiuser system is not exatly like shared remote desktop, altough it can be used like it. It can be even better than conventional internet because now one receiving message can be received between thousands of customers if needed instead of one , in the cost of one message, messages like commercial advertisements, or shared messages etc. Negative aspect is slow data rate for one customer, but using sophisticated data compression methods like Finite State Encoding (FSE), or QDigest or other it is possible to overcome this and different audio- video etc. compressions etc. Instead of 3G or 4G LTE, Wifi phones can be used. Wifi phones can be even cheaper to built than LTE phones and data rates are at 4G 100 Mb/sec to 350Mb /sec. That means 5000- 20 000 "virtual 2G" data rate channels, or 5000- 20 000 customers in inside one Wifi cell. Cheapest chinese Wifi phones start at 5 dollars for large factory order, and cheapest GSM phones about 4 dollars, altough both these cheap Wifi and GSM phones are quite primitive. Using Internet of Things protocols (some of them have megabit sec data rates, at least in"burst mode") can be used to bring costs down for both receiving device and teleoperator. On the uplink if messages that are send "crash" because there are thousands of customers in same radio cell, simply "send again" procedure and message on the phone screen resolves this problem. If user wants to share his information, song he is using or video he is downloading, customer can simply left it be not encrypted and without address code, now everybody in the same radio cell can watch the same video or hear the same song. In free internet for development countries only public domain material or other license / royalty payment free material is used, so sharing information (like song or video) is cost free in free internet for developing countries. Public domain means different things in different countries. In USA early Alfred Hithcock are publibc domain but not outside USA. In England old Michael Powell films are public domain but not outside England. In Japan old Akira Kurosawa films are public domain but not outside Japan. If now could be agreeded that if some information is in public domain in its home country (or in the USA, USA is the most used source of public domain material) that same material can be public domain if not all over the world, at least in poor development countries. Altough material is license payment free, the free internet is still funded using commercials, there will be commercial breaks like between Youtube videos, but not opportunity to skip adverts, and in audio stream adverts too. Using graphical user interface in free internet net is full of pop up window comercials. The more money from advertisers free net collects the better, enven licence payment material can then be used on free intrenet of developing countries. If Wifi and IoT is used devices can be very cheap, or even usual phone components. Maximising roll printed electronics like displays, keyboards and batteries and making rest of integrated circuits so cheap as possible in some chinese OEM factory. Printed electronics solar panel on the phone itself can be used as battery charger, or left battery out and use solar panel as only means of electric power, when sun does not shine phone does not work. Or other similar methods like using body heat as energy source etc. Some IoT low power device can work on these methods. These super cheap phones could be advertisement paid, because production costs are 1-3 dollars per phone and then distributed free, without cost to end user in free net. Even "expensive" "smart phones" and tablet PC:s that cost 10 -15 dollars to make in mass production factory price, can be adevertisement paid with commercial sponsors, phone is painted full of advertisement logos, in the hardware level there is adware etc, user must hear/ see commercials from the manufacturer all the years he uses the phone etc. If there is no way to encrypt information individually in one phone radio frequency cell, or use address code that prevent to open messages if address code is not right, then only way to use this multiuser system is that privacy must left out, every message is heard to everybody inside radio frequency cell and that is the price of cost free internet. Like using skywave radio comunication, at low 20 watts every message can be heard all over the world to everybody, that is real "multiuser system" and worldwide without teleoperator costs, because athmospheric propagation transports the message. There are several athmospheric propagation methods different wavelengths from below 1Mhz to several gigahertz, and some are being examined as means of long range Wifi (hundreads of kilometers Wifi range radio communication). These things (multiuser internet for development countries, slow data rate, many customers using same radio frequency cell simultaneysly, either for receiving data or sending) is possible today. Costs will be 1/1000 - 1/ 20 000 compared to nowadays teleoperator costs for one customer. State owned telephone firms are possible the most suitable users. Over third world almost in every country there is teleoperator that is state owned, or partially state owned, by some rich industrial country, some middle eastern super rich country, or China, or development country itself. Teleoperator business is together with weapons trade and oil business one of the most state owned businesses in the world. And this shared multiuser system will perhaps bring huge profits to teleoperators and expand their customer base exponentually to billions and billions more customers. And it will be cost free for end user. Propably private firms like Millicom that operate only in development countries can use this method too. In development countries, even in poorest ones, have today 4G networks, either Wifi or LTE, and state owned teleoperators are using them. Cheap Wifi device is easy to build, even cheaper using roll printed electronics on those parts that are possible to manufacture using it. The multiuser "virtual 2G" concept inside one 3G or 4G radio cell is capable to bring down costs to minimum and expand advertisement money to maximum. For free internet for developing countries. If 20 kilobits/sec is too slow data rate then higher rate is possible. Obviously all 1000- 10 000 users inside one cell are not simultaneysly using net so data rate in most cases can be higher. This "radio cell" is not like traditional cell phone radio cell that telephone networks uses, normally one radio cell holds thousand individual phones.this "virtual 2G" is just one telephone connection in this normal radio cell of thousands of individual phones that have 100 - 350 Mbtis/ sec data rate for one user in 4G network, But in multiuser content system this one connection of 100 Mbits/sec data rate is shared by thousands users. If 4G wifi connection has 350 Mbits/sec data rate, it is shared between 20 000 users. It is like all 20 000 have the same telephone number, because they all are using same telephone connection that is normally used by only one people (altough they dont have the same telephone number, but principle is that one teleophone connection is used by thousands of people). It is like 20 000 people using the one and only telephone at the same time, (although there are 20 000 phones, using only one telephone connection). They can
t all have same telephone number because when people move and use mobile phone they change radio cell and link tower when they move. Despite the fact that people have different telephone number if there are 20 000 people in some geographical area that fits inside one "radio frequency cell" that is normally used for only one telephone Wifi user, all received messages (downlink) will now be received together with 20 000 different cellphones, and all sended radio frequency messages (uplink) of 20 000 phones will go to one telephone connection "slot" of the link tower. The principle is same as net browser that uses only one net connection but the browser has browser subdomains, or many sub- browsers inside one main browser and now many different users can use only one net connection through these sub- browsers inside one main browser- principle (in this case "main browser" is shared between thousands of phones, there are distributed browser, or distributed operating system concepts etc., many different ones, but the phones used in this shared network can still be Symbian or Android etc. if wanted, and shared network is just simple communication protocol inside Symbian or Android phone, or go step further to complicated distributed operating system concepts, byt because simplicity is needed in free net these high-tech concepts are propably not needeed like distributed operating systems. Dooble browser itself is in its original distributed form and now a "desktop" or sort of simple operating system). Multiuser browsers with one net connection are iSpaces, iCube (previously called iRock or Rocknelt browser) etc. This is the same situation like multiuser browser inside one phone using one net connection, but not used for internet browsing only but all audio and video communication also, and those 20 000 cellphones have the same multiuser browser for one telephone connection to link tower that has 20 000 subdomains. In normal multiuser situation all messages received will automatically go to 20 000 cellphones. For privacy all messages and received information downlink have individual "address code" in the beginning of every received or send message, (text, audio or video) and now phone sees should it react to message, and open the message (right address code) or not to open the message (address code different than in on the phone), and perhaps individual encryption or double encryption for privacy. For tablet PC or laptop this same method can be used, and in the net browser browser "subdomains" that share one browser to several users, like iSpaces Cloud browser/computer. Maximum data rate is 17 kilobits/sec in this 20 000 users inside one 350Mbits /sec phone connection model. For uplink and send messages to prevent data crashes maybe "waiting on the line" model where one has to wait in the line perhaps several seconds or minutes to have free data slot to messages uplink, after pressing "send" button it takes time until phone announces "message send". Realtime data transfer like telephone conversation is difficult using this method uplink, but perhaps possible using slow dta rate and frequent "data crashes"on the line if many people are using realtime data transmission at the same time. Normally in cell phone radio frequency is shared into cells that use "time division multiplexing" and other tricks, one cell has perhaps thousands users and data rates of one user is from 2Mb (3G) to 350 Mb (4G Wifi). But now this one user slot must be divided between thousands of users, and then thousands of these slots (that normally combine only one user with hundreads of megabits data rate) combine one normal "radio frequency cell". Multiuser content model where information downlink must go to everybody in the same one user radio cell area is only possible solution. If information is individually encrypted or have some sort of identity code, altough thousands of phones receive message only one opens it. when user shifts position and enter another cell phone tower link he is in the area of another radio frequency cell and shifts now to use it and there is another 20 000 people in the same user slot that he is using. This is like normal cell phone operation but insead of one user for one telephone connection there is now 20 000 users for one Wifi telephone connection in shared multiuser model, using same frequency, same time slots etc, same time division multiplexing, same everything, sharing one telephone connection. For voice communication perhaps sharing one speech voice channel between dozens of users brings data rate savings. The cacophony of dozens of voices speaking same time would be enormous, but using "voice scrambler" code that military radio phones are using, individually coded voices can perhaps be separated from cacophony, and using "polyglot synthesis" or other to make information audible, at least speech voices of telephone conversations and internet radio speech voices. And using CELT type speech synthesis. Now several speech voice channels can now be transmissed using only one channel for voice. All this shared multiuser content system thing, one radio frequency connection normally aimed for one user shared between many people, "virtual 2G" system, can be used in 3G, 4G, 4G LTE or in Wifi networks. For voice communication, because "wait on the line" model is used, perhaps radio telephone (trunked phone network) type communication (push to talk over cellular) that is already possible in some phones, after one say something on the phone he/she presses "send" button on the screen, then voice is send using data compression, "beep" sound gives signal when data packed voice is send, if signal is not heard that is because data crashes on the upling have prevented message, pressing "send again" button and perhaps now message goes through. This kind of push to talk over cellular limits voice channels to one (instead of two) and makes voice communication possible even in "wait on the line"procedure. Radio protocols like TETRA, its british equivalent, Tetrapol or Project 25 operate at low frequency range (about 100 Mhz lowest) and does not neek link towers, because contact is direct between phones. Motorola IDEN and WiDEN and european OMA PoC are similar but inside mobile phones, and does not neet link tower either. Low frequency gives more range to phones and slower data rates, but one link tower can now cover much large geographical area. Thats why CDMA phones are recommended for development countries because their lower frequency than GSM gives them greater range and therefore less infrastructure (cell phone link towers) are needed. CDMA phones are more expensive by avarage than GSM phones, but only because they are manufactered in smaller quantities, when mass production reaches same level as GSM phones there is no difference in prices of GSM or CDMA. CDMA requires however less infrastructure costs because of greater range. If GSM frequencies go down to same leveI as CDMA, GSM can cover the same area as CDMA also. If frequencies go down to 100 Mhz which is close to TETRA etc. networks, range is even more greater and perhaps athmospheric propagation methods becomes one way of communication to over long distances, athmospheric propagation have several different areas and methods between about 90 Mhz - 430 Mhz, and even more below 90 Mhz. Chinese TD-SCDMA is a rival of western 4G netwoks, and it is specially designed not to include western licenses, so it is cheaper then western versions. Japanase iBurst is not anymore developed cell phone standard, it is capable of low data rates and low bandwidths, and it is in use globally. It has good spectral efficiency as 3G standard, in the Wikipedia “spectral efficiency” page, so it is consuming less bandwidth, but 4G standards consume even less. “LTE-Direct” is a phone standard like Motora iDEN where phones contact each other without the need of link tower, range in LTE -Direct is 500 metres. So it is possible to keep infrastructure costs down if phones don`t need link towers so often than they nowadays do and link towers can be sparsely separated. There is a text “An intelligent extracting web content agent on the internet” 2005 MIng-Huey. Using Ogg Opus as speech / music codec and Daala etc. as video codec no license payments are needed for video / audio codecs. However, if there is commercial codec that require license payment as hardware devioce, if that is more efficient than license free codec, license payment price has no signifigance because more efficient data compression helps to bring data rate price down and so it is significantly more important than one 0,5 -1 dollar etc. license price of commercial hardware codec. More efficient compression saves perhaps many dollars per month from data rate costs, and better service and content can now be offered at cost free internet.“Shruti Drishti” is a special web browser for visually impaired text-to-braille coneversion tool, developed for development countries own use. Previously chinese factories manufactured super cheap “wristwatch phones” that were just basic GSM phones without operating system (some had Symbian), phones where wristwatch type, minimal size display, WAP text browser was optional for extra price, and price for phones was 1,35 - 2,5 dollars. But mobile ineternet made these basic phones outdated, and last models were manufactured 2013 - 2014 about. SoC was Coolsand or cheapest Spreadtrum etc. no mali graphics so WAP text browser was only option. MP3 player was however included. Minimum order of quantity was 10 000 - 50 000 phones for super cheap basic phones. Text browser without graphics is still option in cost free internet, together with efficient text compression, without video/graphics processor phones can be made cheaper. However using printed electronics even 7 inch tablet PCs can be made cheaply, display is most expensive component of tablet PC, if that and keyboard (no touchscreen is needed) and battery can be made with roll printing mass production, super cheap tablet PC, that can be used like Google virtual cardboard phone glasses as 3D glasses also, and together with printed electronics made solar panel (solar panel can be on the backside of tablet PC etc.) or using body heat or other heat source as battery charger etc, and cheap lamp (roll printed manufactured) that will bring light at the evening and uses tablet PCs battery. This phone/ tablet PC / 3D or 2D glasses/ lamp combination can be made very cheaply and paid by commercial sponsors and advertisers together with adware, and internet connection (low data bitrate) will be advertisement paid too by commercial sponsors. In third world and poor countries this is possible.Or combination of primitive 2D or 3D video glasses together with “Android TV stick” style cheap Wifi computer stick, that is cheap to manufacture, but this TV stick works as phone/ tablet PC, not just a TV Wifi middleware. This “Wifi computer stick” / cheap video glasses/ headphones / lamp -combination is all that is needed for audio and video communication and light at night. Extra internet time is possible to earn using “advertisement watching work”, watching adverts for hours and earn some money or credit or internet time.
Cheapest Android smart phone in the world is at this moment Nanotel Acche Din in India, price is just 1,3 euros or 99 indian rupees. Previous Android phone record was Freedom 251, price 3 euros. Year ago in India cheapest Android phone cost 24 euros ( Forme P9). Those new phones are being sold well below manufacturing prices, realistic price for Android phone is Docoss X1 that cost 11,5 euros in India. So “real” prices of cheap Android phones have come to more than half in just one year. Cheapest 4G connection cost 2,5 dollar/month (that is not the subscription line price but the cost of whole months use, data downloads and telephone calls included) in India, however that has use restrictions (limited amount of data / calls per month). In Ukraine there is unlimited 10mbits/sec data line only for 3,6 dollar/month price. In Russia one teleoperator tried even better this and tried to offer total cost free telephone connection that was paid by advertisers and was cost free to end user. However that was abonded. So the cost of electronics is coming down quickly and even more when roll printed electronics (displays, touch keyboards and batteries) are reality (those are in fact possible to manufacture already today using roll printing, if someone just start manufacturing them). So manufacturing cheap smart phones (with Symbian, Linux, Android / Brillo, Windows Phone, Firefox OS / X-Pud Linux, Ubuntu Touch, old Nokia Linux + Symbian together OS, whatever OS or even without OS) that are paid by commercial sponsors and advertising money, and net connection that is also free to end user because it is paid by advertisement money. In development countries (where the customers are) this is possible, like examples of India show, but taking the final step to give the phone free and not charging the 1,3 euro price, and free net connection without 2,5 dollar / month- price. Even more data restrictions than there is in 2,5 dollar line cost can be istalled on free net for developing countries if data is packed efficiently by data compression. And limiting the amount of information that can be accessed to cost free net, these are being done already in different free net scenes ( that are not free to end user, teleoperator still charges for line subscription costs each month). Progressively when time passes more and more content can be upgraded to cost free net, but in the beginning it can be very restricted (but cost free to end user). For operating systems: old Nokia S-series OS (S40?, S60?) had strange hybrid Symbian and Linux together- structure, so Symbian and Linux programs can be theoretically run on the same OS, or in old Nokia phone. S60 became Maemo Harmattan, that became Meego, that became Sailfish. If Symbian apps and old Linux desktop PC programs from 1990s / early 2000s can be run on same telephone / laptop, and that OS is simple and not memory consuming (like nowadays phone OSes) that can be one solution for super simple phone OS. For Symbian (and other) phones synth music sound chips such as Vimicro, Macronix, NEC, Oki, Rohm, CS 9236, were manufactured but now only Yamaha and Dream (previously Atmel SAM) remain. Micronas MAS3515G is old german synth sound chip for phones that also includes speech synthesizer. If there would be some standardised synth sound chip in every low cost phone that would bring the phone a musical instrument. Yamaha and Dream /Atmel SAM have been around 20 years and more so combining old versions of their chips together in one new silicon chip (that uses phones DAC etc. resources, 20 years ago there still was not synth on chip solutions for phones except CS9236) and putting that chip inside cheap phones or integrating synth circuit to processor SoC etc, and for more expensive phones including new Yamaha and Dream chips that have better capabilites, for unified standard sound. Or use chinese Vimicro discontinued product that is propably cheap. Or use phone
s MP3 player as wavetable synth as Antti Huovilainen has proposed. For laptops using FPGA circuit as music synth is perhaps best hardware synth option, there are over hundread of FPGA music synth designs and about 100 of them can be grouped to same FPGA processor because most of these designs are old Spartan 3 or earliar Spartan etc. and one Xilinx Virtex, net even new one, can hold about hundread of these in one one FPGA. Old Symbian phones have softsynths for ringtones, standardising these simple few dozen kilobytes cheap synths for one unified standard for super cheap phones help bring musical content to phones. Earlier there were over dozen softsynth manufacturiers that offered simple softsynths for Symbian phones for ringtones and gaming sound etc. but also for musical synth content. Perhaps Sonivox Audionside EAS still remains, others have been forgotten. That Sonivox soft synth has 256 voice maximum polyphony and minimum 64 kilobytes memory requirement. I remember smallest Symbian soft synths had 7 - 17 kilobytes memory requirement to work properly. Combining these old soft synths for phones perhaps several together under one unified phone standard that is cheap and light on resources (memory and computing power) and this standardised softsynth package can be in every cheap phone, among real hardware synth circuits. For Android there is "Android audio sound scape project" at wordpress.com and its "Android audio filter" 2015. Solutions like "Subtractive synthesis without filters", "Polynomial bandlimited step function Poly BLEP", and "Filter-based oscillator algorithms for virtual analog synthesis", and for old fashioned analog technology in electro-music.com/wiki netpages: "Schematics vault" section, such as "Nicolas super simple VCO", "Simple dual A(S)R", "Super simple resonant lowpass filter" etc, but these are for analog technology.Others are Harmony Systems inc. (Delora) VCO, Pigtronix Mothership note recognition system etc for analog tech. Using phones graphics processor (GPU) like Mali in Android for soft synth processing instead of CPU eases phone processor workload, additive synths and others are possible to build using GPU processing. In China (brand name Cheertone) and in India (brand name Prasad) are being sold cheap "educational tablet PCs" for children, these have no internet connection but just run some program content of their own, including simple music synth, and OS is Android. Turning these childrens toys into powerful but cheap music workstations that run Windows Phone 10 that can use Windows desktop PC softsynth programs, or Linux for Linux PC desktop softsynths, is possible, these would be like Plugiator, itself partly India-made product. Ron Kuivila has made FORMULA Forth- based music description language that is compact. More compact would be APL- based music description language, that would be super dense and low memory requirement. Already is plans to use phone`s MP3 player at music hardware synth circuit (Huovilainen), and is possible to combine many audio compression circuits to one simplified circuit structure (for example MP3 and AAC using partially same circuits for cost effective audio multi codec player). Some delta-sigma modulator types use ringmodulator (or ring oscillator), and as delta-sigma modulators are used at cheap A/D converters these ringmodulators (ring oscillators) in D-S circuit can be used as musical instruments, because hardware music synths use ring modulation. In tinyapps.org are small programs such as Tiny Piano (19 kilobytes) and Synfactory modular synth (119 kilobytes), improved audio / video version is called Studiofactory, these small programs would be ideal for mobile phones. Earlier Symbian phones utilised cheap softsynths which are listed in “Mobile audio technology report and recommendetations” 2007. Modern phone softsynths studies are such as different writings by Jari Kleimola, for example “Logical synth”, and “MOMU: a mobile music toolkit”, “mobile phone XMF extensible music format”. Texas Instruments TMS 320 was released 33 years ago and is still one of the most popular DSPs. There are versions of it that combine ARM core together with TMS 320 DSP, such as TI OMAP, Da Vinci, DA25x (that combines ARM processor and TMS 320 audio processor) and TMS DM series. Abot 20 year ago first ARM7 + C54x series TMS 320 appeared. Older music synthesizers used only one TMS 320 DSP, like Nord Micro Modular and others. TMS 320 is also used as video DSP, and earlier TMS320 together with ARM core was used in mobile phones, for video processing. Because TMS 320 video processor and audio processor share common features, and TMS 320 was used in many old music synthesizers, and unified ARM core and TMS 320 combination existed already about 20 years ago, building cheap mobole phone processor that combines ARM core together with TMS 320 DSP for video processing, but the same TMS 320 circuit that video processing uses is also used, with slight modifications, as music synthesizer that can represent the processing power of Clavia Micro Modular 20 years ago for example. It would be Android phone that has ARM processor but instead of Mali video processing it would have TMS320 based unified video and audio processor and that unified audio and video processor is capable of using for example old Nord Micro Modular software and other old one TMS 320 processor- synthesizer software. The result would be cheap mobile phone that uses TMS 320 instead of Mali as graphics but also as a powerful professional grade sound synthesizer that costs only few dollars. And because modern processor has much bigger transistor count and huge speed improvements over 20 year old designs that were manufactured at 350 nm or 250 nanometer technology, audio processor capacity would require only few percent of TMS320 video processor resources, but still audio processing efficiency is in the level of old late 1990s - early 2000s TMS320 audio processors. Even multiple processor synths like Korg Oasys PC card from the year 2000 that had 4 +1 TMS320 could work on cheap modern mobile phone ARM processor with TMS320 video processor instead of Mali- combination if TMS320 has capacity to process audio also. So cheap few dollar price feature phone could have sound synthesis capabilities that old expensive hardware synths had about 20 years ago. But perhaps extra payment license is needed if user wants to use sophisticated synth software such as Nord Micro Modular or Korg Oasys PC card software on mobile phone. However good hardware synth is included in feature phone if TMS320 is used, and freeware synth programs for it are possible. Because new phone SoCs have MP3 player included, TMS320 audio capabilities can be totally concentrated in sound synthesis if MP3 audio decoding is done at special MP3 player circuit that the phone SoC has anyway. Even MP3 player circuit can be used, with slight modifications, as music synth. Cheapest chinese toy electric pianos are such as Peng Zhan (or Peng Jia) 168B, that earlier cost only 1,65 dollars at cheapest to buy one piece, and that cost even included shipping that one piece to overseas, can be used as super cheap musical instruments for example putting cheap microcontroller inside and microcontroller is programmed as music synth, or small amount of memory inside toy keyboard and using that keyboard as sample player etc. These super cheap keyboards as “real” musical instruments (not just as toys) would have a marketplace at the third world. For making net connection to phones cheaper: “puggle.me” app (that program is on Github so it is perhaps not ready yet) type automated searching for cheapest net connection, and Jana / mCent net app that makes toll free internet connections possible, should be in every cheap phone. There is article “On the role of infrastructure sharing for mobile network operators” by D. E. Meddeur.
For video compression if data rates are going to be minimal, one solution is video super-resolution that expands very low pixel count picture to larger level. For example if 96 x 128 pixels is smallest possible quality, video and image super- resolution could make quite acceptable quality picture of low data bitrate transmission. Not only pixels, but colour value of each pixel and contrast can be downscaled also, and the same method that is used in pixel count increasing can be used on colour bitrate increasing and improved contrast / luminence quality. For transmission video stream or still image if it is downgraded to minimum 96 x 128 pixels, colour value per pixels downgraded also, and luminence / contrast values to minimum. Video / image super- resolution hardware in the phone reconstructs the original picture or video in original pixel count and colour and contrast levels that video steram had before downgrading. Altough original quality is not achieved and information is lost , quality is much better than downgraded 96 x 128 pixels with minimal colour and contrast values. Even in mobile phones are marketed today video and image super-resolution software, and specific hardware processors such as “Xevic image processing on Fujitsu phones”, and Synthesis Corporation Super Resolution Scaler for Mali graphics Android processors. However also colour and contrast values could be “super-resoluted”, not only pixel count increased. for simple implementation: “Low complexity image / video super-resolution using edge and nonlinear constraint.”, “Primitive solution of video super-resolution” Devi 2012. Many commercial solutions for image and video super resolutiona are being sold: vReveal, CEVA multi-image XM4, Neuron Doubler, Video Enhancer, MDSP resolution enhancement software, Multi-frame super resolution toolbox, MSU super-resolution filter, Morpho super-resolution, AMD virtual super resolution, Amped Five, Topaz Enhance, Sony super resolution de-mosaicing processor, Almalence super resolution zoom, AKVIS magnifier, Nvidia dynamic super resolution, “NTT Docomo over complete transform and improved image super-resolution”, SuperResolution Plugin, Sourceforge super resolution free download, Keitoku Giken FE-R1EX processor. If 96 x 128 pixels can`t obtain enough information from picture because artifacts in picture are so small that low pixel count destroys information, video and still pictures must be “zoomed” to such a level inside the widescreen picture frame that super resolution can bring acceple levels of results when the video or still picture is decompressed and pixel count increased after superresolution treatment. This zooming can be made automatically by some computer program, but if it is not a direct live transmission, some person who monitores the quality of input and makes adjustments and “zooms” the picture when needed if video suffers from pixelation even after super resolution treatment. Of course information outside zoomed area of widescreen picture is lost to viever if 96x 128 pixelation is used, but at least inside the zoomed area picture quality is good and video super resolution can represent decent video stream. Other ways to make bitrate even lower is “Video compression with colour quantization and dithering” Raja. For colours japanese “Practical colour coordinate system” and RG system that uses only 2 colours or swedish Ikaros project RGi (RG + luminence) . Also using continously changing display pixel accuracy, for example changing between low 64 x 48 pixels to 120 x 200 for each keyframe of video, now video quality changes constantly even subsecond time, video stream blurs and sharpens constanly and also zoom is pumping trying to keep quality of video acceptable, but at least that kind of “variable pixel rate” helps keep bitrate to minimum, only that minimum amount of pixels needed for each keyframe that manages to make video at least somewhat accpetable is needed for transmission. Video super-resolution hardware in the phone extends pixel count to normal TV transmission levels, or at least 120 x160, and also increases color values and contrast that is in downgraded video stream dropped to minimum level. This variable pixel rate minimum pixel count, and also colours and contrast per pixel, and would be “minimal video and image codec” for super low bitrate transmission, also using same video compression methods that standard video codecs (libde265 etc.) do after this video downgrading bitrate drops to absolute minimum. So this is super resolution method for pixels, colours and contrast added together with conventional video compression for really super low bitrate. The pixel count may vary also on output stage, for example regular TV quality 480 x 600 pixelation can drop as low as 120 x 160 sometimes, on display screen. So video super- resolution takes low variable pixel count 48 x 64 to 120 x 160 about input with minimal colour and contrast values, automatically or manually by some person who monitors the input zooms inside large widescreen frame if pixelation destroys information, video superresolution corrects colour and contras values back to TV quality level and pixel count also, but sometimes when processor cannot hold the workload pixel count of the display downgrades even down to 120 x 160 level. So the displayed video stream blurs and sharpens constantly as information in video stream changes, zoom is pumping constantly and also pixel count of the display suddenly changes from TV quality to low display pixelation and back, but at least data rate of this type of video / image super- resolution codec would be super low bitrate, because super- resolution hardware does video restoration work and tries to present video stream as such as it was before downgrading process that drops pixel count to very low for transmission and drops colour and contrast values to minimum also.
Video and image codecs use Look Up Tables (LUTs) like Colour Look Up Table (CLUT) in order to save information. Vector quantization is used in video codecs like Daala. If there is indexed colours, what if there will be indexed shapes and indexed movements also. So there will be “Shapeform LUT” and “Movement LUT” just like there is colour LUT. These methods are already used in some way in modern video codecs. There will not be just “indexed colour” but indexed movements and indexed shapeforms and indexed textures also. They will be quantized using vector quantization or other method. If now also time is encoded in the look up table, almost everything that happens in the moving picture can be indexed. For example “blue surface moves from top of the picture frame to bottom changing from straight shape to curved and changing colour from blue to grey”. Colours are already coded using three base colour values and the colour comes of the combination of these three values. Now also shapes (and textures) of the picture are indexed and movements also, and can be represented at the combination of indexed values like colour is represented from three color values. Large frame is in video codecs divided into small subframes and then these small subframes are coded (quantized). Indexing also textures, shapes and movements (changing of shape to other in some amount of time) can be encoded if enough small subframes are used. When timing information is encoded in the subframe (time between changes and direction / vector of changing shapeforms) everything that is happening in the picture (colours, textures, shapeforms and movements) is indexed (and quantized) and put in the look up tables. Using look up tables can be like Smacker video codec which encodes (quantizes) information from the picture into look up tables, and when they have been quantized as to save bitrate, also these look up tables themselves are quantized to save bitrate still. And if there is already indexed shapeforms and textures in the memory of the device, and only timing information (time of change and vector / direction of change) is transmitted and “address number” of the required shapeform or texture, now instead of quantizing a single pixel or even large group of pixels, now very large surfaces is indexed and quantized, and there is no matter anymore how many pixels there is in the picture, because this quantization does not count any separate pixels at all but shapeforms and textures. So encoded this way picture can represent zillions of pixels, altough accuracy of these representations is in doubt, how much this shapeform / colour / movement / texture can represent real world. It is more like computer animation that tries to make animated scenes as realistic looking as possible, and not like real video coding. But at least ultra low bitrates can be achieved if no actual pixels are quantized, only small subframes in larger picture frame and these small subframes are encoded using predefined “building blocks” (like three colours that make all colours when they are blend together), that are simple shapeforms, textures, different colours and timing and direction of movement information (vector) that enables moving pictures. When these “building blocks” are mixed together like three colours are mixed to get the true colour with timing information, If these predefined “building blocks” are used that are stored already in device
s memory, they perhaps cannot represent reality as such accuracy that real video coding that uses pixels can do, but ultra low bitrate video coding is now possible because no pixel values are needed anymore and picture when it represented this way can have million or zillion pixels without "pixelation". Scalable Vector Graphics (SVG) is a vector graphics that uses Bezier vectors (or Splini if that is possible) to draw pictures and animation. Vector graphics can be used as video codec so no actual pixel information is needed anymore, resulting picture and movement in the moving picture is just a big mess of Bezier and Splini curves. Pixel count has no meaning anymore, picture can be 2K, 4K, 8K, 16K Omnimax or even zillions of pixels on display because information is not stored at pixel basis but vector graphics only. Simple "building blocks" can be used, so when the camera is filming a face, it has a recognitition program which notes when eyes or hair is filmed, and then uses instead of real world transmission simple signal that indicates "eye" and "hair" and reciving device has inside memory that contains graphical building blocks like "eye" and "hair", the receiver just puts eye and hair in right place in the picture frame and takes those pictures from the receivers own internal memory, no actual picture of the faces is needed to be send. Information of colour and textures of the surface is however needed to be send, and graphical fine-tuning that “building blocks” would represent real world more precisely. Fine tuning is achived using Bezier and Splini curves that transform simple graphical “building blocks” into something that resembles picture that camera is filming. These building blocks can be simple Bezier etc. “wire models” that computer animation uses, building blocks of for example “human” (“man”, “woman”, “child”), “car” “lorry” “truck” “train” “house” “block house” “tree” “forest” “grass” “cat”, “dog” etc, and these universal graphical models are fine tuned to represent reality, if camera pictures a dog, simple wire animation model of a “dog” inside reciving device is used, it recieves precise colour and texture information (has the dog long hair or shorst hair) from the camera, and movement information if dog moves. Movements (moving pictures) are achieved using Bezier and Splini curves twisting and turning with these basic graphic"building blocks". The look of the wire model building block can be slightly altered by vector graphics curves so that individual shapes of objects can be represented. So this is actually computer animation disguised as live TV transmission. Not a single pixel is needed to be transmitted, just vector graphics surfaces and curves information. If the output looks unnatural and like computer graphics, mixed model that Encapsulated PostScrip (EPS) uses can be done, EPS can simultaneusly use bitbap (pixels) and vector graphics ( Bezier curves). Result would be mixed bitmap and vector graphics video coding. Or then use more accurate vector graphics, picture frame is divided into small subframes like modern video codecs do (actually that “building block” model uses subframes also), but inside these subframes pixels are not quantized, instead pixels of the camera are turned into vector graphics curves inside these small subframes, and colour and texture information is added to them. When timing information of movements and direction of the movement inside subframe are encoded, also moving pictures can be represented using only vector graphics that twists and turns these vector graphics curves, no information of pixels is needed to be send. Receiving device can enlarge this moving picture to very large size because picture contain no pixels and have no pixelation. So this is computer animation, and can use same tricks that computer animation does, using vector graphics, but it is a video coding that tries to represent reality using vector graphics (computer animation) instead of pixels, and can be used like other video codecs. The camera perhaps must have as sort of regocnition program so that it can compare or regognite different objects like houses and humans and sent suitable codeword “house”, “human” to receiving device, and receiving device just puts that “building block” of its own memory to display with slight colour, texture and fine tuning differences. Or assemble picture into very small subframes so graphic “building blocks” are not needed anymore and real world objects can be represented with accuracy, using only vector graphics curves inside each small subframe, so no information per pixel is not needed anymore. Also building blocks of very simple textures, shapes and patterns can be used in this more accurate- subframe model also. Actually modern video codecs are like computer graphics also, they take one keyframe of picture and then twist and turn it so movements can be represent, so perhaps abonding pixels and changing to vector graphics curves is not too difficult. Also when larger picture is divided to very small subframes, about 16 X 16, 32 X 32 or 64 X 64 pixels, (if the picture is 1k, 2K or 4K TV format, subframe sizes may vary (for example 1K TV picture can have 16 X 16, 32 X 32 and 64 X 64 pixel subframes all in the same one large frame), then inside these small subframes vector graphics can be used to represent the information of these small subframes, and camera send information to receiving device as vector graphics codewords, not as pixels. Colour of TV transmission uses Colour Look up Tables or fixed look up table of 24 bit (actually 3 X 8 bit) colour table. Now not just colours, but also shapes, patterns and textures of the small subframe can use predefined look up table, and these shapes, patterns and textures can be very simple basic forms. Now, altough TV camera uses pixels, it sends information what it is filming to receiving device as vector graphics codewords, and vector graphics that are divided to small subframes of the picture inside one large frame. Now pixel count of the camera does not matter the same way as ordinary pixel bitmap camera, even primitive low pixel count camera can send information without pixelation. The fixed look up table of shapes, patterns and textures can be use basic forms, for irregular forms like “tree” or “grass” textures that resemble tree, leaf or grass can be used, and real picture of them is replaced by vector graphics codeword of “leaf”, “tree” or “grass”. Because subframes are so small these basic “building blocks” of simple textures are enough to represent reality in enough accuracy. Movement is achieved using timing information in the codeword and direction (vector) of direction that indicates the changes in shape, pattern, texture and colour in some amount of time, and that amount of time is also incuded in the codeword. Using Bezier or Splini curves these changes inside small subframes can be represented twisting and turning textures and shapes of simple “building blocks” and changing colour and brightness/contrast of these simple building blocks while their shape is changing, so image of moving picture becomes reality, when all these small subframes are presented as part of largers picture frame. Because large picture frame is divided to small subframes individual changes in small frames are relatively small and can be represented using vector graphics, not pixels. For example 3D virtual reality can be created using this system and not pixels or 3D “voxels”, in some 3D virtual reality helmet etc. Because display uses pixels to display vector graphics, that picture contain pixelation, but the actual information that is being transmitted uses only vector graphics and 3D virtual reality that uses vector graphics can be made for that accuracy that only display `s capabilty to present vector graphics is the limit of pixelation, not the vector graphics coded information itself. 3D virtual reality can now be zoomed to small details without pixelation. However the perhaps unrealistic nature of vector graphics compared to “real world picture” with pixels comes visible if someone is looking vector graphics picture close enough. The picture of the camera that send information as vector graphics codewords and not as pixels can be enlarged indefinetely large, into 8K or 16K digital Omnimax etc, or pixel count of the display can be zillions of pixels if picture is coded using vector graphics only, not pixel bitmap. More bitrate savings can be made using larger subframes and more complicated “building blocks” like basic shape of human or animal, and then some “video animation generator” creates the picture from the received vector graphics codewords, but altough bitrate savings are now greater the result is perhaps unnatural looking computer animation, that in some way represent reality that TV camera is filming. One possibility is using hybrid method that Encapsulated PostScript does and include both pixel bitmap and vector graphics into a single picture or TV / video transmission, when irregular forms became so complicated that vector graphics has difficulties with them pixel bitmaps are used, when picture has straight geometrical lines etc. vector graphics can be used instead of pixels in video / Tv transmisson. For example if subframe size is 64 X 64 pixels, the shape of the subframes inside one large frame can be some other form than square, for example if motion takes place in subframe, square subframe can be divided into two 32 X 64 pixel frames, or four triangular frames inside 64 X 64 pixel subframe, and Bezier and Splini vector graphics curves are going different directions in these four separate triangles inside 64 X 64 pixel frame. Small and larger square frames can be mixed, and used together with triangles etc., in one large picture frame. When TV picture is transmitted pixels are transformed to vector graphics curves and no pixels are needed in information transmission. Vector graphics can use Korch, Splini or Bezier curves, like Lazynezumi Pro (lazynesumi.com L-system tutorial) uses.
Yes, for the comments on presentation.
However, @leinonen the technical content is great!
Lots of useful ideas.
If phones, tablet PCs or other devices should as cheap as possible, cheapest solutions must be used. Near field communication (NFC) is in some mobile phones. A version of it that uses sound as data transfer method and does not need special NFC electronics, because only microphone and loudspeaker is needed, and that phones have anyway. Tagattitude Near Sound Data Transfer (TagPay) has been in use since 2005, and similar techniques Narette Zoosh (using ultrasound) and NearBytes (from Brazil) are in use. Encrypted voice data transfer can be used as payments etc. similar way NFC uses radio waves. Voice data transfer is ideal for cheap devices because it does need extra electronics, as cheap substitide for radio frequency NFC. NearBytes has 10 cm transfer distance and works even in noisy enviroment. Directional audio loudspeakers (Panphonics) propably makes transfer distance even longer, like acoustic version of Bluetooth. Cheap techniques like acoustic data transfer as substitute for radiowave NFC is needed in super cheap devices that are given free to users of free internet connection (advertisement paid connection by commercial sponsors, limited access to internet and communication compared to customer pays- model). Also what kind of cheap entertaiment free internet connection will have? Public domain movies are either very old, or then very bad c-class movies, and their amount is limited. But Bollywood movies, old and new, are offered in dozens or hundread netpages for free, and amount of them for free download is staggering, over 6000 Bollywood movies is in one netpage (Nyootv.com) alone. All are downloadable for free. In Youtube in “Cinemascope Movies” is over 400 indian films at least, probaply many more. Netpages like apnaview.com, oldmovietime.com, bharat-movies.com are full of Bollywood movies, old and new. So if cost free to end user internet becomes reality Bollywood movies (perhaps mostly old Bollywood movies, the cheapest option) are the most significant entertaiment. Because content in cost free internet to end user must be cost free too, so public domain or other minimal copyright cost material must be used. One way to make super cheap devices to development countries market is musicmaking software and hardware. Free softsynths are vailable at Android and Linux platforms. Making Linux run Android Apps is possible (several conversion solutions are for free), Windows music softsynths can be run on Linux or Android using Wine and Qtractor, Ardour etc. Even iOS apps can be run using Cider APK (Cycada) or iEMU APK, in Windows iPadian is similar solution. For poor countries free apps like free soft synths for musicmaking are the most important free goods. If very cheap device can run free music apps, not only Android and Linux together in same device, but Windows and IOS too, that would be ultimate solution. Windows is offered for discount price to cheap laptops, about 25 - 30 dollars. Hardware memory is restricted to 4GB / 32 GB and also the size of screen is limited, and processor must be cheap (Intel Atom, or Via, for example). Cheap “music computer”, mini PC style small box, small display (made using e-ink or other cheap), DVD-ROM is needed (altough it is nowadays forbidden in cheap Windows requirements), and perhaps slight increase in memory requirements (to 6 GB / 54 GB for example) because some sample-based softsynths have several dozen gigabytes of samples. The music computer can run Linux, Android apps, and it can have cheap Windows also, perhaps in compresssed OS form like some laptops have. Main OS is Linux or Android but it can switch to Windows if needed. Also iOS apps can be used through Cider APK, iEMU APK in Android or Linux mode. Windows increases the price 25-30 dollars compared to without it using only Wine, but now softsynth platform is better and stable. Outside typing keyboard and large screen (sold as additional extras, not in standard configuration) is needed also. Price must be as low as possible. Altough it is possible to run Android programs aimed for ARM processor to x86 processor using conversion software, and different operating system apps and programs will work in Android or Linux (both Windows programs and iOS iPhone and iPad apps), conversion with processor hardware base from ARM to x86 and vice versa makes this difficult. iPadian is (?) an conversion program that makes (?) iOS apps made for ARM processor to work in x86 Windows (?). If that is true, however conversion to operating system to another is diffiult enough, and conversion to one processor architechture to another even more difficult (if at the same time operating system must be changed to another one) so perhaps this cheap “music computer” needs two processors, other is ARM, another cheap x86 (Atom or Via) so that Linux desktop PC, Windows desktop PC, and Android and iOS mobile apps work in same computer. Main point is that because Windows (and Linux) softsynths are offered for free (thousands of them), and Android musicmaking apps also, and to lesser extent iOS apps (for free), and people in the third world countries does not have money to buy a hardware synth, so offering to them at cheap price (or for free) softsynths and other musicmaking software would be great help. So (super)cheap platform device (“music computer”) that can handle four operating systems (Linux, Android, Windows desktop and iOS mobile apps) at the same time and for free (except Widows desktop that costs 25-30 dollars at cheap devices) is needed. It would be little box like mini PC, and desktop PC monitor screen and typewriting keyboard would be extra features not sold with the box itself. But using two cheap processors (ARM and x86) perhaps makes performance not so good, so selling two different devices (other with ARM processor, other with x86) is also possible. Specs should be such that discount price Windos desktop OS (25 - 30 dollars) can be used, or then using Wine emulator instead (which can make softsynth / VST operating unstable and difficult). Selling softsynths to this device is perhaps with DVD-ROMs that have softsynths etc. software. Ecodisc single layer DVDs are cheap to manufacture. Softsynths can be downloaded for free, but net connections in development countries are not the best, so selling DVD-ROMs directly to users, price printed on DVD disc and package, 1 Dollar price collecton discs with music software. Music software like softsynth can be compared to music album. Streaming services like Spotify pay 0,0084 - 0,006 dollar per stream, 10 downloads is an album. But only 0,0016 USD or less (0,0011 USD or less is avarage in Europe) goes to artists, record company takes the largest amount. If music album is compared to softsynth then 60 softsynths make 1 dollar DVD-ROM, 4,7 gigabits or 2 X 4,7GB if double sided single layer disc is used. Softsynths here are those that are otherwise offered for free in internet. Now some amount of money (like spotify payment, but from product that exists on DVD-ROM or memory card) goes to free softsynth developer. Same method can be applied to selling music directly to users in development countries, that music that is nowadays offered for free download to internet and without record company support (indie artists giving their music for free in the internet). If 60 million DVD audio records or DVD-ROMs are sold one music album or softsynth will bring one million dollar money to its maker. Beacuse mobile devices does not have DVD-ROM inside (phones and tablets), SD memory card distribution of content is also possible. Internet connections in third world countries are weak and slow, so downloading gigabytes of softsynths would take ages and broken down link connection propably corrupts the data files, so direct sale of content to customers using memory cards and DVDsis one option. People in poor countries have no money to buy iPhone or iPad, so conversion program that translates iPhone and iPad free musicmaking apps to platform that is cheap is needed. If those free apps cannot be downloaded at Apple store because devices are not Apple manufactured, other distribution channel for those iOS mobile apps that are free or few dollar price is needed. Direct sale using memory cards and DVD-Roms is one option. Also softsynths are “too sophisticated” nowadays, 32 or 24 bit sound with 96khz sampling rate is too much and useless, and same goes to high sample rate sampling instruments also. Lower bitrate 36 khz sampling rate is enough for human hearing system, or even 32 khz (higher sound than 18 khz humans does not hear, and musical notes over 15khz humans does not hear anyway). Simply adding zeros (silence) to this 16 or 18 khz sound will make 22 or 24 khz sound if needed in the output stage. Softsynths have poor dynamic range anyway, about 60 - 70 decibels, so 12- 16 bits is enough. Some softsynths in mobile phones etc. perhaps have 50 db, or less than 48db “dynamic power” (dynamic range) and that is even below to 8 bits chiptune music quality. So not wasting bits low sample rate and bitrate is needed. Some sampling collections for softsynths have dozens of gigabytes high quality (too high quality) samples. If floating point processor is used why not use differential method like DPCM makes 16 bit linear PCM 8 bit DPCM sound, from 32 bit or 16 bit floating point sound 11bit or 10bit floating point sound, but with differential (“DFP”) form would have same quality but lesser bits. 10/11 bits is used in OpenGL graphics. If integer processor is in use 4 bit DDPCM (Dynamic DPCM, bitsnbites.eu- netpage) is enough for softsynths, and using simple dither 1-bit signal can have 4 bit precision, so 1-bit dithered signal with 4 bit DDPCM quality is enough for soft synths. Sony Super Bit Mapping had 22 bit quality using 16 bits, and no audible alteretions in sound, so SBM method in 1 bit sound would have 6 bits precision and signal is still not different than without SBM because changes are below audible range, and this signal can be treated and processed as ordinary 6 bit signal without noise shaping changes to sound (or as 1 bit signal with 6 bits of accuracy) , but it is only 1 bit signal (with 6 bits of accuracy). This “differential floating point” and 1 bit integer that can be dithered/noise shaped to 4 bit, 6bit, or 8 bit DPCM (ADPCM / DDPCM) accuracy can be applied to sampled sounds also not only sound synthesis algorithms, and to general audio (music and speech, any sound) codec also. If 8 bits precision is needed from 1 bit sound noise shaping goes to audible, so if this 1 bit noise shaped to 8 bit DPCM file is played without similar noise shaper codec, difference in sound is noticeable. But 1 bit to 4 bit simple dither or 1 bit to 6 bit Super Bit Mapping sound can be played even without precise codec because difference in sound is unnoticeable, changes of sound will stay below audible level (altough noise shaping is used). Also DTS sound encodind that uses “coherent audio” coding that encodes only differences between channels, so in its maximum (HD form) several thousand sound channels can be compressed and encoded. If same methods are applied to soft synths, different oscillators and filters are used but only difference between them is computed and similarities are copied from polyphonic voice to another, perhaps then softsynths with thousands of voices would be possible. But modern softsynths propably use these methods anyway already. Phase shift keying, frequency shift keying, amplitude shift keying etc. methods use low bitrate, some even are1-bit for default, and applying them in “coherent audio coding” style methods perhaps make low bitrate softsynths, those keying methods use sine waves, square waves and sawtooth waves already, like oscillators of sound uses. Standard waveshapes that are modulated with those phase shift keying methods (some of them are only 1 bit by default) are already used in softsynths. Other bitrate savings using sparse sound synthesis, numerical sound synthesis, “Audio synthesis by bitwise logical modulation”, “Even more errant sound synthesis”, “GLITCH: noisy fusion of math and music”, XOR synthesis, “Evolving sinusoidal oscillators using genetic algorithm”, Vector Phaseshaping Synthesis, “CORDIC based digital sound synthesizer”, “CORDIC and Taylor based FPGA music synthesizer”, “Design of FFTs using CORDIC and parabolic synthesis as an alternative”, Polynomial Transition Regions PTR (yofiel.com has many examples of these), Feedback Amplitude Modulation FBAM etc. Pine64 is cheap (15 dollars) PC platform, Unuiga S905 (25 dollars) and “Kodi box”, cheap chinese set top box PC (30 dollars) are examples of cheap tech. Instead of set top box / mini PC style music computer (with cheap e-ink display, even black and white display, in the device, because Windows needs it according to cheap desktop OS requirements) perhaps Android TV stick / Linux stick style small device that ihas small display that is required and uses memory card only as storage device (DVD-ROm is missing), so in appareance this device looks like Plugiator softsynth platform but instead of DSP it has ARM and x86 processor inside. Cheap large display like desktop PC, typewriting keyboard and musical keyboard / MIDI controller is sold as extra features, not in standard configuration (that has just “TV stick” music computer). Cheapest Android / Linux TV sticks are being sold below 10 dollar price in China, and because they are computers with Wifi, memory etc, they are like cheapest Android phones that are being sold at 10 dollar price range in India and China. Instead of ARM or x86 cheap processor (or processors) perhaps Kalray manycore processor will do, it is for embedded devices and it is reasonably cheap, has good computing power, but it is neither ARM nor x86, so programs written for ARM or x86 processors work equal badly in Kalray. But because conversion programs from one OS to another or one processor architechture to another are emulator programs anyway, processor that runs emulations as OS like Kalray is perhaps suitable for many operating system “music computer”. Any device that can handle four OS (Linux, Android, Windows desktop for softsynths, and iOS mobile apps) is needed. Because in rich countries there is no need for this device it will be sold only in third world countries. For sound (sound codec, sound synthesis, compression, sampling and processing) perhaps vector compression like Additive Quantization (Martinez), also called extreme vector quantization, is suitable. Also “Maximum mutual information vector quantization” 1995 Wilcox, “Soft vector quantization and the EM algorithm”, “Amplitude-adaptive vectro quantization system”, “Musical sound analysis-synthesis using vector-quantized time-variyng spectra”. Other: Waverazor synth, Tone2 impulse modeling synthesis IMS, decoy.iki.fi/dsound: “Sound synthesis”, “Towards timbre synthesis and method for sound…” (Kantz), “Extenting the Macalay - Quatieri analysis of synthesis with limited number of oscillators” (Fritz). There are virtual analogue models that can reproduce exact analogue waveforms, but they need several megaherz bandwith and huge oversampling, Novation Peak uses 24 mhz sampling rate. Delta-sigma modulation (or Takis Zourntos nonlinear oversampling) uses oversampling, and in the Hatami distributed model has low bitrate (Hatami model has been improved to even better now), so virtual analogue synthesis using Hatami or other 1-bit model (or “A higher-order mismatch-shaping method for multi-bit sigma-delta modulators” A. Lavzin 2002) would have low bitrate. Same methods can be applied to general sound codec also. Also music description languages can have code density that has low bitrate, Forth- based (Formula, Ample, Moxie, MacMox, Neon, iMops, Sporth), Common Lisp Music (Scheme, Impromptu, Extempore, Racket), and even APL programming language and its many versions to music generating (altough no known examples of APL applied to music description language exists). Faust (and Faustina), and Futhark (GPU programming language), Lytescript, Bard etc. effective programming languages can be used. “A declarative metaprogramming language for digital signal…” (Kronos) is Common Lisp-based. Earlier music programming languages used only few kilobytes of data, because computers did not have memory and capacity, so early Csound (from 1980s) or earliest PureData versions (from 1996) propably has small memory footprint. Also very early sofsynths from 1980s like “Armonyx” or “Tecmar sound syth” for IBM PC (propably MS DOS programs) were very small programs to nowadays standards. Also “Fractional Fourier shift transform” (Rutanen 2016), “Algebraic integer quantization”, “Sparse atomic decomposition of audio signals”, Xampling, “Non-negative matrix factorization” (Koivunen 2016), “Recent advances in PWM techniques and multilevel inverters” 2014 (space vector PWM SVPWM), Differential Pulse Position Modulation (another modulation method like phase shift keying), “Differential spatial modulation”, “Dynamic differential modulation of sub-carriers”, “Angle differential modulation for odd-bit QAM”, “Duobinary modulation for optical systems”, “Dispersion modulation using allpass filter”, “Abstract frequency modulation”, “1bit quantization and oversampling at the receiver”, multiple description delta coding, “Bit-interleaved coded differential space-time modulation”. Also conversion of different operating system to another: Google Chromebit, and Intel Compute Stick are kind of cross operating system devices, CloudReady, and Igel Software are cloud based cross platform and OS systems. Google Chrome is only wide scale succesful thin client system that has been a success when Firefox OS was cancelled. Free cross platform solution that can handle music software (and several operating systems) is needed. Processor architechture where memory and processor are the same like iCube UPU exists, but it has not gained much use, and is like experimental thing and not gonna make widespread use. Other: “GPU driven killer synth software”, “A ROM-less direct digital frequency synth based on fifth-degree bezier curves” 2013, dinisnoise.org, Direct digital delta sigma modulation (Orino), “A modulation matrix for complex parameter sets” 2011, “Bilevel sparse model polyphonic music transcription”, “Neurogranular sampler” Mironda 2009, “Parallel evolutionary optimization of digital sound synthesis parameters” 2011, “SKMD: single kernel on multiple devices for CPU-GPU…” Lee 2015, “How to implement digital oscillator ?” 2011, “A survey of computational calculi used in musical applications” Sarria 2013, “Harmony: a system for musical composition”, “Another fast, fixed-point sine approximation”, “ARMO (Auto Regressive Moving Avarage filter)”, “Extreme compressive sampling for covariance estimation” 2015, “Compressive phase-only filtering at extreme compression rates” , “Enchancement in DASS for low cost sensing nodes” Bonavolonta 2014, “Perceptual signal coding for more effective usage of bit codes”, “Universal rate-efficient scalar quantization”, “Quantization noise shaping on arbitrary frame expansions”, “Sparse composite quantization”, “Pairwise quantization”, “Analog additive dithering”, “Optimum quantization” James Bruce. OSCulator is Open Sound Control software, cross platform. In new devices perhaps something better than MIDI is needed, like OSC, TUIO, or Nonlinear Labs control method (TCD). MPEG4 had SAOL standard, that can directly control hardware synth (in phone or tablet PC, but those have no hardware synths, only software- ). If phones or tablet PCs will have inbuilt hardware synths, more modern control than 20 years old SAOL is perhaps needed, and those standardised hardware music synths must use most sophisticated modern synth methods. There is MPEG7 AABIFS etc. Even MP3 player circuit can be turned to hardware music synth, and GPU or graphics circuit used as softsynth platform etc. About modulation: Multiple Description Coding (delta / delta-sigma), and “Multi-Pulse Pulse-Position modulation”, “Pulse Group Position Modulation”, “Modulation by multiple pulse per group keying…” patent EP 1477001A 1, (Charles Hartmann 2002), Bit Angle Modulation and Digital Pulse Interval Modulation are different modulation methods. Tinyapps.org has very small music software (kilobytes), so several hundreads of these programs can be packed to small memory space, like old Symbian ringtone/music softsynths that had just few kilobyte memory requirement. Softsynths can have hundreads of voices, so but humans only 10 fingers, so system that uses some algorithm to expand those 10 notes that are played to dozens is needed, for example expanding 8 notes to 24 voices (3X8) so that 8 voices melody line goes on bottom and 8 voices top human played 8 notes. Unlike MIDI control that replicates notes played, algorithm or artificial intelligence program brings variations to those additional notes that are not played by human, so it sounds like theree persons with six hands would be playing variations of melody. for 10 notes (fingers), program would expand to 32 notes (voices of synth etc.) 11 notes played below and 11 notes on top of human played melody. 64 voices / notes version would expand 8 human played notes to 8X8 melody lines played by 8 different hardware or sof soft synths for example, not direct note for note copy like MIDI, but algorithmic /AI generated variations of basic human played melody line. 10 notes would expand to 10 + 6X9 note melody lines, one 10 note line played by human, AI plays 6 other 9 note sequences with different synths/ softsynths. Also sample player/ soundfond player that uses AI to find from samplebank suitable sounds to play with human played keyboard melody . “Automatic position loop for heterogenous systems” 2016 Baier, “DAINO: a higher-level framework for parallel and efficient AMR” 2016, efficient algorithm for longest common subsequence, generic inverted index, dynamic sparse matrix allocation, “DQMRES: a quasi-minimal residual algorithm”, “Sound signals as RGBA textures” (Gallo). ADC DAC structures like “Multi-phase VCO-based ADC” Madhulika 2016 can be used as musical instruments, A/D converters have oscillators, ring oscillators etc, that can be used as musical instruments in “reverse mode”. If polyphony reaches hundreads of voices, virtual reality user interface with video glasses and data gloves is one option for synth control, even in cheap devices (if video glasses and data gloves are cheap also). If synth sound frequency is limited to 7 khz (14 khz sampling rate) and higher frequencies are made using High Frequncy Replication and aural exciter, and synth patches designed accordingly, multiple voices are possible. 48 khz or 44,1 khz sampling rates can be divided for example integer 7, 48 khz would then have slightly less than 7 khz frequency and sampling rate of about 13,7 khz (48 khz divided to 3,5). 44,1 khz would have 6,3 khz sound and 12,6 khz sampling rate (44,1 divided to 7 / 3,5). This 7 khz or 6,3 khz sound can be expanded using High Frequency Replication in final output stage when all voices go through one mono channel or stereo channel, so final output voice will have about 14 - 16 khz frequency, even more if aural exciter is used. Aural exciter and HFR circuits are required in phone / tablet PC for regular sound codec (speech codec etc.) and music reproduction that uses low sampling rates . Aural exciter circuit themselves can be used as music synth (EXCTR synth by Jari Suominen, Spatial Aural Exciter). Unusual MIDI controllers like Samchillain or Udar can be used in mobile devices and soft synths designed accordingly. If human ear has two sensitivity peaks, one at about 6 khz and another at 13 - 13,5 khz (for young people, over 40 years old don t hear over 12 khz frequencies), and highest musical sound heard is 15 khz and highest sound about 18 khz, no higher sampling rates than about 24 - 27 khz is needed if sound is 12 -13,5 khz and everything over that have HFR or aural exciter, and nobody notices the difference. If integer processor is used quantization noise requires some extra bandwith. Because there exists conversion programs that can run Linux, Android and even (desktop) Windows in same processor, and even iOS (mobile) apps (but there is no single one program that can handle this all, several different programs is perhaps needed for all this conversion from several operating systems to another in same device), and even so that they change operating system to another and processor architechture also so for example Windows programs run x86 programs at ARM processors in Android OS, so best is cross-platform enviroments like Webinos that those programs run smoothly from one enviroment to another. Android Go or other simply systems are best at cheap devices, Linux is simpler than Android so for cheap devices similar to Android Go simple Linux version like Ubuntu mobile that had switchable between desktop OS and mobile is best. If “music computer” has integrated DSPs (that is unlikely, because DSPs are perhaps expensive, but some DSPs have ARM and TMS320 DSP together for example, in relatively cheap two processors in single chip package), then Xenomai-style Linux and DSP- RTOS together is suitable for cheap music computer, Linux is for ARM processor and RTOS for DSP, programmed with Faust or Futhark or other programming language. Also Binarized Neural Networks (BNN) can offer high performance with small bitrate and 1- bit signal path, for FPGAs etc. or mobile phone (hardware) sound synth. Not only soft synths but sound recording (mixing desk) can be virtual also, so no hardware sound desk is needed for recording. If user interface of mixing desk is virtual also, using video glasses and data gloves, now cheap muiscmaking and recording/mixing is possible. There are programs like MixerMuse and “Automatic mixing tools for audio and music production” (Gonzales), Landr mastering AI. IArtificial intelligence can be used in music production, AI adds “auto accompanient” to human played 10 or 8 notes on keyboard. Because humans have only 10 fingers but hardware and softsynths have 32, 64 or more voices, these “extra” voices can be played by AI, that adds basic 10 note melody line arpeggios, chords etc. or invents own melodies that accomply and follow human played melody. This AI uses available synth voices of one or many different synths and/or sampled voices simultaneusly. Now full polyphonic voice potential of synths and samplers can be used, not just using them for MIDI controlled note for note melody copying. This AI system should be in music computer either in software or hardware, and perhaps(?) using binarized neural networks. Artificial Intelligence and music has many applications already. Non-academic examples are “Impro-Visor”, “Harmony Assistant”, Google Nsynth, Sony Reflexive Looper, Future Retro Zillion, Breakage drum machine, A.M.Y midi controller, Magellan Jr. 2.0, Refraktions generative midisequencer, Jukedeck free music generator, Dhalang MG iOS app etc. Not only basic melody line inventing following human player, but also AI generated (symphony) orchestra or jazz band backing could be used, programs exist already that can do this. Music computer is aimed for third world market so price must be as low as possible (or free) for software, and hardware (music computer that is just base device that this software can operate) also as cheap as possible. Requirements are such that discount price Windows OS (desktop) can be used at about 25 dollars for Windows OS. But if this cross platform (operates Linux, Android, and Windows OS, and can use mobile iOS apps also) computer is used it must use both x86 and ARM programs also. There exist conversion programs for one operating system and processor architechture to another simultaneusly, and this is possible because “hardware virtualization” is used so these conversions use software emulation of hardware. The base OS is perhaps Linux, and Ontolinux / semantic Linux perhaps, and filesystems accordingly MongoDB or something similar (?). However normal Linux soft synths must work in this device. On top of that Linux is Android, and Windows and mobile iOS programs can be also used (through conversion, or Windows desktop OS at bargain price is used, iOS apps must use conversion programs because Apple does not licence its OS). Music software that has cross platform capabilities is for example Jazz++ sequencer. If base processor architechture is insignificant because hardware emulation is in use anyway, processors such as Kalray (first generation and cheapest, newer models are more expensive), Loongson / Godson (the latter has up to 64 cores), Phytium Mars, Elbrus, or Cell processor, Mill processor, Adapteva Epiphany, EPIC/Itanium, UPU (Harmony unified processor architechture by ICube) etc., but processor must be cheap because components must be as cheap as possible. Small mini desktop PC and/or tablet PC is perhaps the form factor. Because no iOS is used, only program to operate mobile iOS apps, Haiku OS or Darwin OS is perhaps possible (if needed) in device (if not needed to run iOS apps these OS are not needed). And if roll printed electronic components is possible to be used it must be used for these low cost devices. There are concepts like superoptimization, non-blocking data structures, loop-free data structures, partial lock freedom (Efficient and practical non-blocking data structures Sundell 2004). For most compact programming, like using APL programming language and its modern versions. Superoptimization is available for PIC microcontroller etc. If simple and cheap components are used with limited resources. Programming languages with “extensible programming”, “extensible syntax programming languages”: Red, Racket, Lua, Lithe, Forth, Factor, Dylan, XL (XL1), MirageOS (OCaml). Many of these are Lisp-based. If DSP is coupled with CPU in SoC then Faust or Futhark for GPU etc. If software is running hardware virtualization it does not matter so much what base OS is. Not only is possible to build cheap music computer, but for example cheap mobile phone. OS such as Symbian (or old Nokia OS Symbian + Linux together like S60? OS) that uses Android / Linux apps using Alien Davlik or other is perhaps cheapest and simplest (?) for simple and cheap mobile phone or tablet PC, or very simple Linux mobile OS, “Linux Go” with Android apps operation capability. xPUD Linux is cheap and simple cloud OS, but data rate restrictions prevent using cloud computing in free net, at least in significant level, and due to bitrate restrictions cloud computing should be avoided in free net whenever possible.For free internet connection (in third world countries) advertisement paid, data bitrates must be minimal (because of line cost) so Google Chrome or Firefox OS thin client solutions are impossible. Price of electronics is so low nowadays that even 8-core ARM (or some x86 processor that chinese factories began manufacturing 2017) SoC is perhaps 5 dollars or less if “old” (released 2012 - 2015) 8-core budget priced ARM processors with 4-core Mali graphics GPU (older Mali GPUs than most modern) for exmple, are used in these phones or tablet PCs (when for example 100 million unit single order is made for those cheap tablet PCs that are given away with free net connection, advertisement paid), so thin client solutions are not needed. So line connection price is much higher (if telephone connection with net costs 2,5 dollars in development countries per month it is still 30 dollars per year, much higher than cheapest tablet PC/phone price) line connection must use data compression to max in free net, savings with max data compression is much more important than slow decompression time in free net. Maximum advertisement paid solutions with minimal data rate in third world cost free net.
Also for transportation: development countries have mountainous land, roads are bad, so transport via air is one possibility, but no airfields exists, so helicopter is only way. But airships, helium filled, hot air airships or hybrid hot air / helium models can be cheap to operate, so they can be used as cheap air transport in poor countries. If airship is powered by diesel engine (turbodiesel), then fuel consumption is low. SMA Engines had 2016 protype aircraft diesel engine that is clabale to 800 horsepower as 6 cylinder (and perhaps 1100 hp as 8 cylinder version?). Those efficient engines can be used in big airships. Not only mountaineus landscape, if airship (helium/ hybrid hot air - helium) is more cost effective way to air travel than aircraft, airships (large airships, dozens of tons payload capacity) can be used as intercontinental flights form thousands of kilometers. Because speed is slow (150 - 200 km/h) this way of air travel does not interest those who have money to pay, but only poor people. Trip to continent to another will take several days, and because those airships are most cost effective, they don t have sleeping facilities (except aircraft seats) and are otherwise regular aircraft-like accommodation inside, except a way slower. So only poor people of third world countries would use this airship travel, because they have no money for proper aircraft ticket. Russian “Aerolokomotiv” etc. projects that are like train transport through sky, ticket prices and other is almost like train or bus trip, but travel is through air not land. Not only long trips of thousands of kilometers but shorter trips only (mountainous land for example) using smaller airships of few dozen passangers (large intercontinental airships would have 800 - 1000 passanger seats or even more), those airships can be hot air (thermodynamic) or helium, or helium/air hybrid. Largest hot air balloon Cameron Z-750 can carry 32 people, and Cameron makes airships also. Larger airships need helium. Helium has price problems and hot air airships are said to operate only 2- 5% of helium airship operating costs. But hot air airships like balloons need new skin balloon “envelope” after some flying hours because heat worns out balloon skin. Airships are rigid or semi-rigid (blimps) so they not require envelope changing as much as hot air balloons. If helium airships, hybrid airships (Rozier airships) or hot air airships are more cost effective operate than airplanes or helicopters they can offer cheap air travel to poor countries. If operating cost with turbodiesel engines are equal or more than aircraft or helicopter then it is no sense to use airships. Slow speed restrics interest of rich people to travel with airship, but poor people in third world countries would find airship travel interesting if ticket prices are low enough. Maximization of payload is needed, every passenger is weightmeasured before entering airship, and ticket price is for kilogram basis, heavier customer pays more, and seating capacity is in maximum, 70 -80 kg payload per seat, women and children are light so perhaps even 70kg per seat is enough, if heavier than 70kg avarage per seat customers come to flight when payload per seat is set as 70 kg, some seats must be left empty. No luggage is allowed, or extra price must be paid for luggage on kg basis. Airship always takes on using maximum payload, so that ticket price would be as low as possible. Airship interior would be cramped like aircraft, because maximum usage of payload would go to passanger weight and minimal to cabin structure itself. From a small 20 -30 passanger short range airships to thousand or more passanger intercontinental airships this would offer bargain priced air travel. They are so slow that normal western travellers didn t want to use them, but for poor people any cheap air travel option is welcome, for example in mountaineous terrain. Intercontinental flights that last several days would be very uncomfortable because only normal aircraft interior is in use, no beds etc, and 2 - 3 days is length of flight time. But poor people accept this if ticket price is very low. There is also airship builders in Chech republic, Russia and China, not just in western countries, so cheap airships are propably available. Because fuel consumption must be minimum aircraft diesel engines (turbodiesel) are needed. Some of these are normal car diesel engines retrofitted to aircraft use, so they have propably low fuel consumption. Military transport aircraft have “military overload” configuration, and because these airships must provide air travel at minimum cost they too use “military overload” specs altough these airships are civil aircraft. Keeping payload per seat as low as possible, even below 70kg, perhaps 65 - 67 kg is enough, more seats can be put inside airship. People in poor countries are short and skinny and lightweight anyway, and because women and children also travel, light payload per seat is acceptable. It is possible that some seats remain empty if payload limits are exceeded, but keeping payload weight per seat low ensures that when lots of women or children reserves ticket, payload/seat capacity is exploited in full and no single kilogram of payload weight capacity is wasted. And when take off weight is always at maximum overload configuration every percentage of weight capacity is used and ticket prices are low. Also dividing passengers between different flights so that maximum payload is used (different people have different weights, when these are divided that maximum possible seat capacity is always used in every flight, altough some seats perhaps remain empty because payload per seat is so low, but waste of payload capacity is to minimum). Negative aspect is that sometimes passenger must wait next flight to get a seat and ticket to airship. Large number of seats is more economical after all than high payload capacity per seat because sometimes lightweight passangers are coming to flight so they will have seat and “empty kilograms” is not used per seat, 80 - 90 kg payload per seat will propably lead to waste of payload capacity if people are lighter than that weight limit, and in development countries they are (rich people there of course travel using aircraft because airships are so much slower than aircraft).
In netpage wealthdaily.com, in year 2016 is an article “The cheapest internet in the world”. According to it some firm (from India?) has invented revolutionary method that puts internet line cost to minimum in development countries. It makes possible that internet connections have only small fraction of the cost that they have before. However the two internet articles (wealthdaily.com article has link to another one) keep the name of this tech firm “secret”, and again the additude of these writings is how much money already rich investors can get if investing to this new tech, so again western people look at developing countries poor people as money making perspective for themselves, altough this revolutionary technology is aimed to bring cheap internet connection to people who otherwise have no money to buy it. These net articles were written already over a year ago, and the “secret” tech firm (from India?) and its revolutionary net connection tech has not gained much publicity thereafter. If this technology works and it exists, and it is now possible to offer poor people in third world countries net connection at fraction of the cost that it is today, why not offer it now? If revolutionary tech for cheap internet exists it must be put to good use, demand and need for cheap internet is enormous among billions of people that does not have money for net connection prices of today. The “zero rating” principle has gained opposition mainly from rich western telebusiness firms fearing for their profits, and largely they have managed to make zero rating illegal in many countries using “net neutrality” laws. If net neutrality laws are killing free internet initiatives in poor countries, then one solution is separate “zeronet”, cost free internet with restricted access (only public domain and no-royalty payment material available) and censored material, and no net neutrality legislation. Then is “ordinary” internet for paying customers with net neutrality laws and unrestricted access to material (normal internet of today). “Mobile zero rating: the economics and innovation behind free data” 2016 . In India already several cheap internet projects are being made (Reliance Jio and Express WiFi). If commercially sponsored internet (advertisement paid, free to end user) is possible, so that adverts pay net connection price in poor countries, data rates would still be very low, because adverts need to pay for example 0,08 dollar per day per customer net connection price of internet in Africa (according to outdated information from 2012, 5 years from now at oafrica.com netpage). One million people in a day needs 80 000 dollars advertisement money per day, adverts are send to these one million free net users in Africa. But those net costs are from 5 year old data, and telebusiness firms are making profit, in some cases billions of dollars selling those 0,08 dollar per day net connections, and the internet that they are giving to customers is regular internet with royalty payment filled material. First if zero rating is the goal then internet must be “slimmed down” accordingly, making only cheapest no-cost public domain material available (at first phase, later when free net becomes more commercially profitable costly material can be added), this “zeronet” is offered for zero rating, paying customers get full internet access. This zeronet material must also be data compressed to maximum. No HD videos or high quality audio files, worst quality video and audio only, if ever, maybe only textual material is available in zeronet if net connection is very slow. Many firms such as Spotify or Youtube use royalty payment material. They pay for material and then they pay for net connectivity. If royalty payment material is removed from Spotify and Youtube, and only free shareware or public domain material used, now those payments normally paid for royalty payments can be used to pay zero rating net connection also to internet provider, so internet provider gets more money from Spotify and Youtube for zeronet. Zero rating usually means that no extra cost is for internet connection, but line user must pay basic subscription line cost to telebusiness firm however. Zeronet is cost free to end user, all is paid in advertisement money. That revolutionary new internet tech mentioned in net article is partially using phone s own memory, if I understand it right. If 128 gigabytes of cheap magnetic RAM memory cost 4,75 dollars now (dramexchange.com), large bulk orders propably will bring cost below 4 dollars for 128 gigabytes, then cheap telephone or tablet PC will have few dollar price 100 gigabytes or more memory. Now internet connection is not even needed very often because huge amount of information can be in cheap phone / PC already, and data compression increases this data storage even more. Also cheap magnetic data storage can be used in internet routers, now they can have terabytes or petabytes (if hard drive HAMR or magnetic tape with Fujifilm nanocube etc.) magnetic memory, and huge amount of data in data compressed form in near router, no much need for far distance internet connections anymore, almost all can be in near router or near routers internal memory, that would propably bring down net line usage costs. For example whole Wikipedia fits in few gigabytes (about 3- 5 gigabytes with or without pictures and graphics) in compressed form (Wikipedia offline). There are many new innovations in cheap internet tech that is done by development countries themselves. Examples are SupaBRCK, Tala Mobile, Tizeti Nigeria, kamaleon.co.mz, and most importantly Sliide Airtime that has won innovation award. It is similar to Jana / M Cent app or puggle.me app, offering advertisement paid extra internet time, but it can be even more effective if “slimmed down” version of internet with extreme data compression and payment-free internet material would be used. There are lot of other innovative apps, for example africanbrains.net netpage “AIF announces the top 10 nominees for innovation price for africa” 2017 (Lakheni group buying app, Lokole email app, Omolabale Adenle voice recognition software, Dr.CADx artificial intelligence medical software), and “Announcing the winners of the internet.org innovation challenge in africa” 2016 (SaferMom, Mutt mPharma and Tutelia apps etc.). Cheapest mobile net connection in africa was 0,89 dollars for gigabyte of data in 2016. New technology like that mystic new net connection tech that promises only fraction of cost of todays net connections or larger memory capacity in routers etc. can bring cost down much more. If advertisement paid internet connection is possible already in poor countries using these new innovations, why not use it? Search engines are also making huge amount of money (Google). There are “ecological search engines” like Ecosia etc. that despite the fact that they have only few million users (or less than million), they still manage to donate some amount of their profit to plant trees etc. I wonder if Ecosia would be the size of Google, how much money it could earn then, for donations to charitable purposes? And those ecological search engines are private businesses that are making profit for their owners also. So how much money for charity would bring similar model, for example if Google would use Ecosia s earning model? And use this money not planting trees but to bring free net connection to poor people (and doing this would gain billions of new customers to Google). The hardvare device itself would be full of adware outside and inside, outside it is plastered with logos of commercial sponsors, inside it has adware in its internal memory that shows pop-up commercials now and then, no internet connection is needed for this “advertisement campaign”, and it continues several years (as long device is used). So advertiser does not have to pay to teleoperator advertisement message cost for years because adverts are inbuilt in phone s or tablet PC s memory. Instead advertiser pays for user s cost free net connection to teleoperator, and in return user uses phone / tablet PC which is full of adware and adverts and commercials of advertiser. In Brazil some time ago (perhaps before 2010) was “IT Brazil” (aisatotal.net ATM 1088, according to infodev.org/articles “Quick guide to low cost computing devices”) that was commercial laptop PC or similar that was given free to customers, in return commercial sponsors controlled internet searches (“hotkeys”). But no more information about is available, before 2010 cheapest laptop PC s cost 100 dollars about, so commercial sponsors gave free 100 dollar PC to every customer in return for adverts in the machine and controlled limited internet connectivity. This was private business so it was aimed to make profit. But internet connection was not free perhaps and customers had no money to pay internet altough got PC for free? If commercial sponsors were ready to gave 100 dollar device for free in return of adverts inside it, and this happened already about 10 years ago in commercial business aiming for making profit with it, how about few dollar price modern device for free and rest of 100 dollar or so budget to free internet connection, 0,08 dollars a day / 2,5 dollars a month? Internet in zeronet must be “slimmed down” cost-cutted version of “normal” internet. In africa and india 0,08 dollars a day is teleoperator charging for net connection / phone line, teleoperator is using adverts to pay (most) of line costs but still charging that 30 dollar a year / customer. Either cost of internet should be slimmed down that net connection is possible without that 0,08 dollars (for 100 megabytes per day / 0,08 dollars for example), or extra advertisement money that covers that 0,08 dollar a day is sought. Teleoperator may also slim down its profit marging if for example zeronet will have less tax than usual tax of teleoperators. Telebusiness is at large part government- owned business around the world (either rich western / middle east governement owned firms or owned by state of China). If they decide to offer cost free restricted zeronet in poor countries that is already possible. And propably will also make profit with it. Usually net search goes to datacenter perhaps thousands of kilometers away. But if in every phone or PC has large memory, made of cheapest magnetic memory, now it has “mini-internet inside” and does not need internet connection much. This data compressed “info pack” can be in ROM, not in RAM memory if needed. If 128 gigabytes of flash RAM cost 4,75 dollar now and if ordered in scale of billions cost is even smaller. I don t know how much ROM memory costs. If information is stored using cheap mask ROM memory in “internet inside of the phone” structure that would perhaps be the cheapest option. Information is data compressed education and general information and also concerning food, health, clothing etc. Speed of memory can be slow because data compression makes memory slow anyway (decyptrion of data). Then if information is not found in phone s own “search engine” inside phone itself it contacts near router. Router could have terabytes of memory, in RAM, ROM, hard drive or magnetic tape (swedish carousel memory was fast tape memory, similar with wide tapes, about 5 cm / 2 inch instead of regular 1,27 cm / 0,5 inch will rewind fast, and using Fujifilm nanocube could contain even petabyte of information in maximum version, data compressed even more, because of data compression slowness is not a problem because decompressing data is slow also, and magnetic tape is cheapest magnetic mass memory). If distributed search engine is used no supercomputer-driven search from thousands of kilometers away is needed, distributed search engines have been studied for years. There are dozens of distributed search engines, from simple to sophisticated. Now router can contact nearby routers and do distributed search engine search of data. Only if information is not found in nearby routers cheap magnetic memory, it will call faraway internet search. Even human-augmented version can be used, “Human search engine” (Wikipedia) in every village of africa, south america or south asia is a person that does human search engine job with net router. Now instead of supercomputers that are costly, several thousand otherwise unemployed poor people of third world can do human-augmented search engine jobs. In netpage ariadne.ac.uk “Human powered search engines” is more. If distributed search engine works with human augmented form only relevant search results will show perhaps. This hybrid distributed (automatic computer) search / human search engine (human augmented search engine) can give jobs to hundreads of thousands of people in poor countries, and now internet search is possible without expensive data center supercomputers far away. Google gets dozens of billions of earning in net searches, some of that money in poor countries in africa etc. can now go to people that do human search engine jobs in those countries. And that money is not taken from Google because zeronet offers net connection to people that are nowadays without net connection. Semantic filesystems, Ontolinux, semantic search engines, “natural language user interface”, SemanticRank algorithm etc. can be used. Distributed internet models like “Wireless ad hoc network” (MANET, B.A.T.M.A.N etc.) and Xanadu / ZigZag routing protocol can be used in base level. In higher level this distributed network contacts normal internet. “Wireless village” (Yamigo cost free messaging service etc.), push to talk over cellular (phones contact directly without router, Motorola iDEn, wiDEN, OMA PoC), Qaiku and infinit.io message services, Datawind UbiSurfer and Skyfire (now Opera) browsers that had inbuild effective data compression etc. Long distance connections can be done with long range Wifi, up to 500km and beyond is possible, so couple dozen long range Wifi hotspots can cover whole african continent. Using athmospheric propagation that exists even gigahertz range long range connections are possible. Also air balloons or satellites are studied, satellite launch vehicles are going transition phase so couple of years from now satellite launch prizes are just fraction of what they are now. Air balloons in high atmosphere are perhaps still cheaper. But perhaps cheap routers like SupaBRCK make possible that “ordinary” router link internet / xanadu net covers african continent cheaply. No electric power is in rural areas of africa and americas. So solar power is needed. Simple parabolic or other shape plastic mirror which concentrates light to small solar cell is perhaps cheapest solution. In astronomy several fancy optical mirror shapes that make possible following for example star altough mirror stays rigidly unmoved have been studied, Pfund telescope, Herrig telescope and other optical systems. That mirror used in solar cell can be used at night as lampshade, and if cheap yellow led lamp or other lamp comes with phone, and solar charger phone can be now used as lamp at night. Plastic mirror should be foldable soft material. Other electric charging methods are heat charging using body heat or open fire to charge phone with electricity etc. If human augmented search is used ask.com or Quoara type answering service is also possible. Because of low bitrate of zeronet new solutions are needed, sound can be compressed using TwinVQ based or Additive Quantization (AQ,) vector compression solutions, still pictures and videos compressed to max and only small pixel count available etc. Sparse coding methods used with knowledge of human sensory system so sparsecoding audiovisual information etc. Also computer programs need another approach in zeronet. There is tinyapps.org for few kilobyte apps, and demoscene has “four kilobyte art”, programs that need only 4 kilobytes to work. Such minimal apps and programs are needed in zeronet. Old VIC20 games used only 3,5 kilobytes, and Atari 2600 had 4 kilobytes with only 128 bits of RAM. Emulators which run these games on ARM processor etc. exists, and old Acorn Archimedes or RiscOS games are possible also. Thousands of such games can be in phones ROM or RAM memory. In netpages like myabondonware.com are thousands of old computer games, that are few kilobytes each or few dozen kilobytes. Using emulator ARM processor can use them (QEMU etc.). Sinclair Spectrum alone has over 20 000 games. Phone / tablet PC and router can use cheap mask ROM in “info pack” (over half of its memory capacity or 3/4), and cheapest (slow) RAM for the rest (less than half or 1/4 of its memory capacity). Router if it is large hub can use magnetic tape, magnetic tape carousel memory or hard drive also. All is data compressed so memory is slow anyway. If phones or tablet PCs are suddenly being build in billions of units, price per one unit will be very low. Semiconductor manufacturers perhaps must change 300mm wafers to even larger than 450 mm to keep pace of the demand. Those cheap processors are manufactured using outdated tech, 40 - 20nm, or even 65nm if cheap manufacturing capabilities of semiconductor fabs of India are used (India is building semiconductor plants). So if billions of processors are needed fast then wafer sizes of 550 mm or even more is needed using old 65nm etc tech or similar. Non-silicon manufacturing of thin film transistors in “glass motherbords” already uses 680 X 880 mm wafers etc. Price of cheap silicon SoC is very low in massive orders of 100 million units per order etc, and even octacore processor with quadcore Mali GPU can be used, but those CPU and GPU are older models, already going out of production, for super cheap tablet PC or phone SoC. Then battery and display can be made with roll printed electronics, cheapest option, and magnetic memory cheapest possible (but internal memory capacity perhaps as high as 100 gigabytes about, if needed, in ROM and RAM, cheapest is mask ROM perhaps, most of memory, then RAM). Also memristors are going to large scale production from 2018 onwards, and this “memristor ratioed logic” is different from transistors. Memristors can be used as CPUs or memory (Intel XPoint etc.). Newest ARM cores are designed for about 16nm manufacturing, if these are build using old, like 65nm manufacturing how it affects the processor? Data compression like Finite State Entropy, Qdigest, ODelta compression (Gurulogic Microsystems), Octasys Comp (Ypo P.W.M.M van den Boom, compression method), IDBE text compression, Albert W. Wegener and Clinton Hartmann (“group keying”) data compression patents, Quadrature Amplitude Modulation or (spherical) vector quantization / spherical ADPCM with Vector Phaseshaping Synthesis, Feedback Amplitude Modulation (FBAM) with delta /delta sigma, extreme vector compression called Additive Quantization AQ, “New approach based on compressive sampling for sample rate enchancement in DASs” Bonavolonta 2014, “A novel pseudo-parallel delta sigma modulator” Johansson, Svensson 2014, Takis Zourntos: “oversampling without delta sigma”, “The faster-than-fast Fourier transform” news.mit.edu 2012, “Anamorphic Stretch Transform”, unum computing, inexcat / propabilistic computing (Rice university, iSlate tablet PC India, now PC can have large but inexact GPU for example octacore with 8% error rate and small but exact CPU, and octacore CPU with 4 exact, or 4 inexcat cores with 0,25% error rate. Erroneus computing needs “non blocking data structures” or other solution that small mistakes don t crash whole program, program is in small snippets that are stored in memory when processing so when error comes small snippet can immediadly rerun and not crash computer etc. Or use added error correction code in computer programs and hardware error correcting encoder/decoder. iSlate India is tablet PC with inexact processor and uses solar power to work). Data compression using floating point accuracy: it is possible that standard 64 bit FP number has about 2000 bits (mantissa) accuracy using software computing. If now first 64 bits of this 2000 bit number s accuracy is used to represent ANOTHER FP number, now one FP number is two FP numbers with about 2000 and 1936 bits of mantissa accuracy, and so on: 10 X 64 bit FP numbers occupy first 640 bits of one 2000 bit FP number s accuracy, still about 1360 or more bits for information left for FP numbers in “back row” and about 1900 bits or less in “front row”. Compression ratio is massive, and information can be text or picture or video in these FP numbers (bits of mantissa precision). But computing 2000 bit FP number will take for ages in phone processor etc. But it is the best compression method. Actually using unum computing on the hardware that increases FP accuracy is better way to use “several FP numbers inside of one (one mantissa with extented precision). Unum computing is faster than software computing mantissa expansion and can perhaps lead to similar accuracy, so one IEEE FP number when using unum extented accuracy can have several other IEEE unum numbers inside of its extented (mantissa) accuracy, these can have their own FP numbers inside their extented accuracy etc, until accuracy potential has been spend, and one (unum) FP number now holds hundreads or thousands of other FP numbers inside itself. For sound using High Frequency Replication for frequencies over 7 khz, new video compression methods, “Video compression using colour quantization and dither” Raja, Unicode table compression in FOX”, Opera unicode compression, “Text compression and superfast searching” Khurana 2005, VSEncoding, zhen.org/blog “Go vs. Java: decoding billions of integers per second”, “Parquet encoding definitions”, Quote notation number system, Q number format, Bounded integer sequence encoding BISE, Zero displacement ternary number system ZDTNS, order-3 tribonacci code (Sayood), “A new number system for faster integer multiplication” Hashemian 1996, “Between fixed and floating point by Dr. Gary Ray” chipdesignmag.com, “floating-point division and square root” Viitanen 2013. Or then abandon integer base altogether and use some exotic number formats like Magical skew number system (Elmesry, Jansen, Katajainen) or Munafo PT system and multiple base composite integer at MROB.com, Balanced ternary Tau BTTS, or Paul Tarau s number systems. etc. Using dithering to maximum effect: all information, text, picture pixel and hertz of sound is encoded with only 1- bit, this 1 bit is dithered differential (delta) value. Dithered to 4 or 5 bit precision (with or without sign bit) Additive Quantization AQ vector compression (1 bit dithered to 4 bit, then 4 bits used for AQ, even more information can be stored in 1 bit) or logarithmic (4bit log is 7 bit integer). Or 1 bit dithered to 8 or 9 bits, dithering noise affects information making it partially erroneus, AQ used again or logarithmic (8 bit log is 14 bit integer) or floating point 8 bit with sign bit, delta (differential) values perhaps not needed anymore in AQ and logarithmic. Or 10 bit FP plus sign bit OpenGL format number, needs 2 bits to be succesful, otherwise dithering noise very loud. “Gal s accuracy tables revisited” method can be used in logarithmic information, boosting 4 bits to 14 bits max precision. Also “metallic numbers” at neuraloutlet.com, another logarithmic. “Dither and data compression”, “Optimal subtractive dither for near-lossless compression”, "Improvements related to data compression " 1993 Kirk, “Nonuniform dithered quantization” 2009, “Data compression with low distortion and finite blocklength” 2017. Analog circuits are made at 28nm or even smaller. Combining analog dither with 1-bit digital signal in 1 bit analog/digital circuit, then when dither expands this 1 bit to 4 or 8 bit digital values or more if AQ etc. is used, multibit processing is digital, so this DSP has two circuits, 1 bit analog and multibit digital. Magnetic phase memory can be analog also, 1 bit dithered. “A VLSI analog computer/math co-processor” Cowan 2005. Analog processing math / DSP can be superior to digital. Trachtenberg speed system is collection of simple algorithms for computation, improved by V. Jadhav among other his simple math algorithms. If similar simple number shifting trics are used in hardware then actual computing could be avoided so even inexcat processor can get right answer, only when algorithms does not offer simple solution computing must use processor s 0,25% inexact computing integer ALU or FP. Sound is 1 bit dithered so it is DPCM, delta values for picture pixel1 bit dithered also, and text, unicode or other text format. Even magnetic memory can be roll printed, but very large and slow. So large memory cards and memory slots inside PC and router is needed if roll printed slow memory is used. Router perhaps needs even rackmount memory space for super large but slow and cheap roll printed memory, and tablet PC size 86 X 54 X 5 mm standard PC card slot for swappable memory card. This few dollar device is given for free together with zeronet connection, because net connection costs more in yearly basis or many year basis than phone or tablet PC, that is used many years. If other commercial sponsors join zeronet “advertisement campaign”, then new better model of device is given to user and older given away or thrown away, so this is “buy, play, throw away” cheap tech (commercial sponsors buy the device). At least semiconductor manufacturers would welcome this kind of massive new use of processors in scale of billions instead of millions. Operating system that uses emulator to use old PC programs of 1980s (abondonware) and OS perhaps (old) Symbian and Linux (Tukaani Linux? etc. based) together and Android Go also, all three are based on Linux (Symbian and Linux can be swapped in OS so three operating systems Android Go, Linux mobile and Symbian is in same phone. Old Linux x86 programs used through emulator. Old Windows Mobile that is simple and Windows CE based that needs only 1megabytes OS memory in simplest form is also one option, using emulator for ARM processor or new cheap chinese x86 processors). Unikernel (Mirage) is one OS concept but very different than OSes today. Device can also be like Android TV stick with cheapest video glasses, the small stick PC has small thumb keyboard that uses front and backside of device and very small black and white display. Almost all data goes to video glasses connected with wire to stick PC. Speech synth is a must, because people cannot read or write, hardware speech synth either simple or more complex like Yamaha Pocket Miku. Also cheap integrated chip analog TV and radio is needed in zeronet device, similar that are being sold with chinese cellphones. TV, radio and speech synth perhaps can be done using printed electronics. If technical solutions for cheap internet connections exists, and cost of hardware is as small that it is today (Allwinner A33 quad-core SoC cost 4 dollars in 2015), and advertisement money that goes to internet advertising is in steady rise, why not open “zeronet” to billions of people who do not have money to pay for internet, and open it now? Even without internet connection this device with large data compressed memory (cheap ROM and RAM) with ongoing inbuild “advertisement campaign” can be used as PDA (personal digital assistant) with useful “info pack” and thousands of games inside. For PC sized devices 2,5 inch hard drive size memory card that is plugged to laptop PC like memory card but size is 2,5 inch HDD, PC can now have slots for either HDD or large and slow but cheap magnetic chip memory. Internet suite is old concept, not just basic browser, but not operating system either, how internet suite suits for zeronet I don t know. Even without internet connection or not even connection to nearest router this zeronet device (commercially sponsored) would be tremendously useful. With even strictly limited “zeronet” connection it would be even more useful. If in africa only 9% of people today have internet connection, if it would be offered for free to 100% coverage now teleoperator firms can have only 9% of profit margin of today and still collect the same amount of money due to much larger customer base. Actually they get more profit than that because those that use commercial internet today continue to use it because zeronet is so slow and restricted. There is Spotify and video service Machinima offering content for free. They use material that has royalty payments but still manage to offer that material for free. Spotify has not made profit, has been years in operation, but still Spotify is considered as big success and has almost monopoly in its business sector. Internet in poor countries need similar “netify” service that offers people of third world internet for free of any cost (“slimmed down” zeronet, with no subscription line costs that teleoperators charge) and uses roughly similar business model like Spotify or Ecosia browser (but instead of planting trees money is used to cost free internet connection), those apps that offer only partial solution, like Sliide Airtime or Datawind UbiSurfer, are only partial solutions. Microsoft Rewards gives money (points) to those that use their browser, and offers 650$ to those that change Macbook to Surface. Using 650$ could 100 - 200 people in poor countries get cheap internet phone and Microsoft 100 or 200 new customers instead of just 1 in rich countries, and Microsoft Rewards in third world would bring much more people to their browser than in rich countries. Large datacenters are not in africa or south america, so slightly distributed model of internet is best, projects like Meshnet, Google groups: Building a distributed decentralized internet, “IPFS is the distributed web”, “Remote I/O: fast access to distant storage” Foster, “Realm: performance portability through composable asynchrony” 2016, “ECO: efficient collective operations of communication” 1996, “Towards parallel computing on the internet: applications” Sundarajan 2006 section “models for wide area network”, “Silicon Galaxy” system area network for collective communications, and “The case for network coding for collective communication on HPC” 2015, last two are supercomputing not internet, but similar to web. In the 1980s were “ROM pack” memory cartridges used, but RAM memory cards replaced them. Cheap mask ROM memory card as an option (with RAM memory card also in the phone) would be very cheap, but non-erasable. However cheap price/capacity is more important than RAM writeoff sometimes. What this device (phone, tablet PC, or video glasses + TV stick PC) needs is common enviroment, zeronet, where all different partial solutions for cheap internet can come together and form cost free internet for development countries. Less than normal telebusiness tax is perhaps possible for zeronet connections and devices so net connections can operate cheaper, this is almost charity operation anyway so tax levels should be low too, or nonexant if zeronet does not bring profit in first phase (year or months) when it starts (but it is aimed to make profit after some time, like Spotify). Cost free internet to customers in third world, commercially sponsored, is propably possible already, using all new internet cost reduction techniques, if these would be together. Even few megabytes per day is enough, if it is free of cost and data compressed efficiently. But those who have money to spend will use regular paid internet in third world, because zeronet is so restricted, so it is no threat to regular internet, it just offers net to those who cannot pay for it and do not use internet nowadays, and will not use net in the future, if it is not cost free (because they have no money to pay), so some form of cost-free option, without subscription line payments etc. is only chance to get them to internet.
Because now is possible to make cheap printed electronics, either using roll printing or inkjet or laser printer, that can be used to make electronic devices for third world. Medical / health care and agriculture / food production can use almost no cost electronics, like printed electronic thermometer that cost almost nothing, in third world. Because they cost almost nothing they can be given away for free by aid organisations etc. For consumer electronics there can be three different devices: personal digital assistant roll printed, that has no internet connection, only contact is physical contact with another device. It looks like Android TV stick or very small pocket calculator, and audio is by headphones only, and visual information using simplest form of video glasses made like Google cardboard virtual glasses. No display in device itself. Headphones (acoustic electric elements) and video glasses (black&white display) are made with printed electronics also. Battery is made using printed electronics, and touch keyboard and solar panel also that brings power to device. CPU is roll printed, if it is made at western factory hybrid printing with cheapest silicon memory (Dramexchange.com has “flash memory project price” 4,8 dollars cheapest for 128 gigabytes, so 0,0375 dollars is for 1 GB and about 300 megabytes is 0,01 dollars) can be used in CPU cache and for other memory, instead of RAM mass memory is cheapest mask ROM mostly and little RAM. If CPU is made at third world using inkjet printer or book printing machine also memory is roll printed. Cheapest “memory card” can be used in some devices (it is mask ROM, not RAM, mostly). There can be several different devices so if audio and video are headphones and video glasses only that can makes things cheaper, and solar panel only is separate thing. These can be plugged to device to another. Also some may lack battery so electricity is from direct contact to another device s battery or solar panel. Touch keyboard, printed electronics made, is multi-tap or chorded keypad type, only 6-7 keys are needed in chorded keypad, so Android TV stick shape device can have simple keyboard. Electronics is moulded on plastic like microchip package, battery and innards can not be changed, plastic mould is the surface of the device. So anot her option for device shape is like PC microprocessor in its plastic package, but in this case “microprocessor” is the whole device, CPU mmeory and battery etc. packed to cheapest SoC imaginable, cheap printed or otherwise cheap circuit and battery in shape and size of large PC processor, one side has 6-9 key multitap or chorded keypad, simple touch keyboard, and pressing those 9 capasitave membrane keys is only way to control device, video glasses and headphones are plugged directly to processor mould when used and solar panel also. So in this case “microprocessor” or SoC that otherwise is planted on PCB board, is the whole device itself, no PCB board or addtional package is needed for SoC, The SoC is itself a package, and membrane keypad on one side of chip is the controlling means and simple flat headphone jacks or video glasses are plugged to plastic mould of processor directly, and even small flat “memory card” can be used for additional memory if needed. This “microprocessor” can be put to pocket and user may have several different cheap 0,03 - 0,05 dollar price devices, one comes with ARM processor, another has x86, third C-RAM, fourth is additional GPU, fifth is another processor which is new structure and comes with modern new operating system, sixth is DSP, etc. Price is so low that these devices can be given away for free, like chepest headphones, video glasses, and solar panel, paid by aid organisations or commercial sponsors. Altough they are cheap, using them is possible to watch video, listen to music, play computer games, use it as PC etc. using clumsy 6-9 key keypad chorded keyboard or multitapping. Altough operating systems and games are 20 - 30 years old, concerning the usability of these devices, 20 year old Windows, Linux or early Symbian is perhaps OS for ARM or x86 devices. Electric cables needed for headphones, video glasses and solar panel are printed also, they are flat, not square, and electric contact plugs flat also, few millimeter wide. Cost minus CPU and memory is perhaps 0,01 dollars for battery, less than 0,01 dollar for keypad, solar panel 0,01 dollar, video glasses 0,01 dollar, headphones 0,01 dollar. CPU is 0,01 dollar, memory 0,01 dollar also. Together about 0,03 dollar device, with 0,03 dollar video glasses, headphones and solar panel. Price is so cheap that these devices can be given away for free by aid organisations or commercial sponsors. CPU licensing is problem, licence cost is much higher than electronics printing cost. But because latest tech is PragmaticIC 1 micron process for transistors, 1 micron is 1000 nm, those printed CPUs are much slower and larger than silicon chips, so 20 year old designs can be used. Electric power usage is concern, so mobile version of processors can be used. But because manufacturing volumes are much larger than silicin chips, for example A33 SoC cost 4 dollars, and Intel Atom quad-core tablet PC chip 5 dollars couple a years ago, about 1 dollar is perhaps license price for them. Manufacturing yields are about 10 000, or 100 000 tablet PCs or phones per manufacturer and model at largest, if not extremely popular model. But printed electronics can be made at scale of million or over at ease, so this 100-fold or 1000-fold increase in manufacturing brings similar amount of money to CPU license holder. And roll printed CPUs are about thousand times or more larger and slower than silicon chips, so they are thousand times less useful, so that way also license cost must be lower than ordinary silicon chip. License price can be tied to processor size, so if printed version is 100 times larger and slower it has 1/100th license cost and 1000 times larger 1/1000th license cost. Usability of large printed chips is so limited that license must be cheaper than ordinary silicon chip, however license holder can get same amount of money from printed electronics version than original silicon version because manufactering is much faster and bigger mass production than silicon chips. However printed versions are enormously slower and larger and worse than silicon version, however amount of profit earned is the same, even using 1/100th or 1/1000th license payment. These PDA style devices can be used as MP3 and video player, and typical PC duties, as game console etc, but without internet connection. Inkjet printers have 14400dpi accuracy In Epson Stylus and Screenjet 3100, meaning 1,76 micron. Laser printers have 9600 X 9600 dpi accuracy, 2,65 microns, offset book printing machines 600lpi, meaning perhaps 16 color cells per line, so 9600 dpi is accuracy, gravure and flexigraph book printers have higher accuracy, and some chinese factories advertise 14400 dpi book printing machines or package printing machines. If printed electronics is manufactured at third world country book factory, perhaps 2,3 or 2,5-2,55 micron transistor dimension is possible, and 3,5 micron if inkjet printed in some office in third world cheaply. Agfa has patented US pat 8232043B2 “Method of making lithographic printing plate” litho printer that uses 350-390nm laser, that is smaller than 1000nm printed electronics accuracy, but for book printing. However litho press can be changed to printed electronics production instead of books and pictures. Second kind of device can be “internet receiver”, that has silicon CPU and GPU but everything else printed electronics, including colour display, and is like cell phone or tablet PC. But unlike them it can only receive information, not transreceive. This can be solution for using internet without pay, it is like radio or TV, and device has inbuild analog TV and radio like chinese cellphones, or digital radio also or digital TV (if third world countries have digital TV or radio ransmissions), and radio frequency receiver for internet. It cannot however send signals. Because it only receives, it can be used for free like radio. It scans multicast internet transmissions, multicast is one frequency transmission that can received by millions of people like radio or TV transmissions, in internet. Using NVOD (Near video on demand) style transmission, even Wikipedia with its 5,5 million pages can be used. 5,5 million pages has each 0,01 second time slot, usingdata compression and no pictures, dictionary is send toreceiver, when right page apperas it is automatically downloaded by receiver when right page mark is in the dictionary list, 0,01 sec makes 12 hours for 5 million pages, using 40-channel NVOD that is 0,3 hours, approx. 0-0,3 hours and avarage 0,15 hour (9 mins) to wait until searched Wikipedia page arrives. There can hundreads or thousands multicast internet mobile pages, some of them using NVOD if such as Wikipedia has large number of pages. If intermet connection can t be used then analog radio transmission can be used. Before analog modem internet connections ended they achieved 1 megabit/ second speed using data compression. Using analog audio channel transmission in ordinary radio, internet can be received. Public radio broadcasts in third world countries can include several channels dedicated to internet transmission. Even better would be analog TV transmissions, on e Tv channel uses several megaherz, so large amount of internet content can be transmitted. SECAM / OIRT TV is used in many ex-communist countries and it uses large bandwidth. Digital radio or TV would be even better internet transmission platform, but few third world countries have them. However nowadays even poorest countries have 4G (LTE, Wifi) internet connections, and 5G is coming with its gigabit/sec speed, so using radio or TV channels to multicast internet seems futile. Internet receiver is like ordinary radio, it only scans channels like radio, and those channels reach every internet receiver using same frequency and time slot shared with millions of users. Radio frequency communication: not perhaps nothing to with previous but US pat 5313457A1 “Code position modulation system”, WO 2002065640A1 “Digital modulation device” Walker, “Multiple pulse per group keying” Charles Hartmann. Price of internet receiver is few dollars. Third device is internet transreceiver, ordinary cell phone or tablet PC, it comes with free internet connection and is commercially sponsored and is given free also because net connection is more expensive than device itself. Internet receiver can be sold, because it costs only few dollars and does not require paid net connection. It even cannot “connect” to internet, it just “listens” to it. So it is like radio, free of charge. There can easily be thousands of multicast channels, some like Wikipedia need perhaps few dozen NVOD-style channels if they have large amount of information, and other way around several net pages can use one multicast chanels if only 0,01 seconds is avarage page time. No audio or video (except regular internet TV and radio channels that are provided for free) and perhaps not even still pictures, only data compressed text. Allwinner A33 SoC cost 4 dollars 2015, single board Arduino type computers use it, price 8 dollars about. Now cheapest are 7 dollars (processor SoC 3,5 dollars?). Pine 64 uses 64 bit quad-core processor, price 15 dollars when introduced. So perhaps 64 bit quad-core processor cost 7,5 dollars then? If price if dropped like cheaper single board computer (about 20%) now processor SoC perhaps cost 6 dollars? So internet receiver can use 3,5 dollar 32 bit quad core SoC (without internet transreceiver, only receiver is used, making SoC cheaper), and free net “zeronet” transreceiver can use 6 dollar(?) 64 bit quad SoC. Both devices look like tablet PC, but display, touchscreen and battery is made using prunted electronics. Receiver is about 4-5 dollar price, and transreceiver (regular tablet PC) about 6,5 - 8 dollars, but paid by commercial sponsors. Operating systems are ordinary Android or Linux mobile, or Windows (Windows Surface Phone? Windows Phone? Windows Mobile?) for processors that are ARM or x86, and x86 not necessirely Intel Atom or Quark, Intel and AMD have licensed their x86 tech to chinese manufactures, and IBM its POWER processor. Because batteries, solar panel, touch keyboard (not necessiraly touchscreen), headphone acoustic elements and display can be made using coarse printed electronics, they are propably cheap. Cheapest magnetic memory available like mask ROM or cheapest slow RAM can be used as memory, and memory amount is low anyhow in these cheap devices. Only mermory and CPU and GPU is ordinary silicon chip. However cheapest, almost no cost PDA device has everything made to cheapest possible way, so even CPU can be roll printed, even roll printed using almost standard inkjet printer in third world country itself. Only way to share information using PDA or internet receiver physical contact to another device or"information loading station" where plugging device to loading station information can be changed between station users and loaded. Information station can be in every village, if not, every town at least. Altough internet receiver and transreceiver use normal ARM or x86 chips, PDA and other printed CPU or printed GPU devices can use 20 year old ARM or x86 or modern version with license (such as VIA Nano, Intel Atom /Quark, ARM A-55), but also some other, Linux /Windows compatible are Transmeta Efficeon, Elbrus, Kalray, Loongson and Cell and Mill processor. Some other like old Lisp machine (TI), Apollo PRISM or TRON/BTRON, old Cray processor 20-30 years ago if they are more efficient than x86 from same age. Transputer, systolic array, stack machine, dataflow, TTA, unikernel, Synthesis kernel (Massalin),"NISD: a framework for automatic narrow instruction set ", lambda calculus directly in hardware (inkjet printed hardware description language, SECD, CAM), hardware code abstraction, GreenArrays, Knupath Lambda Fabric, Rex Computing REX 256, GRVI Phalanx, Xputer HPRC are new concepts, and Warp processor too (that is different from old Warp machine, and “Hardware-based design flow” 2011 Altera floating point method suitable for Warp processors and other). All these need different programming and operating systems than normal chips. Large and slow printed electronic chips can be made faster using inexact computing (Rice university) or making processor asynchronous. Asynchronous and inexact principle can be combined, asynchronous processing corrects mistakes of inexact computing and rerun parts of program that are erratic. 0,25% mistake is suitable for CPU if speed and die size is improved, and 8% for GPU and signal processing, altough audio signal with 8% error rate is perhaps quite bad, but signal processor can leave errors incorrected, CPU need correction like memory circuits that have own error correction if inexact processing is used. Because GPU is more inexact than CPU and so more transistor efficient, mobile SoC can have large GPU and small CPU, otherway around than nowadays, and programs use both GPU and CPU capabilities simultaneusly (GPGPU computing), not only CPU and separate graphics only GPU like nowadays. Even unified CPU/GPU like UPU Harmony can be used. Inexact computing suits better for established x86 or ARM designs, almost normal programs can be used, asynchronous processing needs totally differint operating system. If CPU is other than ARM or x86, new concepts (inexact & asynchronous, unum computing etc.) can be used. Operating system must be so new for them too, like those aimed for internet of things, some fancy OS concepts are planned for IoT. Or Argo OS, supercomputing OS, for mobile phones. It is for supercomputers, Linux based, but has asynchronous and error correcxting capabilty and parallel processing for multicore processors. Thee was something called “Green processor” academic research project years ago, if I remember the name right (Green secure processor?) for low power chip. Cheap magnetic memory is needed, dramexchange.com has 4,8 dollars 128 GB project flash memory price, so 0,0375 dollars for GB and 0,01 dollar price for 300 megabytes. That cheap memory can be used in hybrid roll printing as cache memory etc. and also as regular memory. But even cheaper is magnetic tape memory for large mass memory. If regular LTO data cartridge is made as PC standard, every desktop PC has LTO tape drive, with take up tape roll inside PC because LTO is one reel format, 6 terabytes is possible to store in tape, 48 TB in future and 154 TB in Fujifilm Nanocube. If LTO tape standard is made available in every desktop PC, tape prices go to same level as old VHS videocassette prices, if manufactured millions of units (like video cassettes). For max 6 -154 terabytes of memory per cassette. Also tape memory can be improved using same hybrid memory principle as hard drives, smaller 1 terabyte amount of RAM (1 terabyte of cheapest and slowest flash RAM cost about 40 dollars) can be used with 6-154 TB tape memory. LTO Accelis two reel cassette instead of single reel can be used, but using standard normal LTO cartridge tape reel size, so LTO Accelis two reel cassette would be about 110 -115 mm X 150 mm size, about the size of old Betamax video cassette. But both LTO one and two reel version cassettes should be thinner, about 16mm, if 12,7mm is the tape width, when now LTO is 21,5mm, that s too thick cassette for home PC. One reel LTO cartridge fits inside home PC and two reel version should fit also in PC towers. Smaller two reel cassette using half of 960 meters LTO tape legth, and half of tape width from 12,7 mm to 6,35 mm, this is for small laptop and even tablet PCs. Data capacity is 1/4 of bigger cassette formats, so from 1,5 TB to 38 TB. Accelis two reel format is faster but lower data capacity than standard LTO. smaller 6,35 mm tape makes also tape heads cheaper I think. But 8mm tape can be used if 6,35 mm tape is not produced anymore, with 26% more capacity. Cassette size can be 80 - 87 mm X 107 mm, only slightly larger than C-cassette which is 64 X 101 mm. 480 metres of LTO tape fits in that casette in one reel of two reel cassette, one reel LTO cassette has 960 metres max. Cassette thickness can be 8,7 - 9 mm with 6,35mm tape, even slimmer than C-cassette that is 9,5 mm. But even slimmer version can be made for mobile phones, using previous cassette size but only 3,81 mm tape width cassette can be only 6 - 6,4 mm thick. Altough it has 15 - 20 mm more height than C-cassette it has almost same length and it is thinner. Capacity is slightly over 15 % of LTO single reel cartridge, so from 900 gigabytes to 23 terabytes. Even smaller can be 480 m tape one reel micro-LTO tape width 3,81mm and 75 X 76mm cassette size, take up reel is inside mobile phone. The narrower the tape width is more cheaper tape reading mechanisms (tape heads) can be used I think. But tape heads, electric motors etc. consume power, are more expensive than simple memory card RAM memory slot in phone or PC, so cheap phones and PCs use regular memory card to save device cost, however magnetic tape manufactured mass production like C-cassettes is cheap making with cheap hybrid memory (tape together with flash RAM) for example virtual reality possible in mobile phones using virtual glasses, 40 dollars is cheapest terabyte RAM. So LTO cassette prices can be about same price range as empty C-cassette in its heyday, if manufactured in similar scale (millions of units). And LTO now has 6 terabytes (or 1,5 TB previous generation), in future will be 48 TB and using Fujifilm Nanocube 154 TB. However cheapest phones and PCs don t use tape memory, because tape mechanism is bulky and more expensive than memory card slot. A way to make processor cheaper is computational RAM (C-RAM) used processor-in-memory (PIM) devices like Venray and Micron Automata, that C-RAM tecnology makes possible even silicon chip fast processor to be made at same price as printed PDA, but operating system and programming is diferent than regular PC. However because meory price is so low nowadys, 128 bigabytes is an awful lot of transitors for price of only 4,8 dollars. C-RAM in “dram-optimized” form, using cheap memory facturing as logic IC. Billions of dollars are spend in printed electronics research and that have large publicity. But nobody notices that is possible to make super cheap ICs like printed electronics, but much smaller than 1000 nanometer of nowadays printed electronics. However price is same. C-RAM chip at 90nm is 120 times smaller and 120 times faster than 1000nm printed electronics. 65nm C-RAM is 240 times faster and 240 times smaller. That is enormous advantage. But price is almost same as printed electronics. Why nobody notice it? So making super cheap electronics C-RAM in dram optimized form is much better than printed electronics, that gets all the publicity. Even cheaper should be mask ROM (MROM). If CPU is manufactured at mask ROM factory instead of regular CPU factory, MROM transitors are much simpler than regular CPU, and simplicity of MROM structure puts restrains to design, that must be simplified from regular CPU. But still flash RAM or mask ROM price is so much cheaper than regular CPU per transistor, and in same class to almost no cost per unit printed electronics. And no new factories and manufacturing techniques need to be invented like printed electronics, factories do already exist. Speed and size is so enermously better in C-RAM or “computational MROM”. “C-MROM” is not memory chip at all, it is just CPU chip that is manufactured at mask ROM factory. C-MROM must be simple logic, no match for similar size regular CPU, but price is much cheaper, and price in same class to printed electronics. Why all publicity goes to latter when C-RAM or C-MROM offers so much more speed and shrink? At same ultra low price. In third world market is need for ultra-cheap electronics so C-RAM and C-MROM (a term I invented, because I don t know what else it should be called) is suitable for their needs. As long printed electronics does not reach similar shrink levels C-RAM and C-MROM should be preferred method for ultra-cheap electronics. Regular CPUs have “MROM units” and that like cheap flash RAM can be used with hybrid printing with printed electronics. C-RAM has been studied in processor-in-memory (PIM) and other ways, most interesting is DRAM optimized super cheap version, and C-MROM can be made also, but C-MROM is like ordinary CPU made according to MROM manufacturing principles, so simplified transitors and less layers etc, and simplified logic. If RAM price is only 0,01 dollar for 300 megabytes, that is lots of transistors for such as small price, same as printed electronics but much faster and smaller, and if they can be turned to logic (C-RAM) that would be much higher efficiency than printed electronics. Nobody seems to notice that, altough IoT and cheap electronics is what is wanted nowadays. MROM should be even cheaper, and more easily turned to logic chip (MROM memory is not turned to logic, but CPU manufactured among same cheap principles that make MROM chip). So fast and ultra-cheap electronics is available but nobody seems to be interested. Why? 1 gigabyte at 0.0375 price is lots of transistors almost at no cost, if some of those transistors can be turnet to logic, making CPU or GPU from that gigabyte. Even antique 250nm process from 20 year ago is 16 times faster and 16 times smaller than best printed electronics at 1000nm. There is huge difference between 16nm silicon manufacturing and 1000nm printed electronics, C-RAM and C-MROM can fill that gap, and price range difference also between printed and silicon electronics, but C-RAM and C-MROM can be made at same price as printed electronics, so its hugely more effective option. India is opening silicon fabs at 90 and 65nm, if chips can be ordered there at cheaper price than chinese 90 or 65 nm chips. 110nm, 130nm, 180nm etc is still much faster than printed electronics, and 350nm also. No matter if silicon chip is slow, only thing to worry is price and power consumption in third world market super cheap devices. Manufacturing license can be some sort of shareware, free license cost, etc, like Parallel Ultra Low Processor PULP, ePUMA, J-Core etc., or paid but very small price like 0,01 dollar for processor at max for super cheap chips. Software license also free, like old abondonware from 20-30 years ago, those processors have such small capacity anyway. Printed electronics is needed if ARM or x86 is made, C-RAM and C-MROM need different programming and OS so software must be written for them. License can be for software similar to hardware, very small, 1000 dollars for unrestricted amaount of manufacture, million devices with that program only has 0,001 dollar licence payment, billion devices can have million dollar payment for software usage etc. MROM research has ceased when MROM became unfashionable, but some innovations is made: “The mask ROM and method for making same” US pat 20050070060A is three diensional MROM and EEPROM (like primitive Hybrid Memory Cube?), US pat 8606193A “DRAM and MROM cells with similar structure” 1994, RAM and MROM memory made at same manufacturing process to same chip, Philips “JMROM” theoretical concept “High speed fast cell mask ROM”, and CSMC ltd patents like WO2013086908A1 and Macronix patents. But because printed electronics is free of size limitations super large, even square meter class “macroprocessors” can be made. If Cell processor (Playstation) or Kalray 256-core is made using printed electronics it would be huge size. And huge slow compared to silicon version. But processor can be folded like hankerchief to about 10 X 15 cm (4 X 6 inch) package, if processor design take care of folding points so that processor folds smoothly. And printing can combine ordinary silicon RAM or ROM memory to printed logic circuits, or even optical componets like mass produced optical fibre, Nufern makes 1,8 fibre as mass production, and 1,3 micron is proposed, so 1,25 -1,8 mass produced optical fiber can boost printed electronics. Even optical computer, digital or analog, can be made using roll printing and 3D structure like Hybrid Memory Cube stacking fibres on top of each other for many layers, using polarton lasers.Also analog electronics can be roll printed, “Configurable analog chip” (J. Hasler) for example, it has low power consumption so it can be made at square meter class, and neuromorphic chips also. Also KLT transform can be used in analog form, and “analog magnetic memory” like pat US 5504699A 1996 with analog computer. Hitachi Ising machine, “The Feynman machine AI” and SSRLabs processor are other concepts. Not only CPU or GPU (or UPU Harmony, combined CPU/GPU), but also DSP, ASIC, FPGA, CGRA (configurable array), microcontroller etc. can be made with C-MROM, C-RAM and printed electronics. VLIW microcontrollers like CEVA-X and Improv Jazz seem to be effective. If Intel Quark has 0,025 watt power consumption, making 100 -200 core Quark processor with about billion transistors, it still has about 2,5 watt power requirement. Other way to make slow processor faster is to use mixed radix- residue- combinatorial, complex- or other number systems instead of straight binary, reduced instruction set residue number system, “Signal processor with reduced combinatorial complexity” etc. And using reconfigurable- or “unconventional computing” (reversible-, chaos-, and stochastic processors). Bipolar transistor / ECL, or all new transistor type designs, other than silicon semiconductor etc. However these are not suitable for legacy processors like ARM or x86. Making purely GPU -based processors (GPU almost without CPU circuits) using OS like TxLinux RTX Vxworks, PTask, GDVE, Yellow Dog Linux, Futhark programming language etc. is possible for parallel processing OS. Using memristor based tech, unikernel instead of normal OS, and super simple integer ALU that uses Trachtenberg speed system improved by V. Jadhav. Trachtenberg system is mental computing so it uses simple algorithms that make even complicated calculations simple using very simple number shifting tricks etc. Those simple algorithms can be used to make decimal ALU that is in the software if not in hardware, decimal computing fast, and computer can have instead of 1- 8 integer ALUs 64 simpler decimal Trachtenberg ALUs that are fast, but however cannot replace completely need for normal integer ALU. Simple microcontrollers like Corvus and ZPU can be made using printed electronics. Ways to improve slow ARM or x86 processors in 1 - 5 micron or larger made, is to make processor slightly inexcat, 0,25% is proposed (Rice university), GPU can be 8% error rate and signal processing elements too. Making signal processing (audio, video, computer graphics) analog instead of digital can also lead to speed improvement, and slow 1 -5 micron made FPU replaced with analog floating point unit. Also reducing cache memory to minimum make transistor count smaller but processor is after that even slower. However if printed electronics processor reach square metre size, smaller transistor count makes processor more manageable and cheaper. If digital floating point unit is used unum computing can be added to FPU. When old legacy ARM or x86 processors or old GPU or DSP designs are not used, multicore parallel processing, asynchronous, Warp processor, Simultaneus sampling data acquisition architectures 2015 etc. methods can be used with totally new CPU or DSP structures and all new operating systems. For mobile phones or tablet PCs costing about 2-8 dollars simplified assembly where no separate plastic case and PCB board is not used but SoC, touchscreen and battery and memory is all wrapped in casted plastic, phone cannot be opened and battery changed because all innards of phone are cast in plastic, cheap battery would be making hole in plastic casing, pour battery materials in and closing hole with plastic, battery is simply a hole in phone s plastic casting that is enclosed and contain battery material. So phone is large “microchip” that has touchscreen, casted in plastic case, but no PCB board is needed if all components are packed inside SiP or PoP package. Only touchscreen, memory card slot and outside contact jacks are not cast under plastic surface. However If price and performance is compared together, C-RAM and C-MROM are most effective, and in wide margin compared to printed electronics or even ordinary silicon chips. So in cases where legacy ARM or x86 processors are not needed devices for third world countries should use super cheap but still effective C-RAM and C-MROM chips. Also IoT should be interested of them, they offer super cheap computing fast and are much faster and smaller than printed electronics at about same almost no-cost price per chip (300 megabytes RAM cost 0,01 dollar at cheapest, and 300 megabytes is lot of transistors for that price). So another option instead of using printed electronics in IoT or cheap electronics for third world countries is DRAM optimized / MROM optimized (meaning manufacturing process uses mask ROM factory production, the chip itself is not memory circuit but CPU made along cheap MROM manufacturing principles) C-RAM and C-MROM circuits that are not only super cheap and at same price level to printed electronics but are also much faster and smaller, efficiency can be hundreads times better using them than printed electronics but price is about the same. Somehow nobody seems to notice that. And Panasonic Aquarea is heat pump system innovation, for third world household heating.
Now when internet of things is coming and those are standards that are cheap to use, so net connection does not cost as much as regular mobile phone or tablet PC connection, data rates are slow (if max. 10 megabits / sec. data rate is “slow”) and standards like LTE-M, NB-IoT and EC-GSM-IoT do not require new hardware (network link modems, only receiving device is needed to be made for IoT receiver), they only use existing mobile network in a different way. So no hardware costs for teleoperator, and using IoT is cheaper than regular telephone / data network. Why not using IoT standards for making telephones and tablet PCs that use IoT standard for audio-, video-, and text communication? Data compression and data rate restrictions (low quality audio and video) is possible, and using IoT communication standard. When net communication is very cheap, and IoT is that, development countries can have free to end user internet communication using IoT standards. Advertisement costs that nowadays cover only portion of costs in development countries internet, perhaps can cover all costs if communication is by cheap IoT standard. And then when user base is expansing (if billions of people that have not connected to internet can now have free internet access) increases money that free internet for third world can use. IoT is cheap communication by nature, so using it with data compression in third world for free internet to end user is possible. And some IoT standards use regular mobile networks and they don t require any new hardware for teleoperator, so cost of operation is very cheap. So telephones and tablet PCs for free internet are like ordinary telephones or tablets, they only connect using IoT standard for communication, not mobile phone standard. Large number of teleoperators operating in the third world are government owned, by some rich industrial country, middle east country or China. If they would offer IoT based no cost to end user net connection in third world that would be huge success, also commercial success. Not only IoT standard is possible but similar low-cost mobile data communication that makes possible cheap internet connection in third world, no cost to end user, advertisement & sponsor paid. This communication standard would be specifially build to low cost transmission, low data rates with data compression. Existing teleoperator link network could be used like those IoT protocols, and phones and tablet PCs can connect with it like they connect internet today. Only software is needed for teleoperator to use this network, because like those IoT standards, this low cost communication requires only suitable software, it uses ordinary radio telecommunication links like those IoT standards. Even that software can be made free for teleoperator to use, but not necessirely open source (it could be closed source, but no cost for that software is teleoperator need to pay). There are attemps to make cost free internet, balloons in stratosphear or satellites, cost is billions of dollars. So if only software is needed to turn ordinary telecommunication network to cheap free internet, all that money is saved. However this low cost internet has data rate restrictions like IoT. But at least it is free to end user. So some organisation that attemps to make free internet for developing countries, could make suitable software for it that uses ordinary telecommunication links like IoT does in low-cost IoT communication. So no new hardware, only software is needed for free internet teleoperator. Data rates would be small because that telecommunication must be advertisement and sponsor paid, and data compressed to maximum. Ordinary phones and tablet PCs can be used as receiving devices, with software or hardware modifications so that they can receive those IoT- style low cost optimized messages from teleoperator s links. Cost of (China or India made) mobile phone or tablet PC is so small nowadays that it is not the cost of phone or tablet PC but internet connection cost that makes majority of people in poor countries impossible to use internet. If only software patch is needed to remove that situation, that software should be used. Also even wired, not only wireless networks could be used, so that wired internet networks like ethernet could operate at low cost mode too in development countries. So “zeronet” cost free to end user internet in development countries can be made using just software modification to existing telecommunications network. Low data rates are inevitable because low cost of operation is the goal, but at least it would be payment free to user. IoT is new emerging industry, free internet for third world could be similar new industry, if only software is needed to turn 4G or 3G network to IoT style low cost low data rate communication, with simple graphics, low quality audio and video. Simple Linux (QT mobile etc.) or Android (Android Brillo IoT-style, or other Android version OS) or even Windows (old Windows mobile OS was based on Windows CE that itself needed only 1 megabyte in smallest form OS) or Symbian. Or simply ordinary Android phone that can be used as Linux - IoT type communication. Data compression (for audio, video and text, and for browser and graphics) is needed in operating system, because data speed is low and data rate restrictions used (per month, -week, -day limited amount of data allowed). But if ordinary 3G and 4G networks can be turned not only IoT but payment free internet networks also, why this opportunity is not used, in third world countries? If only software patch is needed to make it true. If IoT is billion-dollar industry this new free internet in development countries could be even bigger industry and business opportunity.
There has been indication that internet user data has been used in political manipulation, and that internet operators have done things to prevent this in the future. But instead of preventing, it can be allowed in third world if cost free limited internet is opened there. Instead of preventing sale of information for political purposes this kind of information sale should become allowed in third world in order to collect money for free internet costs. Money collected this way can even surpass money collected for advertising purposes. For example Ruissia is interested in political influence in Venezuela, and China in Africa. Instead of Cambridge Analytics - style cover operations politically interesting internet user information could be sold freely, paid in cash. This information can be sold to political parties of that third world country where information is gathered, and also sold to foreign political forces that are interested in political influence in that country. This kind of information is valuable to political forces so it must not be sold cheaply. Money collected this way can be much larger amount than selling information just to commercial (advertising) purposes. Is this the way to make free internet true in third world at last? I have written more about how this kind of politically motivated information sale can be done in text “Free internet for third world” which is in Robin Hood Coop Forums, and things about net neutrality etc. are there also. Selling data collected from internet users in third world for not only commercial but political purposes also can generate enough money that at last cost free limited interned line can be opened in third world, and collecting and selling information for political purposes can collect far more money, many times more, than just collecting information for advertising purposes. Zero rating, no line payment costs for end user, no any cost for user using limited net connection in third world. Selling information not only advertisers but for political forces also can be that combined amount of money that makes possible cost free net in third world. Another way to multiple money collected for free net is to use metasearch engines instead of just one search engine. Some of those collect almost 70 search engines and browsers / webpages together (Searx), and (discontinued?) LIMMZ over 50. Metasearch engines often block ads, but if instead of blocking ads metasearch engine collects ads from all those 50-70 webpages and show them all to those who use that metasearch engine / browser? Because free net uses phones as most used net device, small screen of phone is full of flow of adverts when metaserach engine (which is default and only available search engine in free net) is used. So rendering adverts with only text can make adverts smaller so that several dozen of them fits in phone screen. Collecting ads from 50 -70 search engines or browsers and webpages that metasearch engines use to collect information, then makes number of ads 50-70 times more than using just one search engine. So money collected by using advertisements then multiply with 50 - 70 times more. That is enough to maintain cost free internet in third world. Actually because those lesser known search engines gather less ads than Google, Bing or Firefox, number of ads is not 50 - 70 times more if compared to famous browsers like Google or BIng etc. Those metaserch engines seems to use not only other search engines but browsers and netpages to gather information. So metasearch engine not actually gather 50 - 70 other pure search engines, but 50 - 70 netpages. But collecting ads from all of those pages and sending them all to one customer can increase amount of advertisement money to such level that limited free net is possible. Negative side is enormous flow of ads to single device, altough rendered to text, contionous flow of ads almost cover phone screen. But if it makes free net possible this gathering ads from search engines / pages to one one device should be used. There is also browser Brave, made by founder of Mozilla. It uses 20 search engines, and blocks advertiser s ads and replaces them with its own ads. It however pays 55% advertiser money back to advertiser. It also pays money to people who use Brave browser, in “pay to surf” model, 15% of money it collects from adverts. So Brave browser is ideal for third world free net browser, it is metasearch engine, it pays to advertiser 55% of usual amount of advertisement money and 15% to browser user, and that extra 30% what it gets after that can be used to maintain free net connection. That means Brave can use 30% more money from adverts than usual browser to free net connection in third world. And it pays money to browser user, something that free net users in third world will welcome. So free net in third is not only cost free, it also pays money to people who use it. There is Jana /mCent app and others that use similar principle. Brave uses also own cryptocurrency for transactions, IOTA or Moeda cryptocurrency is best for third world. Altough there are hundread or more search engines, there are just few web rendering engines / layout engines / browser engines (those three are same thing). Actually only one remain, Webkit and its Google version. Only Firefox uses some different, and Firefox is ported to Webkit also (Firefox for iOS). Microsoft, Opera, Google and almost all significant browsers and search engines use Webkit or its Google version, and some small players use some other or their own. Firefox already uses Webkit too in iOS. So metasearch engine can use mostly Webkit in third world for all of its 50 -60 search engine and browser / netpage information gathering. Webkit is like Google, only one browser and browser engine dominate market. This simplifies free intrenet for third world. Free smartphone made as cheaply as possible, with cheap video glasses and cheap data gloves, and cheap headphones and solar cell charger made by roll printing can be given for free to free net user in third world, because manufacturing them costs only few dollars, and actual net line costs are several dollar per year and used in many years, so net line cost is higher that equipment cost. UbiSurfer browser uses data compression and charges only 0,5 dollar per month for net connection. MyPhotoZip is fractal based image compression, so it uses lots of time for encoding but decoding should be fast. Video codec that uses MyPhotoZip for “real” frames and then interpolates frames in between can be truly effective video compression, because image compression ratio is 1500. Video when encoded can be stored using some fast video codec and then in device memory transferred to MyPhotoZip (“real” video frames that video codecs use, they then interpolate frames in between, and no fractal compression is needed in this interpolation). Using maiximum data compression and those money collecting methods free intrenet in third world can be true. One way is to use hybrid of centralized server model with distributed system like distributed filesystems (IPFS) and IOTA / Tangle and distributed search engine or browser like “Dweb page” / IOTA. This distributed model can use LTE Direct for example as partially distributed internet and also centralized link tower / server to phone model at same time. So centralized model can use less resources and be cheaper if part of its jobs is done in distributed way with central servers and link towers having less load than ordinary all centralized model. However this distributed part of free net must also me monitored and censored like other free net. Those who want uncensored material must pay for regular internet. Radeeus and Spotify (earlier) and internet (OTT) GT TV were services that were half centralized server / half disributed model. If only one phone / net device is given free to one family in free net, or whole family, including children if they have old enough, get free device, that is possible if enough money is collected for free net. Or only one device in family has net connection, others given to family are simply build portals (screen and headphones) that connect this one net device with thin wire or wireless. There is so called kids safe browsers, that can be in devices given to children, when they become full age browser changes to normal. Bingle was browser that showed both Google And Bing together in same screen, and there were others, Searchzooka, SearchOpener and ShinySearch. Favitt showed several (5-7 or so) browsers in one screen. Similar several browsers open in same time (with all their adverts) can perhaps be used in free net, however perhaps phone screen simply becomes too cramped and single metasearch engine / browser that shows all adverts (from search engines and netpages it scans for information) in one browser page is best for phones. Metasearch engine is called search aggregator, perhaps adverts aggregator (that collects adverts as much as possible) is needed in cost free inernet for third world. Metasearch engine can collect not only information, but adverts from 50-70 netpages / other browsers and search engines, to fund cost free internet for third world. In Wikipedia is articles about pay to surf-, pay to click-, and pay to surf models how to pay internet time. According to pay to click article already 1997 adretisers paid from 0,005 to 0,25 dollars if user clicked advert. so if in India cheapest monthly subscription line cost is 2,5 dollar or 0,5 dollar in UbiSurfer browser (those data plans have limitations, they are not for unlimited surfing for whole month), only 100 clicks to adverts in UbiSurfer during month or 500 clicks in other lines is enough to bring money for cost free intrenet. Actually now cheapest phone contracts are about 3,5 dollar per month, but still it is about 700 clicks (if 0,005 dollar price is used). So actually cost free net in third world could be started right now, it could have started already in 1997, 22 years ago. I cannot see any reason why those billions of people without net connection are still without net. They can simply click adverts and buy themselves net time, if clicking is needed for example even one click per minute avarage net time, and user must watch/hear short advert, it is still cost free internet for user. Bill T. Gross have made many innovations about commercial use/payment of intrenet. Those people who need free net are poor, but they still have to eat and have clothing and other daily needs, so adverts biased towards cheap food, clothing, educational aid, and cheap consumer electronics / entertainment are used in free third world net. Chinese bulk product manufacturers have real market opportunity to sell their products directly to billions. I have written in Robin Hood Coop Forums article about “Food mailbox, what it is?”, about group buying pools, where people in third world can make mass orders and make dircect factory orders from product makers, bying clothes, consumer electronics and food directly from food suppliers around the world Slimming down internet to save cost (costly / traffic heavy net content removed from free net, no gigabit size files etc, data compression to maximum) makes “zeronet” cheap to use. Cost of electronics is neglible today and even before, earlier chinese manufacturers made wristwatch phones, watch- size phones that were either basic phones but some were also feature phones with Symbian OS. Price was from 1,3 - 2,5 dollar upwards. But when Android feature phones came to sale it drove those small phone watches out of market. Wristwatch size phone has small screen like screens before iPhone, but still large enough to watch video etc altough small. So wristwatch size feature phone is possible, manufactured using roll to roll printed components if they are suitable, silicon procesor SoC, given for free with net connection and comes with roll printed video glasses, headphones and cheap data gloves. So big screen is not needed. If free net user wants to bigger screen “iPhony” iPhone clone can be sold, user must pay to get this phone, that too made from cheapest possible components and not iOS of course but Linux, Android etc. Linux, Android etc can also be in free wristwatch phone delivered with free net connection, billions of units manufactured / delivered. Manufacturing cost is minimal, below one dollar if roll printing is used in some parts, and roll printed solar cell charger or Peltier element charger (using body heat to charge phone battery etc) comes with phone also, separately from phone. If cost problems of cost free net are solved simply that way that user clicks adverts in order to earn internet time, and that makes free internet in third world possible, why it is not used? What is the reason behind this resistance to not to offer cost free internet in third world? Even clicking advert after every minute in internet is acceptable if that makes cost free net true. And during that minute also other ads are shown in phone screen constantly. If 1997 advertisement techniques made this free net possible, why it is not used in past 22 years already? And not even today? For billions of people without net connection, because they have no money to pay for smartphone and after that about 3,5 dollar per month subscription fee (about 40 dollar per year). So cheap internet device can be given them for free with free (limited) net connection in third world. Free net connection would help tremendously people of third world. One advert per minute is avarage rate, if for example 30 minutes net time is needed, then user had to watch 30 ads in a row to have that 30 minutes. Youtube style system where about 7 seconds per ad user must watch and then can skip ad could be used, that means 3,5 minutes watching ads in a row and then user can use 30 minute net time, if 30 minuutes more is needed then 30 ads / 3,5 minutes more watching commercials etc. This in case if one ad click is needed for one net minute. Of course when user uses 30 minutes net time, other adverts are shown in screen all time like normal internet is showing ads, and if video is 30 minutes, in screen is running text adverts sometimes at same time when video is running, etc advertisement techniques on top of that 1 ad per 1 minute. So other adverts are also in use on top of that 1 ad per 1 minute, but user does not have to click them. That 1 advert per minute is avarage time and user can decide what order he/she wants to watch ads, 1 per minute or 5 in a row for 5 minutes net time etc. If not even restricted zeronet, internet where all content that is expensive to net provider is removed, is still too expensive that cost free alternative in third world cannot be offered, then it is possible to lessen data rate offered. Less data means less costs to internet provider. And somewhere is that point where lessened data rate costs meets income from advertisers and commercial sponsors, and that point is the cost free internet starting point. Cost free net has been possible since early 2010s when chinese phone manufactureres started to offer super cheap phones, so cost of electronics has not been problem for past 10 years. Less data rate can be compensated using data compression, very effective versions of data compression for images, video and audio. Even just text can be used, no pictures or sound. And text can be compressed also. UTF32 uses only 21 bits for text. It is 2 million values. If only few hundread values are used for text characters, numbers and other marks, and IDBE text compression character blocks, and rest of 2 million values for complete words and complete sentances in memory library, ROM or RAM. 0,5 million most used complete words and 1,5 million most used complete sentances can be used. The multilingual nature of UTF32 is not used, this 21 bit encoding uses one language only (other languages need their own 21 bit encoding), and it uses only 21 bits not 32 like UTF32. Additional 5 bit “header” can be used if needed in cases where 32 values are used to define is that 21 bit section ending with point or question mark, does it have quotation marks etc additions, so end result is 21 bit encoding with 5+21 bit encoding if needed. That makes text compressed to really small bitrate using 2 million values, and library is so small it fits in phone s memory etc. Or use 24 bit encoding which is 16 million values, but that makes library very big and memory consuming. Wifi phones are simple phones, similar can be used in free net in third world. If cost free internet to end user has been possible in third world already past 10 years, why it is not used? There is cost free apps offered, but altough they are free to download they are advertised. So they have advertisement budget. Because there is millions of apps, free and paid, advertising free apps in free “zeronet” is one way to collect advertisement money for cost free net. Because there is huge number of apps, advertisement money collected showing ads about apps can surpass other ads shown in zeronet, like adverts about food and clothing (most people using free net in third world spent their money in food and then to clothing and after that to cheap consumers products if they have any money left). Free apps with adverts they bring with them can even be pre-installed in phone itself, so no net capacity is needed to transfer ads. In phone is continuing “advertisement campaign” where ads are in phone s own memory and pop up in screen etc, not transferred using internet to phone, saving net capacity. Those ads come with free phone that is given away with free net line, so net line can be partially free of ads, however net line provider collects advertisement money also from those ads inside phone itself and also ads transferred through internet. There is plans to use cars, busses and motorcycles as moving internet links in countryside, and in cities there can be free Wifi spots that are privately owned (not by net line provider) in cafes and other places. There can also be “sneakernet” style internet, like “USB dead drop”, but in this case those fixed “information loading stations” that use direct physical connection to phone, no wireless communication, are either state-owned or net line provider owned. In Cuba monthly sneakernet is provided in many years, and that has worked well. All that to lessen internet provider s costs. So collecting all those methods together and using them together can be solution to cost free net, if advertising money is not enough even in restricted and data rate limited zeronet.