How reliably does this work over FM transmission or VoIP? (If it doesn't I would guess and say that you'd have to add some mechanism to detect reliable frequency bands like in existing modems)
In your experience, how reliable is the communication between 2 devices when using speakers and microphone (no cable) at distance ~3 meters? What bandwidth do you typically achieve in such conditions?
Nicely done, I'll note for the readership here that a similar technique works fine at 300 baud across the room using FSK.
Something you might consider is to write a web camera backend for it and use the screen/webcam. That has a lot more bandwidth and has higher SNR as the selector effect of the web camera is so much better than that of a microphone.
FSK across the room? Does that have noise restrictions?
I’ve found all the phase modulation schemes work poorly across the room. DSSS works ok but is very slow. I would be curious how reliable FSK is in that situation.
With 1270Hz/1070Hz you've got over 3 complete cycles to identify the tone at 300 baud, at higher frequency pairs it gets better. The person who wrote it was Tom Lyon at Sun but I bet I could reproduce it with a gnuradio-companion flow graph. That would make it pretty easy for anyone to play with it.
I don't have first hand experience but I believe a big problem with sound channels is multipath propagation/challenging frequency response. FSK naturally deals quite well with those; it's similar to OFDM but more robust due to single carried symbols. You can extend the symbol period as long as you like while using an adequate number. Finally in challenging environments it'd a good idea to use error correction, not just error detection as done here -- you can (in theory) achieve reliable transmission over arbitrarily noisy channels by adding redundancy.
I have experimented with FSK modulation myself and was able to achieve quite reliable transmission across the room [1]. Although the bandwidth is quite low (8-16 B/s), it works most of the time.
Using an audio cable (for better SNR), I have managed to achieve and sustain 60kbps for long periods time - so 3MB file should take ~7 minutes to transmit.
The original goal was transmitting ~10kB in a few seconds over an audio cable, so I didn't try to improve the transmission speed much further.
I guess that adding error correction coding will allow increasing the highest possible bit-rate (currently I'm using only CRC-32 for error detection).
I'm just sitting here and staring at this sentence: "The modulated signal rides on a carrier of a given frequency, but the base band signal got no fixed frequency at all. Because of this, we have the possibility to encode the two-dimensional I/Q signal onto the one-dimensional RF signal without losing anything." I may need to sleep on it...
I guess it should also work on MacOS, since it depends on `numpy` [1] and `portaudio19-dev` [2] Debian packages (which should be installable via Pip/Homebrew).