Looks interesting. I haven't tried other fundamentally different approaches different than the one I am currently using. I only recently found that my approach is very similar to DTMF, but I use 6 tones instead of 2 and I have sound markers and error correction integrated in the protocol.