Instead of external things, we could just use our trusty N900 - using bytes-> tones conversion, it could send pre-encrypted data over any analog voice, be it internal GSM modem (aka normal call), or 5$ PSTN phone super-easily modified - replacing speaker and mic with 4-part 2.5mm jack, that we would put into N900 headphone socket. N900's sound output would go to PSTN phone mic input, and PSTN speaker output, would be redirected to N900's mic input.

N900 would take our encrypted (by whatever software we like) data (be it file, voice message, whatever), and convert bytes to tones. Tones would get send via voice link, and transformed into data on another N900, at opposite side of conversation.

In every case, both or one of N900's could get replaced by computer with same FOSS software installed (our data-to-tones converter).

Tones are quite resistant to noise (that's why they were used in early times of telephony, when lines were not so clear), and of course, data->tones transformation would contain adjustable % of "repair data", like some zip archives (in case part of message was translated with mistakes). Benefit of this approach, is that we don't care about integrating encryption into tones - we're just plain sending data as tones, not caring if bad guys are able to decode them into bytes, easily. The thing is, that data is encrypted *before* getting transformed into tones (and decrypted *after* transformation back to data, at receiving side).

It would be rather similar to sending "mail" than real-time conversation - I don't think you can send comprehensive set of tones fast enough, to make it real-time.

Additional benefit, is that we can send it by *any* means of sound transfer - modified PSTN phone is just and example, it can be ham radio, walkie-talkie - whatever. This way, we regain possibility of p2p data transfer, even if whole internet and mobile/PSTN telephony gets shut down (or some of them, selectively). In narrow-band of 8khzsound, we have quite a lot clearly distinct tones, so data->tones transfer could be quite effective.

I'm just not sure, if there already is a FOSS data-> tones converter (as said, it doesn't need to offer any real encryption, just brain dead conversion, as effective/redundant as possible). If there *is*, we're 95% advanced with fully functional project.

/Estel

// Edit

In case of total blackout and need to send data over longer distance - where even establishing relatively clear voice channel isn't possible (but super noisy link, that we can abuse to send *some* sounds) - think ham radio over loooong distance - we could also use reduced set of tones (like 2 or three distinct ones), to achieve same thing, just muuuch slower.

Super extreme case, is when even tones are totally incomprehensible, but *changes* are possible - I know that HAM people use it, sometimes, to talk via morse code, over extremely weak links (changes in noise pattern can be interpreted as morse signals). This could be used to transfer *very* small files, using their HEX representation (or some more efficient conversion to A-Z 1-0 symbols) via morse. We already have text-to-morse-to-led automatic converter - no reason, why it wouldn't be possible (and easy) to make it generate loud noise, instead of flash blinks.

You wouldn't want to send megabytes of data this way, but would still allow to send *data* (sloooooooowly) in some super-extreme situations.

/Estel