From: "Dave Emery" <die@pig.die.com>
I can't quite see how this would work unless the voice was run at a very low level relative to the data.
I believe there are also maximum rates on the data when used with voice (4.8 kbps?) and the modulation doesn't use echo cancellation.
This whole thing has an interesting security consequence - retrieving the voice under data off a crude alligator clip type 2 wire wiretap may be difficult because it requires knowing the data going in both directions. So if one sends random or cryptographically secure pseudo random data one might be able to do a reasonable job of hiding the voice channel from simple wiretapping technology even though it is analog rather than digital. And generating random data with a noise diode and a UART is easy to do and very secure. It is already known that retrieving both streams of data from a two wire tap (voltage only) when neither data stream is known or predictable is difficult or even nearly impossible to do with modern near-end-echo cancelling modem modulations such as V.32 and V.34, and without this capability the voice would almost certainly not be intelligable over the roar of the modem tones. Granted this is not the kind of security one gets from strong encryption, but it is easy to do and makes wiretapping a lot harder.... If I were concerned with securing my calls from most threats other than the TLA's I think spending 640 bucks for two of these modems and a little box with a UART and a noise diode might be a cheap and dirty voice privacy hack. Certainly it would defeat the idly curious and the private eye divorce investigator types... Dave Emery N1PRE (PGP signature out for repair)