PhaseSeeker:
I don't mean to be rude but I think you haven't read a single word of the file I sent you because if you go to the voice mode part it is quite clear that the audio is sampled and sent encoded in a digital format to the computer connected to the serial line.
I did read a few words. But I have no interest in learning all about a piece of hardware I will never use so I certainly did not read, and don't intend to read, the whole document. On the other hand if you want to refer me to one or two pages ...
What exactly is G.721, GSM or IMA ADPCM format? Do you mean that the modem sends a series of numbers at one of the standard baud rates (which rate?) and the numbers represent the digital values of the sampled sound?
Post the program you are using for testing.
...R