It could work, though you would want to do some trial and error with different algorithms; in particular, you'd want to ensure that the compression doesn't scre

It could work, though you would want to do some trial and error with different algorithms; in particular, you'd want to ensure that the compression doesn't screen out the quality of the call. It would not be a trivial problem to solve, as these compression technologies are designed to be used for standard speech patterns, not (to the outside observer) garbled speech patterns. They normally do an amazing amount of compression while still sounding largely unmodified to your ear. I'd make an encryption algorithm with a passkey that is handkeyed into both adapters, breaks every 100ms of speech into 10ms pieces, and reorders them according to the algorithm. Certain parts of speech would probably still get muddied by the time they were passed in and out of compression. I'm sure someone's already successfully tried it and you can buy their wares somewhere on the internet (maybe with a pit stop at a certain place on their way to you).