I love the PO-35! It is such a great little device! I had this idea about “soft-bending” its LPC based synthesis engine by feeding it a "hacked" data transfer. For example, we could cross breed the parameters from two recorded sounds, apply some transform or generate the sound data from scratch by using some algorithm. It seems the data transfer uses DQPSK modulation at 7.8K symbols/s, i.e., around 4K bytes/s considering both left/right channels. I guess it wouldn’t be difficult to modulate/demodulate the data, but it would be extremely hard to figure out the exact symbol encoding (phase to bit pair), how the bit streams from both channels are merged, if it contains any parity or data correction bits, the actual format and meaning of each field, including CRC/checksums, sound data, pattern data, etc.

So, I guess I’m writing this on the small chance that SC would be willing/allowed to give some hints about it, please??