It's good to see that someone who is familiar with MusyX sound groups is working on decoding the file format to separate samples and find coefficients/previous.
The actual ADPCM algorithm is already known.
The actual ADPCM algorithm is already known.