Maximizing the usage of PSG channels

Maybe this is a too much maximized example. This is the "Destroy all of them" channel technique :)

Yes, the encoder is simple after all. It approximates each sound chunk by 3 tones located in its 3 frequency maxima. Tones are reproduced by square waves on psg, by sines on scc.
On TT the sampleenvodet is more elaborate in order to use one single channel per sample.
The encoder finds the pitch of each chunk and approximate in the time domain the chunk.

1xPSG:

• 2 Songs
• 16 ISR Samples

  Using:

• 3 music channels
• 1 percusion channel (drums, and other SFx)
• 3 ISR Sample channels

Not too optimized, but hey! only 16kb and works fine.