@uctumi: Thank you!
@Tobikomi: A short explanation (if you want an excruciating long version, I can message you on IRC or something :) )
1. Record sound off an ST/e running stspeech.tos (since my disk drive is messed up, I just used the STeem emulator.)
2. Downsample to 8000hz, clip to hell and back, and then export as a n 8bit headerless PCM file
3. Load that as a Digidrum in maxYMiser. I can't really speak for vortex tracker or Arkos or whatever else, but they probably have support digidrum as well.
The fun part is trying to cram all that speech into little 6 little 8Kb samples.