(transcribed from voice memo, 2020-04-27)
record spoken phrase/percusive beats/etc into sample
detect transients by peak/threshold-following
trigger permutations of cut-up sample playback, eg:
abcdefg
gfedcba
abababa
fedfedf
optionally have multiple recordings of same phrase/beats, to add variety to chose from
- try with web audio
- progressive web app (saveable to homescreen, fullscreen, offline?)