Y
Hacker News
new
|
ask
|
show
|
jobs
by
piecerough
593 days ago
It's very related to LLMs. Though instead of text tokens you are working with audio tokens (e.g. from SoundStream). Then you go to audio corpus, instead of text corpus.