[CrypticShift]

> predict the next millisecond of audio based on previous milliseconds of audio

Not milliseconds, but AudioLM [1] already does it with just seconds, for speech (and piano). Results are already very convincing (to me).

[1] https://google-research.github.io/seanet/audiolm/examples/

[nilozd]

yes but I think she's talking about something more like real-time, generating new output as you go through with the input (maybe like slicing windows from a stats. perspective)