search

mustass / diffusion_models_for_speech

Deep Learning course project repository.
https://kurser.dtu.dk/course/02456
1 stars 0 forks source link

Conditional generation generates too short audio (< 1sec) #27

Closed mustass closed 1 year ago

mustass commented 1 year ago

Subsampling in the collator is obviously bad idea when we generate conditionally.

mustass commented 1 year ago

I am silly