search

mustass / diffusion_models_for_speech

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

Collator to use all the data #8

Closed mustass closed 1 year ago

mustass commented 1 year ago

Since LJ dataset is not that big, it could be something to consider to add some little code to use all the data and not discarding the ones that are shorter than the audio_len.

panosapos commented 1 year ago

Done by @sandorfoldi, by zeropadding the signal, in order to match the audio_len