Dataset

#3
by yukiarimo - opened

How many total audio files (and what’s the hour count across them) have used to train the model? Is it like LJSpeech-sized?

And is voice real or synthetic (TTS distillation)?

This comment has been hidden

The model is trained on about 120 hours filtered from 200 hours. And no, I did not use the voxcpm2-synthetic-en-v1 dataset; that was generated for a separate project a while ago.

Sign up or log in to comment