Puyuan Peng

You can find the .pth files here https://huggingface.co/pyp1/VoiceCraft/tree/main

Please find details in the paper. We use 4 A40 GPUs and the biggest model took a little over 2 weeks to train

HF spaces is up and running. regarding colab, make sure you rerun the first 2 cells after it restarts

That's expected, you should still be able to run (I just tested it) 

did you click the load model button?

it seems that the encodec model is not downloaded to `pretrained_models`, could you check if it's true? and if https://github.com/jasonppy/VoiceCraft/blob/master/gradio_app.py#L111 is happening?

I'm working on it, in the mean time, please use Gradio through Google Colab https://colab.research.google.com/drive/1IOjpglQyMTO2C3Y94LD9FY0Ocn-RJRg6?usp=sharing

HF spaces is up and running. I uploaded the colab notebook to reflect longer duration supported by newer TTS enhanced models. I personally found that 3~4s is usually enough. for...

Thanks! 830M TTS enhanced and 330M TTS enhanced (to be uploaded) are trained on gigaspeech + lightlight. I recommend using 830M TTS enhanced to evaluate.

> Hi @jasonppy -- I'm curious, if you can spare the details, how exactly did you train the TTS enhanced model compared to the base model? Is it a separate...