search

Plachtaa / VALL-E-X

An open source implementation of Microsoft's VALL-E X zero-shot TTS model. Demo is available in https://plachtaa.github.io/vallex/
MIT License
7.59k stars 756 forks source link

Generates different intonations, emphasis, etc. on every new inference for same inputs #112

Closed RahulBhalley closed 11 months ago

RahulBhalley commented 11 months ago

Is there some way to control this randomness of intonations, emphasis, etc. with some random seed generator like Huggingface's Stable Diffusion pipelines?

RahulBhalley commented 11 months ago

Btw thank you so much for this project open sourcing! I could've never built this myself. :)

Plachtaa commented 11 months ago

It's basically nucleus sampling as all GPT models do, you can try setting a seed by yourself

RahulBhalley commented 11 months ago

Ok, setting different temperature worked for me.