search

huawei-noah / Speech-Backbones

This is the main repository of open-sourced speech technology by Huawei Noah's Ark Lab.
557 stars 115 forks source link

gradio demo #3

Closed AK391 closed 3 years ago

AK391 commented 3 years ago

gradio web demo

CLAassistant commented 3 years ago

CLA assistant check
Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
You have signed the CLA already but the status is still pending? Let us recheck it.

AK391 commented 3 years ago

Is there a way to speed on inference on cpu?

AK391 commented 3 years ago

@ivanvovk

ivanvovk commented 3 years ago

@AK391 The only thing you can try is to set a lower number of timesteps during generation. In that case, expect quality degradation. Try setting temperature to 5 (or even higher) and timesteps to 1-5, and check the quality. 1-3 iterations should be enough to be faster than real-time on CPU.

AK391 commented 3 years ago

@ivanvovk thanks changed temp to 5 and timesteps to 3, it is faster now on cpu with some quality degradation but seems reasonable.

AK391 commented 3 years ago

@ivanvovk trying to add this gradio demo to huggingface spaces, getting this error https://github.com/huawei-noah/Speech-Backbones/issues/4 possibly because the install step for monotonic align is not completing due to permission issues on huggingface, do you know a way around this, thanks

ytyeung commented 3 years ago

This is a separated demo system, better to keep it in a separated repository.