search

Camb-ai / MARS5-TTS

MARS5 speech model (TTS) from CAMB.AI
https://www.camb.ai
GNU Affero General Public License v3.0
2.53k stars 206 forks source link

Replicate support #40

Closed platform-kit closed 3 months ago

platform-kit commented 4 months ago

I see there's an official demo of Mars5 on Replicate

https://replicate.com/camb-ai/mars5-tts

Could you add the cog.yaml and predict.py files that were used to create that demo to this repo? That way users like myself who use replicate can riff on the code in a way that makes it easier to share and host demos of results.

arnavmehta7 commented 4 months ago

Hey @platform-kit https://github.com/Camb-ai/MARS5-TTS/pull/41

This PR has been merged and you can find the code under cog folder. There's a TODO to stop uploading the file as a "temporary link", returning the Path for the audio didn't work for us, someone can try taking that up :)

mattt commented 4 months ago

@arnavmehta7 What error were you getting when you returned a Path directly?

arnavmehta7 commented 4 months ago

@mattt It was showing some "failed to decode bytesUrl" when I was using Path.

arnavmehta7 commented 4 months ago 
>>>>> Saving file at /tmp/aud.wav, sr: self.mars5.sr
ⅹ Failed to decode dataurl: missing data prefix

@mattt specific logs if needed

Just checked the linked discussion, I see the issue.

arnavmehta7 commented 4 months ago

@platform-kit Are you still keen to raise a PR with the fix 🙂

RF5 commented 3 months ago

I believe this is fixed now.