KoljaB / AIVoiceChat

Low latency ai companion voice talk in 60 lines of code using faster_whisper and elevenlabs input streaming
238 stars 46 forks source link

Playback voice stops & starts #4

Closed 1030 closed 10 months ago

1030 commented 10 months ago

Hello, this is incredible! I got it set up fairly quickly. It works great, except that the voice only says a few words at a time, then pauses for a few seconds, then continues, then pauses etc.

Would you have any advice for how to eliminate this playback problem? My internet speed is fairly good, so I doubt it's that.

Thanks for all your work on this, it's the best execution I've seen of this idea to date.

KoljaB commented 10 months ago

Just tried it with my 10Mbps connection, it worked smoothly here. Hard to say what goes wrong. Do you upgrade to the latest elevenlabs python library (pip install --upgrade elevenlabs)? The script uses the elevenlabs API exactly like documented. Maybe their servers are overloaded...

1030 commented 10 months ago

I figured out the problem! I was trying to use "eleven_multilingual_v2" which currenly doesn't support streaming! See https://github.com/elevenlabs/elevenlabs-python/issues/114

I switched the model back to monolingual v1 and the stuttering playback fixed!

KoljaB commented 10 months ago

Ah, good to know that. And sad they still dont support it for their most capable model. Will close the issue then, thx for reporting,.