search

jianfch / stable-ts

Transcription, forced alignment, and audio indexing with OpenAI's Whisper
MIT License
1.6k stars 177 forks source link

VAD + Demucs memory leak #191

Closed MadMaxNice closed 1 year ago

MadMaxNice commented 1 year ago

When using VAD + Demucs, memory leak Container needs to be restarted every 24 hours image

jianfch commented 1 year ago

Does the memory climb with either one or only when both are used together? It would be helpful if you could provide the script you're using that caused this behavior.

MadMaxNice commented 1 year ago

Unfortunately, I can’t say for sure whether the problem occurs when using VAD + Demucs at the same time, or when using some separate component. I disabled VAD and Demucs, and implemented audio processing through "pyannote VAD". And everything has been fine for a week now

The previously used script is simple: There is a docker container with a deployed API The audio is loaded into it, it is recognized, and a response is returned. Nothing fancy :)