OpenVoice
OpenVoice copied to clipboard
HIgh Memory Usage
Hi, Thanks for this great repository. It is amazing work.
My problem is when I initialize OpenVoice's BaseSpeakerTTS, It uses ~3 GiB memory and ~1 GiB video ram. I think that, it consumes too much resources. Do you have any idea to optimize it ?
After experimenting a bit with the provided ipynb files my Vram usages goes above 2.4GB. Less would really be nice especially if this would go up with further requests, but until now I'm impressed with the speed and quality of the speakers. Nicely done :)
This is normal. The pytorch lib does require such memory
Do you know any way to decrease this usage ?