Spaces:

---

ayf3

/

numberblocks-one-voice-clean

like 0

Sleeping

I am working on a Numberblocks voice extraction project using pyannote/speaker-diarization-3.1 to extract a specific character's voice from 124 audio files. CPU processing is extremely slow (100+ hours), while GPU can reduce it to 2-4 hours. This is an open-source voice technology research project for voice cloning and TTS research. Code and results will be publicly available on Hugging Face for the community.