README.md

metadata

license: apache-2.0
datasets:
  - CSTR-Edinburgh/vctk
pipeline_tag: audio-to-audio
tags:
  - RVC

• 01/2025: Add beatrice-v2 (CPU friendly) version 30 Epochs (& 800 epochs-perhaps overcooked.)

Trained using Mangio Fork easiergui(rejekts) of RVC WebUI 08/2023:

• Epochs: 250
• Pitch extraction algorithm: harvest if not rmvpe
• Time estimation: 6h
• Dataset: VCTK - Picked Voice-sample -> JarodMica/audiosplitter_whisper for multi-voice short splits, group by speakers.

Usage: You can use the models vc2vc with w-okada/voice-changer.