v2.0.0
·
114 commits
to pytorch-transformers
since this release
- Switched from GPT-NEO to the transformers library, making switching models much easier
- Reduced VRAM usage
- Introduced float precision argument to further reduce VRAM usage
- Switched from YourTTS to xTTS v2, improving voice quality
- Support for multiple alternating speakers for narration and dialogues
Full Changelog: v1.1.0...v2.0.0