Starstruk Logo MarkStarstruk

v0.1.0

lipku/LiveTalkingv0.1.0Feb 23, 2024by yuantuo666

AI Summary

This is the initial stable release of the Amphion project, featuring core implementations for VALLE, VITS, and SVC tasks, along with improved G2P modules and various training recipe enhancements.

Key Highlights

  • Core VALLE implementation
  • VITSSVC implementation
  • NaturalSpeech2 addition
  • Improved G2P LexiconModule
  • Dynamic batch size for VALLE

New Features

  • Core VALLE code
  • VITSSVC implementation
  • NaturalSpeech2
  • G2P module refactor
  • Dynamic batch size
  • Better training recipes
  • Pretrained models and acoustic feature extractors

Full Release Notes

## What's Changed
* update README.md by @lmxue in https://github.com/open-mmlab/Amphion/pull/1
* Amphion Alpha Release by @RMSnow in https://github.com/open-mmlab/Amphion/pull/2


**Full Changelog**: https://github.com/open-mmlab/Amphion/commits/v0.1.0-alpha

## What's Changed
* add core code of valle by @lmxue in https://github.com/open-mmlab/Amphion/pull/4
* Refactor G2P module and related process by @lmxue in https://github.com/open-mmlab/Amphion/pull/5
* Debug VITS for multi-speaker training by @lmxue in https://github.com/open-mmlab/Amphion/pull/6
* Fix bugs for extracting whisper features by @RMSnow in https://github.com/open-mmlab/Amphion/pull/7
* Resume for the SVC's vocalist pretrained ckpt by @RMSnow in https://github.com/open-mmlab/Amphion/pull/8
* Fix tts inference bugs by @lmxue in https://github.com/open-mmlab/Amphion/pull/13
* Fix bugs for FS2 inference on updated phone_extractor by @ChenX17 in https://github.com/open-mmlab/Amphion/pull/12
* Add dynamic batch size for valle by @HeCheng0625 in https://github.com/open-mmlab/Amphion/pull/17
* Revert "Add dynamic batch size for valle" by @lmxue in https://github.com/open-mmlab/Amphion/pull/18
* Add VitsSVC implementation by @viewfinder-annn in https://github.com/open-mmlab/Amphion/pull/14
* Better training recipes for evaluation and vocoder by @VocodexElysium in https://github.com/open-mmlab/Amphion/pull/20
* Add dynamic batch size for VALLE by @HeCheng0625 in https://github.com/open-mmlab/Amphion/pull/19
* Update pretrained tts models by @lmxue in https://github.com/open-mmlab/Amphion/pull/29
* update acoustic feature extractor of TTS by @lmxue in https://github.com/open-mmlab/Amphion/pull/28
* Improve the G2P LexiconModule of TTS by @treya-lin in https://github.com/open-mmlab/Amphion/pull/30
* Add workflow to check code format with black formatter by @BakerBunker in https://github.com/open-mmlab/Amphion/pull/27
* black format on changed files to make workflow work by @lmxue in https://github.com/open-mmlab/Amphion/pull/33
* split and process valid set by @lmxue in https://github.com/open-mmlab/Amphion/pull/25
* Refine Whisper and WeNet Contents Extractor by @Adorable-Qin in https://github.com/open-mmlab/Amphion/pull/32
* Fix VitsSVC model infer bug when using nsfhifigan as generator by @viewfinder-annn in https://github.com/open-mmlab/Amphion/pull/37
* Debug for issue 23 by @ChenX17 in https://github.com/open-mmlab/Amphion/pull/34
* Control the Diffusion SVC Inference Steps in Args by @RMSnow in https://github.com/open-mmlab/Amphion/pull/38
* Add NaturalSpeech2 by @HeCheng0625 in https://github.com/open-mmlab/Amphion/pull/35
* Amphion v0.1 Release by @RMSnow in https://github.com/open-mmlab/Amphion/pull/39

## New Contributors
* @treya-lin made their first contribution in https://github.com/open-mmlab/Amphion/pull/30
* @BakerBunker made their first contribution in https://github.com/open-mmlab/Amphion/pull/27

**Full Changelog**: https://github.com/open-mmlab/Amphion/compare/v0.1.0-alpha...v0.1.0