v2.0

aiola-lab/draxv2.0Mar 31, 2026by lipku

AI Summary

Performed a major architectural refactoring to modularize the codebase, introducing a plugin-based architecture for digital human models, TTS, and streaming.

Key Highlights

  • Refactored code structure to move models into the 'avatars' directory
  • Adopted plugin-based architecture for models, TTS, and streaming
  • Added RTMP output support
  • Integrated Alibaba QwenTTS
  • Consolidated common logic into BaseAvatar and BaseAsr

Breaking Changes

  • Major architectural refactoring to plugin-based system
  • Directory restructuring of core code modules

New Features

  • RTMP output support
  • Alibaba QwenTTS integration

Full Release Notes

1,重构代码,数字人模型和音频特征代码移到avatars目录下
2,数字人模型、tts、传输方式改成plugin方式接入
3,传输方式整理成单独的类,放到streamout目录下,添加rtmp输出
4,数字人推理和回贴公共部分代码整合到BaseAvatar中,各模型只需要实现自己特有部分代码
5,音频特征的切分代码整合到BaseAsr中
6,tts添加阿里云qwentts