diodiogod d0430d846d Version 5.4.8 ...

Address CosyVoice3 ROCm precision handling (issue #323)

Technical details:
- Run the bundled Qwen LLM with BF16 autocast on ROCm
- Keep the CosyVoice3 flow model and HiFT vocoder in FP32
- Disable unsupported TensorRT loading on ROCm
- Forward LLM worker failures instead of decoding empty tokens

2026-07-13 11:59:53 -03:00

..

impl

Version 5.4.8

2026-07-13 11:59:53 -03:00

__init__.py

Implementing CosyVoice Engine Core

2025-12-17 14:29:02 +01:00

cosyvoice_downloader.py

Add CosyVoice3 model variant support with lazy downloading

2025-12-29 21:52:13 -03:00

cosyvoice.py

Version 4.24.7

2026-03-27 22:05:56 -03:00