Starstruk Logo MarkStarstruk

v3.91

jianchang512/pyvideotransv3.91Dec 14, 2025by jianchang512

AI Summary

Update adds new ASR models and voice support including Fun-ASR-Nano-2512, Huggingface ASR channels, and CosyVoice3 voice cloning. Includes bug fixes for voice channel selection and batch processing.

Key Highlights

  • Added Fun-ASR-Nano-2512 speech recognition model support
  • Added Huggingface_ASR channel supporting moonshine and nvidia models
  • Added CosyVoice3 voice support
  • Fixed voice channel selection error
  • Fixed batch speech-to-subtitle 2nd+ file error
  • Optimized model download error messages

New Features

  • Fun-ASR-Nano-2512 speech recognition
  • Huggingface ASR channel (moonshine 8 models + nvidia English model)
  • CosyVoice3 voice cloning

Full Release Notes

<details>
<summary>v3.91更新日志</summary>
<div>
1. 修复 配音渠道 选择错误<br>
2. 修复 批量语音转字幕 第2个以后报错问题<br>
3. 增加 Fun-ASR-Nano-2512 语音识别模型支持<br>
4. 增加 Huggingface_ASR 语音识别渠道，支持 moonshine 的8个模型和nvidia的一个英文模型<br>
5. 增加 CosyVoice3 配音支持<br>
6. 优化 模型下载错误提示
</div>
</details>



##  v3.91 完整包下载(3.9G 含large-v3-turbo模型)

> 如果未安装过旧版本，请在此下载完整版，如需cuda加速，需有英伟达显卡并且安装cuda12.8及cudnn9


百度网盘下载地址:  https://pan.baidu.com/s/11Ho7BMFTs3vO2fGkiCpPkQ?pwd=1234

huggingface.co: https://huggingface.co/mortimerme/repocollect/resolve/main/win-pyvideotrans-v3.91.7z?download=true

 

## v3.91 补丁包(320MB)

> 不可单独使用，仅用于覆盖旧包，与完整包区别是不含有`torch`模块
> 因torch从 2.5 升级到 2.7，补丁包不含有torch，直接下载覆盖补丁包可能导致无法cuda加速，建议下载完整包覆盖
> **覆盖后如果闪退，请下载完整包**

百度网盘下载地址:   https://pan.baidu.com/s/18QxJZL9tcZjdbNzrrdqhuw?pwd=1234

huggingface.co: https://huggingface.co/mortimerme/repocollect/resolve/main/win-PatchUpdate-v3.91.7z?download=true