Overview
dsh-voice-studio
Voice Studio for DeepSeek Harness: a sidebar panel that manages GPT-SoVITS voices (training, switching, speaking) and RVC covers (voice conversion, separation, mixing) on Windows.
README / EN
Package documentation
Registry summary
Voice Studio for DeepSeek Harness: a sidebar panel that manages GPT-SoVITS voices (training, switching, speaking) and RVC covers (voice conversion, separation, mixing) on Windows.
dsh.pub verifies the pinned bundle contract, runtime facts, and distribution semantics. The complete README remains in the source repository.
Read the full README on GitHubLIMITATIONS
Known limitations
- 仅 Windows（依赖 PowerShell / WMI / explorer） - 翻唱需要 Aurabot 风格的 RVC 目录布局，或自行在设置里指到等价结构 - 6GB 显存：TTS 引擎与 RVC 不能同时常驻，翻唱/训练会自动暂停引擎 - 训练小模型（GPT-SoVITS s1/s2）在 6GB 上只能选 v2，v2ProPlus 会 OOM ---