What is omnivoice-studio?

OmniVoice Studio is an open-source desktop application for AI-powered voice work, it runs entirely on your own computer, requires no account, and needs no API keys or internet connection for its core features. It positions itself as a local, self-hosted alternative to paid cloud services like ElevenLabs. The main capabilities are: voice cloning (give it a 3-second audio clip and it can reproduce that voice speaking new text), voice design (adjust parameters like gender, age, accent, pitch, and emotion to create a custom voice), video dubbing (take a YouTube URL or video file, transcribe it, translate it, re-voice it in the target language, and export an MP4), and a dictation widget that activates with a keyboard shortcut from any application and automatically pastes transcribed text. It also handles vocal isolation (separating speech from background music), speaker identification (figuring out who said what in a recording), and batch processing of multiple files. The app supports 646 languages. It automatically detects whether you have a compatible GPU (supporting CUDA, MPS, and ROCm hardware acceleration) and falls back to CPU if needed. The app includes an invisible AI audio watermark feature for generated audio. It can also be controlled by AI assistants like Claude via an MCP server (a standard integration protocol). Installation is available as a desktop app for macOS, Windows, and Linux, or via Docker, or by running from source. The full README is longer than what was provided.