YoVoice
An open-source, local, offline desktop AI audio studio integrating voice cloning, text-to-speech, and music generation. Built on TypeScript/Electron, it supports macOS and Windows, and all inference is performed locally.
Timeline (last 90 days)
YoVoice is an offline desktop AI studio for macOS and Windows that integrates voice cloning, text-to-speech and music generation in a TypeScript/Electron client, with all inference performed locally.
YoVoice's inference uses Apple Silicon MPS or Windows DirectML/CUDA acceleration, has no per-character billing, and audio data is not uploaded to third-party endpoints.
YoVoice offers fast few-shot voice cloning, capturing timbre, accent and speaking pace from just a small amount of reference audio.
YoVoice's features cover multitrack scoring, full song generation with synchronized lyrics, conversational TTS and ambient sound design, positioning it close to a lightweight DAW.
YoVoice's audio files, model weights and project workspaces are all stored on the local file system, and it is explicitly labeled zero-telemetry.
On first launch, YoVoice prompts users to download the required open-weight models, including TTS, cloning and music checkpoints, into a local cache directory.
YoVoice's project repository is at https://github.com/leemysw/yovoice.git, started via npm install and npm run dev.
The original article does not give YoVoice's specific model names, generation quality comparison data or latency metrics, so the actual gap between its voice naturalness and commercial APIs cannot be judged.
YoVoice addresses cost structure and data sovereignty rather than an absolute quality ceiling; it is worth attention for creators seeking zero marginal cost and unlimited iteration, while cloud services still have their place for commercial projects seeking out-of-the-box best audio quality.
All Facts (9)
YoVoice is an offline desktop AI studio for macOS and Windows that integrates voice cloning, text-to-speech and music generation in a TypeScript/Electron client, with all inference performed locally.
50%UnverifiedYoVoice's inference uses Apple Silicon MPS or Windows DirectML/CUDA acceleration, has no per-character billing, and audio data is not uploaded to third-party endpoints.
50%UnverifiedYoVoice offers fast few-shot voice cloning, capturing timbre, accent and speaking pace from just a small amount of reference audio.
50%UnverifiedYoVoice's features cover multitrack scoring, full song generation with synchronized lyrics, conversational TTS and ambient sound design, positioning it close to a lightweight DAW.
50%UnverifiedYoVoice's audio files, model weights and project workspaces are all stored on the local file system, and it is explicitly labeled zero-telemetry.
50%UnverifiedOn first launch, YoVoice prompts users to download the required open-weight models, including TTS, cloning and music checkpoints, into a local cache directory.
50%UnverifiedYoVoice's project repository is at https://github.com/leemysw/yovoice.git, started via npm install and npm run dev.
50%UnverifiedThe original article does not give YoVoice's specific model names, generation quality comparison data or latency metrics, so the actual gap between its voice naturalness and commercial APIs cannot be judged.
50%UnverifiedYoVoice addresses cost structure and data sovereignty rather than an absolute quality ceiling; it is worth attention for creators seeking zero marginal cost and unlimited iteration, while cloud services still have their place for commercial projects seeking out-of-the-box best audio quality.
50%