OmniVoice Studio
A fully locally executed open-source voice studio supporting voice cloning, AI dubbing, video dubbing, audiobook generation, and transcription. It integrates 14 TTS engines, supports the MCP Server protocol, and is released under the AGPL 3.0 license.
Timeline (last 90 days)
IboChat's /speak endpoint tries three voice paths in order: GPT-SoVITS (best voice cloning, takes about 5–18 seconds, supports only English, Chinese, Japanese, and Korean, and permanently occupies about 3.9GB of VRAM once loaded), OmniVoice Studio (supports any language, takes about 30–40 seconds, called via an OpenAI-compatible endpoint), and Edge TTS (Microsoft's free neural voices, no cloning capability, supports 24 languages, zero latency, never fails).
IboChat runs on a single GPU with 4GB of VRAM, with GPT-SoVITS and OmniVoice sharing the same GPU; the author solves the audio corruption caused by resource contention by starting GPT-SoVITS on demand, using a watchdog thread to kill the process after 120 seconds of idleness, and clearing OmniVoice's VRAM before each call.
IboChat's voice cloning flow from YouTube is already automated: yt-dlp fetches the clip, ffmpeg trims it, OmniVoice generates the voiceprint, and it is cached per character.
OmniVoice Studio can be deployed with a single command and supports Windows, macOS, Linux, and WSL environments.
OmniVoice Studio requires only a three-second audio recording to clone a voice, needs no additional training, and the voice timbre can be saved and reused.
OmniVoice Studio has 14 TTS engines mounted at its backend, automatically selecting the most suitable engine based on the task type.
The OmniVoice Studio official website indicates support for as many as 646 languages.
OmniVoice Studio can detect different compute backends including CUDA, MPS, ROCm, and CPU, and automatically determine the execution path.
OmniVoice Studio runs entirely locally, requires no account registration or API Key, and audio never leaves the local machine.
OmniVoice Studio can launch an MCP Server, allowing tools that support the MCP protocol, such as Claude and Cursor, to directly invoke its speech capabilities.
6 more timeline events
All Facts (14)
IboChat's /speak endpoint tries three voice paths in order: GPT-SoVITS (best voice cloning, takes about 5–18 seconds, supports only English, Chinese, Japanese, and Korean, and permanently occupies about 3.9GB of VRAM once loaded), OmniVoice Studio (supports any language, takes about 30–40 seconds, called via an OpenAI-compatible endpoint), and Edge TTS (Microsoft's free neural voices, no cloning capability, supports 24 languages, zero latency, never fails).
50%UnverifiedIboChat runs on a single GPU with 4GB of VRAM, with GPT-SoVITS and OmniVoice sharing the same GPU; the author solves the audio corruption caused by resource contention by starting GPT-SoVITS on demand, using a watchdog thread to kill the process after 120 seconds of idleness, and clearing OmniVoice's VRAM before each call.
50%UnverifiedIboChat's voice cloning flow from YouTube is already automated: yt-dlp fetches the clip, ffmpeg trims it, OmniVoice generates the voiceprint, and it is cached per character.
50%UnverifiedOmniVoice Studio requires only a three-second audio recording to clone a voice, needs no additional training, and the voice timbre can be saved and reused.
50%UnverifiedOmniVoice Studio has 14 TTS engines mounted at its backend, automatically selecting the most suitable engine based on the task type.
50%UnverifiedThe OmniVoice Studio official website indicates support for as many as 646 languages.
50%UnverifiedOmniVoice Studio can detect different compute backends including CUDA, MPS, ROCm, and CPU, and automatically determine the execution path.
50%UnverifiedOmniVoice Studio runs entirely locally, requires no account registration or API Key, and audio never leaves the local machine.
50%UnverifiedOmniVoice Studio can launch an MCP Server, allowing tools that support the MCP protocol, such as Claude and Cursor, to directly invoke its speech capabilities.
50%UnverifiedIn OmniVoice Studio, ROCm is only available on Linux as optional support for AMD graphics cards.
50%UnverifiedOmniVoice Studio is open-sourced under the AGPL 3.0 license and is completely free.
50%UnverifiedOmniVoice Studio's video dubbing pipeline can transcribe the original video to text, translate it into the target language, re-read it using a cloned voice, and export the result as an MP4 file.
50%UnverifiedOmniVoice Studio offers seven functional modules, including voice cloning, timbre design, video dubbing, story audiobooks, a timbre library, and transcription.
50%UnverifiedOmniVoice Studio can be deployed with a single command and supports Windows, macOS, Linux, and WSL environments.
50%