OpenVoice
Instant voice cloning TTS by MyShell requiring only a short audio reference.
About
OpenVoice, developed by MyShell with researchers from MIT and Tsinghua University, clones a voice from a short audio clip and then generates speech in that voice with detailed control over style parameters including emotion, accent, rhythm, pauses, and intonation. Tone color is decoupled from those style controls, which also enables zero-shot cross-lingual cloning: the target language does not need to appear in the reference audio or the training data. Version 2, released in April 2024, improved audio quality and added native support for English, Spanish, French, Chinese, Japanese, and Korean. The model builds on established TTS and VITS techniques and has powered voice cloning on the myshell.ai platform since May 2023, handling tens of millions of requests within its first months. Both versions are open source under the MIT license, permitting free commercial use, which makes the project a common starting point for developers adding voice cloning to their applications.
Should you use OpenVoice?
Pick it when
Pick OpenVoice when you want MIT licensed voice cloning from a short clip with separate control over emotion, accent, rhythm, and pauses, and you need the cloned voice to speak languages missing from the reference audio.
Look elsewhere when
Skip it if you do not need cloning, since MeloTTS from the same team runs on CPU with no GPU, or if you need native German, Italian, or Russian, which Fun-CosyVoice 3.0 covers alongside zero-shot cloning.
Alternatives to OpenVoice
- MeloTTS
Same MyShell team and MIT license with real-time CPU inference, but fixed voices only, so pick it when you do not need cloning.
- Chatterbox TTS
Similar 4 GB footprint with emotion control, a low latency Turbo variant, and 23-plus languages, but without OpenVoice's separate control of rhythm, pauses, and intonation.
- CosyVoice
Apache-2.0 zero-shot and cross-lingual cloning in nine languages with streaming near 150 ms, broader coverage than OpenVoice's six native languages.
- F5-TTS
Flow matching zero-shot cloning with a fast real-time factor, but its pretrained checkpoints are CC-BY-NC where OpenVoice is MIT for commercial use.
Reviews (0)
Leave a Review
No reviews yet. Be the first to review!
Details
- Category
- Text-to-Speech (TTS)
- Price
- Free
- Platform
- Local/Desktop
- Difficulty
- Intermediate (3/5)
- License
- MIT
- Minimum VRAM
- 4 GB
- Added
- Apr 3, 2026