OpenVoice
myshell-ai/OpenVoiceReviewed by AI
Checked by RepoPicks Radar on
What is OpenVoice?
OpenVoice, from MIT researchers and the company MyShell, copies the tone of a voice from a short reference recording and then reads text in that voice, with control over emotion, accent, rhythm and pauses. Version 2 speaks English, Spanish, French, Chinese, Japanese and Korean, and the voice can speak a language the original recording was not in. It is a Python package with three demo notebooks, so you install Python, install the package and run the notebooks; everything runs on your own machine and the licence allows commercial use.
OpenVoice is free and open source (MIT) and can replace ElevenLabs, Respeecher, and Murf.
How easy is it to install OpenVoice?
C · install 1 or 2 tools · Windows
- Windows C
- macOS C
- Linux C
The same on all three systems: install Python 3.9 or newer, install the package from source and open `demo_part1.ipynb` to `demo_part3.ipynb`, since there is no installer, no release file and no graphics card requirement. The old pinned versions make the install the hard part rather than the machine you use.
OpenVoice on GitHub
- Stars
- 38,192on Nov 10, 2026
- Downloads
- —all releases
- Growth
- +58stars in 7 days
Stars over time
Facts
- Runs on
- Windows, macOS, Linux
- You need to install
- Python
- To build from source
- Python
- Disk space
- 100 MB–300 MB · estimate
- Accounts or keys
- OpenAI (only for the extra demo notebook; `openai` and `python-dotenv` are in requirements.txt)
- Cost
- Free · MIT licence for both V1 and V2, free for commercial and research use according to the README.
- License
- MIT (Permissive)
- Can replace
- ElevenLabs, Respeecher, Murf
- Last change on GitHub
- Apr 19, 2025
OpenVoice risk signals (5)
- AttentionDependencies with known vulnerabilities
Pinned dependencies with known vulnerabilities (OSV): gradio: GHSA-26jh-r8g2-6fpr, GHSA-279j-x4gx-hfrh, GHSA-34rf-p3r3-58x2
automatic check
- Attentionvoice cloning impersonation risk
The repository describes itself as "Instant voice cloning" with "Accurate Tone Color Cloning" from a reference recording, so anyone's voice can be copied from a sample and made to say anything, in a language they do not speak.
AI review
- Attentionunmaintained
The last commit is 2025-04-19, 568 days before this check, there are no releases and 317 issues are open.
AI review
- Infoold pinned dependencies
requirements.txt pins `numpy==1.22.0`, `gradio==3.48.0`, `librosa==0.9.1` and `faster-whisper==0.9.0`, versions from 2022-2023 that often refuse to install on a current Python.
AI review
- Infoaudio watermark dependency
requirements.txt includes `wavmark==0.0.3`, a watermarking library, so generated audio may carry a hidden mark.
AI review
Evidence is shown in its original language.
What Radar could not check about OpenVoice
- Whether the install still succeeds today with those pinned versions and a current Python
- Where the voice model files are downloaded from and how big they are
- How close the cloned voice really sounds, and in which languages
- Whether a watermark is written into every output and whether it can be read back
- Whether the known gradio problems found by the checks affect the demo notebooks you run locally