GitHub radar
VoxCPM: a web-based voice cloning tool (30 languages)
Create any voice cloning from only text or short audio file using VoxCPM2 created by OpenBMB. A browser based tool with no installation.
VoxCPM2 is a speech synthesis model developed by OpenBMB, the creators of MiniCPM model series. It’s a 30 language model with 48kHz sound quality and trained on more than 2 million hours of multi-lingual speech data at 2B parameter. You can play around with the web demo here on Hugging face. There are three different modes available. Design a new voice from given text. (No Audio required) Clone a controlled voice from your own audio sample. Advanced mode which includes complete style control feature. Just record a small clip and type any text you want to be read, it will come out in that voice. Or if you don’t have audio recording you can describe the voice by typing its age, accent, tone etc, it will generate for you. No installation required and no need of a high end GPU.
Why a vibe-coder should care
If you are making videos or podcasts then try VoxCPM2 for quick voice cloning without any installation. Just go to the web space and get started. As compared to other local TTS models, this doesn’t need a GPU or Linux. It’s multi-lingual and can also be used for creating voices from text.
How to install
Copy this and send it to your agent — Claude Code, Codex, any of them:
Open this OpenBMB voice synthesis demo: https://huggingface.co/spaces/openbmb/VoxCPM-Demo — explain what it can do and walk me through cloning a voice for text-to-speech narration.
Nothing to install — runs in the browser.
Open on Hugging Face▌ More finds