GitHub radar
A small model to clone voices on laptops!
Audio8 published a sub-1GB voice synthesis model. It copies any voice based on a brief reference sample, and runs locally on your PC.
The Audio8-TTS-Preview-0.1b is a zero-shot text to speech (TTS) model with voice cloning. Input some text and a brief reference audio sample, and the model will output speech in the desired voice. The primary model is ~170MB. Supported languages include English and Chinese. Experimental languages include German, Spanish, French, Italian, Japanese, Korean. This model is completely offline. There’s no cloud, no account, no API key required. It uses the standard Transformers library to load the model.
Why a vibe-coder should care
No more needing to pay for a TTS service or uploading audio to the cloud to get a new voice for videos or other content. This is the smallest voice cloning model out there. It runs locally on your existing PC, and only needs a few seconds of audio to reference.
How to install
Copy this and send it to your agent — Claude Code, Codex, any of them:
Set up the Audio8 TTS model from https://huggingface.co/Audio8/Audio8-TTS-Preview-0.1b — install it locally and show me how to synthesize speech using a short reference audio clip to clone a voice.
A regular laptop with 8 GB of RAM or more — the model is under 1 GB.
Open on Hugging Face▌ More finds