ποΈ Reverb-TTS is now on Pinokio: 1-click install
Clone a voice from 3β10 seconds of audio, entirely on your own machine.
Reverb-TTS (https://huggingface.co/software-mansion/Reverb-TTS) by Software Mansion is a small 297M-parameter
text-to-speech model. Give it a short clip and the exact words spoken in it, and it will say anything you type in that
voice.
What you get
- π£οΈ Voice cloning: upload a 3β10 s reference clip and its transcript.
- π§ Preset voices: pick one and start right away.
- π 7 languages: English, German, French, Spanish, Italian, Polish, Portuguese.
- β‘ Lightweight: about 300 MB of int8 weights by default, with int4, fp16 and fp32 options.
- ποΈ Controls for speed, sampler steps and seed.
- π API built in: call /generate from Python, JavaScript or curl.
How to use
- Click Install. Pinokio sets up the environment and the right PyTorch build for your GPU.
- Click Start. The weights download automatically the first time.
- Pick a preset or choose Clone a voice, type your text, and hit Generate.
β οΈ The model weights are CC BY-NC 4.0, so non-commercial use only. Only clone voices you have the right to use.
ποΈ Reverb-TTS is now on Pinokio: 1-click install
Clone a voice from 3β10 seconds of audio, entirely on your own machine.
Reverb-TTS (https://huggingface.co/software-mansion/Reverb-TTS) by Software Mansion is a small 297M-parameter
text-to-speech model. Give it a short clip and the exact words spoken in it, and it will say anything you type in that
voice.
What you get
How to use
β οΈ The model weights are CC BY-NC 4.0, so non-commercial use only. Only clone voices you have the right to use.