Open-source speech recognition, text-to-speech and voice conversion that you can run locally for free.
Robust open-source speech recognition models.
Faster Whisper transcription with CTranslate2.
A C/C++ port of Whisper that runs on CPUs and small devices.
Speech recognition with word-level timestamps and speaker diarization.
The code for the small, open-weight Kokoro-82M TTS model.
A fast, local neural text-to-speech system.
Fluent TTS based on flow matching.
An open-source text-to-speech system.
A generative speech model for everyday dialogue.
A deep learning toolkit for text-to-speech.
Voice conversion that can train on 10 minutes of audio or less. Only use voices you have the right to use.
Every link goes to the project's own GitHub repo or official site. Check each project's license and docs before you use it.