Kokoro-82M
A tiny 82M-parameter TTS model with quality above its weight.
⬦ Free and open source
Released under the Apache-2.0 license. There is no affiliate link on this page. The button goes straight to the project.
Free and open source, released under the Apache-2.0 license. This page carries no affiliate link. The button goes to the official project.
Get Kokoro-82M →Kokoro is the current open-TTS darling: an 82-million-parameter model small enough to run on modest hardware while sounding surprisingly natural. Crucially it is Apache-2.0, so unlike some open voice models you can use it commercially without a lawyer.
If you want local speech synthesis with no license landmines, start here.
The good
- Only ~350MB, runs almost anywhere
- Apache-2.0, the most permissive license here (commercial OK)
- Punches well above its parameter count
The catch
- Fewer voices than the big models
- Newer, smaller ecosystem
- Not a cloning tool
Pricing
Free and open source.
Alternatives to consider
- Piper Fast, lightweight neural text-to-speech that runs offline.
- OpenAI Whisper The open reference model for speech-to-text.