Kokoro is a small open-weight text-to-speech model, with just 82 million parameters, that turns written text into natural-sounding speech quickly and cheaply.
In plain words
Kokoro is small for a voice model, yet its creators say it sounds about as good as larger models while running much faster and costing less. Its weights use the Apache licence, so anyone can use it in personal projects or commercial products. It builds on an existing speech design called StyleTTS 2 and can be installed as a Python package.
A quick example
A developer building an audiobook tool could install the kokoro Python package, pass in a chapter of text and get back an audio file read in a chosen voice, all on their own computer.
Where you'll meet it
- Voice output for apps, assistants and audiobooks.
- Low-cost speech generation on local hardware.
Related
Sources
- Docshexgrad/Kokoro-82M, hexgrad (Hugging Face) · read 29 Sept 2026
- Docskokoro, hexgrad (GitHub) · read 29 Sept 2026
Checked by WhatIsAI against the sources above · 29 Sept 2026. Spotted an error? Report it.