All open source
Open SourceVoice & Media5k

Kokoro TTS

Tiny 82M-param open TTS with strong quality and fast offline inference.

Kokoro is a lightweight open-weight text-to-speech model with a small parameter count (around 82M) released under the permissive Apache license. Its small size makes it fast and inexpensive to run while producing competitive-quality speech, and it has become a popular open TTS choice.

Repository

hexgrad/kokoro

Language

Jupyter Notebook

What you'd build with it

  • Generate speech cheaply and quickly in apps where a small, fast model is preferred
  • Run TTS locally or on modest hardware without heavy compute requirements
  • Add voice output to AI agents under a permissive commercial-friendly license

Tags

ttslightweightoffline