Kokoro 82M

By Hexgrad

Kokoro is a compact 82M parameter text to speech model with open weights. Despite its small size it targets audio quality on par with much larger systems while remaining significantly faster and cheaper to run. Packaged via the kokoro inference library, it supports multiple languages and easy deployment from personal projects to production services.

New Audio Gen 4

Released: December 25, 2024

Overview

Open-weight 82M-parameter TTS model that delivers high quality speech at low cost, designed for fast, production-ready deployment across several languages.

🔊Text to speech