Kokoro TTS

Kokoro TTS is the open-weight, Apache-licensed text-to-speech model: 82M parameters deliver high-quality, multilingual audio with superior speed and efficiency.

Kokoro TTS runs on the open-weight Kokoro-82M model: this is a high-fidelity, Apache-licensed solution for speech synthesis. The core technology, built on the StyleTTS 2 architecture, achieves production-level audio quality while utilizing a lightweight 82 million parameters. This compact design ensures ultra-fast, cost-efficient deployment across diverse environments, from large-scale commercial APIs to edge devices. The system supports multiple languages (English, French, Korean, Japanese, Mandarin) and offers a curated library of voice packs for versatile content creation (audiobooks, podcasts, training).

https://huggingface.co/hexgrad/Kokoro-82M