OpenVoice is an open-source instant voice cloning audio foundation model by MIT and MyShell in 2026, enabling accurate tone color cloning, flexible style control (emotion, accent, rhythm), and zero-shot cross-lingual capabilities. V2 supports native multilingual (English, Spanish, French, Chinese, Japanese, Korean) with improved quality under MIT license. Ideal for developers, researchers, content creators needing efficient, customizable TTS and voice cloning—free for commercial/research use.

Chatterbox is a family of state-of-the-art open-source TTS models by Resemble AI in 2026, featuring zero-shot voice cloning, expressive emotion control, paralinguistic tags, and multilingual support (23+ languages). With ultra-low latency (sub-200ms in Turbo), built-in PerTh watermarking for responsible AI, and MIT license, it outperforms many closed-source alternatives like ElevenLabs in blind tests—ideal for developers, voice agents, games, audiobooks, and creative applications.