Resemble AI open-sourced Chatterbox Turbo, a fast text-to-speech model that runs locally with sub-150ms latency. It offers three variants: Turbo for English speed optimization, Multilingual supporting 23 languages with voice cloning, and an expressive English version. The MIT-licensed tool includes zero-shot voice cloning from 10 seconds of audio, watermarking, and expressive controls. In blind tests, it outperformed ElevenLabs in 63% of comparisons while being faster. Drawbacks include occasional overacting, tail-end audio artifacts, slow CPU performance requiring a GPU, and ethical concerns around voice cloning capabilities.