Chatterbox TTS

Chatterbox TTS is a state-of-the-art open-source expressive text-to-speech model developed and open-sourced by Resemble AI. Released under the Apache-2.0 license, Chatterbox delivers human-like voice synthesis with zero-shot voice cloning capabilities and built-in synthetic audio watermarking via Perth.

### Core Architecture & Capabilities - **Expressive Zero-Shot Voice Cloning**: Clone any voice timbre and cadence with as little as 5 to 10 seconds of reference audio. - **Deep Prosodic Nuance**: Accurately reproduces colloquial pacing, conversational pauses, and emotional emphasis across natural dialogue. - **Built-in Perth Watermarking**: Embeds imperceptible cryptographic watermarks directly into the audio waveform to identify AI-generated audio and prevent misuse. - **Local & Self-Hosted**: Run completely offline on local GPUs (PyTorch) or deploy as an internal microservice without external API costs. - **Permissive License**: Apache-2.0 license enables research, modification, and commercial application development.