Chatterbox

Chatterbox is a text-to-speech model family for developers and creators who want to generate voices on their own hardware. It suits narration, voice agents and interactive media, with models for English and multilingual speech.

• Voice cloning: Generate speech using a short reference recording without training a separate voice model; voice conversion scripts are also included.
• Expressive delivery: The original English model lets you adjust emotional intensity, while capitalization in the input can affect emphasis.
• Multilingual voices: Clone voices across languages with Chatterbox Multilingual, or choose dedicated models for Chinese, Hindi, Latin American and Spain Spanish, and Brazilian and Portugal Portuguese.
• Fast speech models: Turbo focuses on fast English generation and accepts tags for non-speech sounds such as laughter and breaths; Nano supports CPU inference for on-device use.

The family is free and open source under the MIT license, including models that allow commercial use and modification of weights. You can self-host on GPUs in an air-gapped environment without an account or API key. Local generation keeps reference audio and speech processing on your hardware. Every generated audio file includes Resemble AI's inaudible PerTh watermark, designed to survive common audio processing.