Hume AI Releases 3B Multilingual Text-to-Speech Model

Hume AI has introduced Tada-3B-ML, a new 3-billion-parameter model for text-to-speech (TTS) synthesis. Released on Hugging Face, the model is designed to generate natural-sounding speech with a high degree of expressive control, a key challenge in creating human-like voice interfaces.

A central feature of Tada-3B-ML is its multilingual capability. The model supports a broad set of languages, enabling developers to create voice applications for a global audience. Supported languages include:

• English
• Spanish
• French
• German
• Mandarin Chinese
• Japanese
• Korean
• Hindi
• Portuguese
• Italian

This release contributes to the growing field of expressive and multilingual speech generation. By aiming to capture the subtle prosody and intonation of human speech, models like Tada-3B-ML allow for more nuanced and emotionally resonant applications in areas like voice assistants, audiobooks, and accessibility tools.

An Important Note on Licensing

While the model weights are publicly available, they are governed by a custom license from Hume AI, not a permissive open-source license like Apache 2.0 or MIT. The terms focus on research and non-commercial use and include specific restrictions. Potential users should review the license carefully before integrating Tada-3B-ML into their work.