Hugging Face releases Phonon TTS, an on-device speech synthesis model

On November 5, 2026, Hugging Face published a blog post introducing Phonon TTS, a high-quality text-to-speech model designed to run directly on-device. Hugging Face says Phonon TTS delivers leading results in benchmark testing, though the company does not name the specific benchmark involved. The post does not state the model's parameter count, nor does it describe where the model can be downloaded or how it can be deployed. As of now, the model's real-world performance remains to be verified, and no independent measurements have been reported. No pricing, licensing terms, or availability details beyond the blog announcement have been provided, so prospective users currently lack the information typically needed to evaluate or adopt the model. The development marks Hugging Face's continued expansion beyond its core model hub into dedicated inference products, though details on performance and accessibility have not been released.