# Gradium's Phonon TTS launch runs speech models on-device with open weights

> Original title: A Closer Look at Phonon TTS. Launching High-Quality On-Device Speech That Tops the Benchmarks.

- Source: Hugging Face Blog
- Published: 2026-11-05T00:00:00.000Z
- Bad Influence: https://badinfluence.news/items/t7eunn03rjza0jl5v9bsjnu51
- Original: https://huggingface.co/blog/phonon-tts

## Summary

Gradium says its Phonon text-to-speech models run on-device without a server or network connection. Phonon 31m for English is Apache 2.0 and usable commercially, while Phonon 40m covers five languages under a non-commercial license. The post reports Phonon ranks first in German and Spanish on intelligibility on the Hugging Face Open TTS Leaderboard and leads on time-to-first audio, with Transformers support from v5.21.

## Why it matters

The post details Phonon's on-device design, licensing split, and leaderboard results, giving developers concrete tradeoffs for choosing a local TTS model.
