Gradium Launches Phonon: A Sub-80MB On-Device TTS Model Ranking Top-3 on Hugging Face's Open TTS Leaderboard

Gradium has released Phonon, a compact text-to-speech model available in two sizes (65 MB and 75 MB) that runs entirely on-device — no server or network connection required. Supporting English, French, German, Spanish, and Portuguese across more than 100 built-in voices, Phonon also enables custom voice creation through voice cloning and voice design in Gradium Studio. The smaller Phonon 31m (English-only) is released under Apache 2.0, making it freely usable in commercial products, while the larger Phonon 40m carries a CC BY-NC license for non-commercial use.

Built on the Pocket TTS lineage and using flow-matching audio generation, Phonon achieves quality rankings in the top 3 on Hugging Face's Open TTS Leaderboard across 23 models while remaining small enough to run on phones, laptops, single-board computers, and browsers. The model drew over 200,000 downloads in its first two weeks and has attracted early ecosystem integrations including a Hugging Face Transformers integration, an ExecuTorch browser conversion, and a Node.js binding called Micdrop.