Phonon TTS brings on-device speech to phones and watches — and tops the speed charts

OFFICIAL Written by GPN · Nov 5, 2026, 10:03 AM ET · Phonon TTS · Tech

Most voice AI lives in the cloud, sending every sentence to a server and back. Phonon, a small text-to-speech model from Gradium, does the opposite: it runs entirely on your own device, with no server and no network needed — and what you type stays with you.

Phonon comes in two sizes, around 65 MB and 75 MB, and together they speak English, French, German, Spanish and Portuguese, with more than 100 built-in voices. You can also create your own voice in Gradium Studio, and the model handles the details other speech systems often trip over, like phone numbers, dates and codes. It runs on phones, laptops, single-board computers and in a browser tab. Phonon 31m speaks English and is released under Apache 2.0, so it can be used freely even in commercial products; Phonon 40m covers all five languages under a non-commercial licence.

The launch has drawn a crowd: Phonon was downloaded more than 200,000 times in its first two weeks. On Hugging Face's Open TTS Leaderboard, it ranks first for intelligibility in German and Spanish and in the top three in every language evaluated, next to models up to 100 times its size. Speed is where it stands out most — Phonon tops the CPU streaming benchmark for time to first audio, with the 31m model at 28 milliseconds and the 40m at 38 milliseconds. The launch demos show the range: Phonon talking an Apple Watch wearer through a coffee brew with Wi-Fi off, counting reps on an iPhone with no signal, and voicing a robot and a sci-fi game entirely on the player's own machine. It is also available in Transformers from version 5.21.

Facts gathered from

Hugging Face

Phonon TTS brings on-device speech to phones and watches — and tops the speed charts

Facts gathered from

More on this