Phonon brings multilingual text-to-speech to local devices

Phonon, a compact text-to-speech model, is designed to generate speech directly on phones, laptops, single-board computers and in web browsers, without a server or network connection. Its developers have released two versions of about 65 MB and 75 MB. Together, they support English, French, German, Spanish and Portuguese, with more than 100 built-in voices. Custom voices can be created in Gradium Studio through voice cloning or voice design.

The Phonon 31m model is available under Apache 2.0, including for commercial products. Phonon 40m supports all five languages under a non-commercial license. The system is based on the Pocket TTS lineage and uses flow matching. It also handles phone numbers, dates and codes, according to the developers.

On Hugging Face’s Open TTS Leaderboard, Phonon ranked first for intelligibility in German and Spanish and placed among the top three in every language evaluated. The models led the CPU streaming benchmark for time-to-first audio. A dedicated inference framework, ptts, supports more than 10 CPU, GPU and NPU backends, while Transformers support is available from version 5.21. Phonon was downloaded more than 200,000 times during its first two weeks.