On 2026‑10‑21 I discussed GradiumAI’s open‑source TTS suite Phonon on X. The suite ships with models all under 50 million parameters to enable on‑device speech synthesis while keeping naturalness across languages. The flagship model Phonon‑31m has exactly 31 million parameters and is released under the Apache 2.0 license. GradiumAI claims Phonon provides the broadest inference support to date, meaning it can run on a wide range of hardware and operating systems.
- Model: Phonon‑31m (31 M parameters)
- License: Apache 2.0
- Design goal: < 50 M parameters for on‑device deployment
- Inference claim: Broadest hardware/OS support among open‑source TTS suites
- Access: Announcement thread https://t.co/d3qa9qmnzY