Gradium has announced the open-sourcing of its Phonon text-to-speech (TTS) model, which boasts 31 million parameters and is optimized for on-device implementation. The English version of Phonon, which is packaged in a compact 64.5 MB size, is capable of generating speech in approximately 7 milliseconds on a standard laptop CPU.
This release under the Apache 2.0 license aims to provide developers and researchers with a powerful tool for integrating TTS capabilities into their applications. As on-device models become increasingly important for privacy and performance, Gradium’s Phonon sets a new standard for accessibility and efficiency in TTS technology.
Compiled automatically by the Tech AI Newsdesk from public AI-news sources and summarised in our own words.
Frequently asked questions
What is the size of the Phonon TTS model?
The English version of Phonon is packaged in a compact 64.5 MB size.
How fast can the Phonon model generate speech?
Phonon is capable of generating speech in approximately 7 milliseconds on a standard laptop CPU.
Under what license is the Phonon model released?
The Phonon model is released under the Apache 2.0 license.
