LaunchAI ModelsNovember 6, 2026

Underdog demos 32MB speech model that generates speech and music

Read original source →techcrunch.com

The open model has 31 million parameters and runs in-browser without an internet connection, powering a retro-themed game called Woof. It follows Underdog's Bark, a 167MB speech-to-text model for edge devices.

People · Sigil Wen

1 source

More stories today

Open the live feed