Underdog shows off a 32MB speech model in new demo
Despite its small size, Underdog's new model can generate both speech and music.
TL;DR
- Underdog's new voice model, Woof, is only 32MB and runs in a browser demo.
- The demo features interactive elements like a typewriter and music tuner, with Woof generating speech and music locally.
- The model has 31 million parameters and can run on most devices without an internet connection.
- Underdog aims to enable private AI experiences and allow users to own their data.
- The company has also released a 167MB speech-to-text model (Bark) and a 27-billion-parameter model for laptops.