Underdog has released an interactive game built with Woof, which the company describes as a 31.7 MB frontier text-to-speech model that runs in a browser. The game is set in a retro-tech room with a typewriter, CRT monitor and sound-producing desk lamp; its introduction takes a jab at cloud providers. Users can type text for the game to speak aloud and rotate a tuner to generate different kinds of music.
Underdog says Woof can run on most devices without an internet connection; the model has 31 million parameters and is available for anyone to download. A model this small is unlikely to replace larger models, the article says, but may be sufficient for local assistants or personalized podcasts. The release comes amid broader efforts by startups and researchers to create smaller models for edge devices, alongside the development of larger models that require substantial processing power.
Weeks earlier, Underdog launched Bark, a 167 MB speech-to-text model designed for edge-device transcription and dictation. Founder Sigil Wen told TechCrunch he began building an app because he wanted an AI notetaking system that runs on users’ devices and lets them own their data, rather than relying on services such as Granola. Wen said he believes most capable models available today could run on existing devices within months, and described Underdog’s aim as training open, capable, American-made models that support private AI. Underdog has also released an unnamed 27-billion-parameter model that runs on laptops and has a waitlisted app positioned as a locally hosted alternative to Muse and Instinct.