What is new
Underdog built a small game to show what Woof can do. The scene is a room with a retro tech theme, a typewriter and a CRT monitor. If you type anything on the typewriter with your keyboard, the game speaks it back, and a tuner can be rotated to generate different kinds of music.
The game intro warns about what it calls intrusive cloud providers in the town. Everything runs in the browser through Underdog's speech generation model. Underdog founder Sigil Wen told TechCrunch he started building the app.
Size, access and open questions
Woof is 32MB and has 31 million parameters. Underdog says it is open for everyone to download, and that it can run on most devices without any internet connection. The company has not published licensing terms, a list of supported devices or browsers, or audio quality benchmarks.
The demo follows Bark, a 167MB speech-to-text model Underdog launched weeks earlier, also designed for edge devices. Underdog has also released a 27-billion-parameter model that can run on laptops, and it has a waitlisted app it positions as a locally hosted alternative to Muse and Instinct. Wen said Underdog wants to train the best open, American-made models that let people have private AI.
What it means for creators and sellers
A small model that runs offline may be handy for quick voice or audio drafts when you would rather not send files to a cloud service. For now Woof is a demo inside a game, not a production tool. TechCrunch notes that such a small model likely won't replace bigger models.
It is worth opening the demo in a browser to hear the output and judge whether it fits your workflow. Before building anything on the download, check the licensing terms and how the demo handles your input, since Underdog has not detailed either.
How this was made: the text was written by AI from the source material, checked against the source and reviewed by an editor. The illustration was generated by AI for AISTETIKA and is not a frame from the story.
