Underdog releases 32MB open speech model that runs in-browser
The 31-million-parameter Woof model generates speech and music locally, with no internet required, and is available for download.
1 source · cross-referenced
- Underdog released Woof, a 32MB text-to-speech model that runs in-browser without an internet connection.
- The model has 31 million parameters and can generate both speech and music.
- Underdog also offers a 167MB speech-to-text model and a 27-billion-parameter model for laptops.
Underdog, an AI startup, released Woof, a 32MB text-to-speech model designed to run locally in a web browser without an internet connection. The model has 31 million parameters and can generate both speech and music.
The company demonstrated Woof in an interactive browser-based game, where users type text that is spoken back by the model and manipulate a music tuner to generate different sounds. All processing occurs on-device.
Underdog also provides a 167MB speech-to-text model called Bark, released weeks earlier, which is similarly designed for edge devices. Additionally, the company offers a 27-billion-parameter model that can run on laptops.
Sigil Wen, Underdog’s founder, stated the company aims to enable users to run capable AI models on their own devices, emphasizing data ownership and privacy. He described a goal of making most capable models runnable on existing consumer hardware within months.
Woof is available for download, and the company positions its models as alternatives to cloud-based services like those from major AI labs.
- Nov 11, 2026 · Hugging Face
Hugging Face proposes token pooling-aware training to improve compressed multi-vector retrieval
Trust79 - Nov 11, 2026 · TechCrunch — AI
Fractile raises $720M Series C to develop AI chips and compiler stack
Trust79 - Nov 11, 2026 · The Verge — AI
Google’s Tabletop AI experiment lets users build custom collaborative apps via waitlist
Trust79