Voice AI Startup Underdog Showcases 32 MB Model Woof

As reported by 深潮 TechFlow on November 6, citing TechCrunch, voice AI startup Underdog has launched a browser game based on its new voice model Woof, showcasing text-to-speech and music generation features.

The model is approximately 32 MB in size, contains 31 million parameters, and is available for download. In the game, users can input text to hear speech output, and generate different types of music by adjusting dials. Underdog said the model can run offline on most devices.

Source: 深潮 TechFlow