Infinity AI has launched a new foundation video model capable of generating realistic AI characters that speak, trained using a video diffusion transformer driven by audio input. This approach aims to overcome limitations of existing tools, which either only perform lip-syncing or lack speech capabilities. The model, which has undergone significant training investment, can animate diverse image types and handle singing, though it struggles with animals, cartoons, and can distort known identities. AI
IMPACT Enables creators to generate more realistic and expressive AI-driven video characters with speech capabilities.
RANK_REASON This is a new product release from a startup, not a frontier model release from a major AI lab.
Read on HN — machine learning stories →
- Andrew
- Elon Musk
- Frank Sinatra
- HeyGen
- infinity
- Infinity AI
- Luma
- Mona Lisa
- Runway ML
- Sidney
- Synthesia
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
