1 paper · 1 filter
Aviad Dahan, Moran Yanuka, Noa Kraicer +2
Existing video personalization methods preserve visual likeness but treat video and audio separately. Without access to the visual scene, audio models cannot synchronize sounds wit…