Google launches Veo 3 at I/O, a video model that also generates matching sound and dialogue

Google DeepMind/Veo

At I/O Google launched Veo 3, a video generation model that for the first time generates matching sound along with the picture: background noise, birdsong, and dialogue between characters with accurate lip-sync. It came with Imagen 4 image model and Flow, a new AI filmmaking tool.

• First video generation model to create matching audio: ambient sounds, music, dialogue.
• Accurate lip-sync between generated video and dialogue.
• Launched with Imagen 4 (image model) and Flow (AI filmmaking tool).
• Available to Gemini Ultra subscribers in US; enterprise access via Vertex AI.