BREAKING
Google Flow demos Gemini Omni
How audio steering works
1Speak in video
2Model analyzes audio
3Edit applied
Any-to-any multimodal model
Inputs
Text and images
Audio and video
Outputs
Short video, ~10s
Native audio
Gemini Omni Flash preview
Where it's available
Platforms
Gemini app, Flow
YouTube Shorts, API
Transparency
SynthID watermark
C2PA metadata
Consistency still a challenge
AI NEWS BLITZ
Google Flow has shown off a new Gemini Omni feature that edits video using audio.