Perform Instant Video and Audio Content Audits
Find specific moments in long videos without watching them.
Use Gemini's multimodal capabilities to analyze video and audio files directly to extract timestamps, summaries, and key insights.
The Scenario
You missed a long recorded town hall meeting or a technical webinar and need to extract the key action items and specific mentions of your department.
Before & after
You would watch the video at 1.5x speed, taking manual notes and scrubbing back and forth. For a 60-minute video, this takes at least 45–60 minutes.
Upload the video file directly to Gemini 1.5 Pro and ask for a timestamped summary or a specific insight. The AI processes the visual and audio data in 3–5 minutes.
The Prompt
Analyze this video and provide a summary of the key arguments made by the speaker. Also, provide a timestamp for when they discuss [TOPIC]. [UPLOAD_VIDEO]
Because Gemini 1.5 is multimodal, it can 'watch' video and 'listen' to audio alongside text, treating them as part of the context window.
Source
Release notes | Gemini API | Google AI for Developers"we can create applications that humans like to interact with... processing the audio we are also looking at the video"