---
title: "Media Processing"
description: "Media Processing skill for Vellum — processes video, audio, and image files through a multi-phase AI pipeline."
canonical_url: "https://www.vellum.ai/docs/skills-reference/media-processing"
md_url: "https://www.vellum.ai/md/docs/skills-reference/media-processing"
related:
  - "/docs/skills-reference"
  - "/docs/skills-reference/acp"
  - "/docs/skills-reference/amazon"
  - "/docs/skills-reference/app-builder"
  - "/docs/skills-reference/browser"
  - "/docs/skills-reference/chatgpt-import"
  - "/docs/skills-reference/computer-use"
  - "/docs/skills-reference/contacts"
---

# Media Processing

## What it does

Processes video, audio, and image files through a multi-phase pipeline — ingest, analyze with AI (Gemini for vision, Claude for reasoning), and generate clips or summaries.

## Setup required

Requires Gemini API key for visual analysis.

## Permissions

- Gemini API key required for keyframe/video analysis
- File access permissions for media files

## Common prompts

| You say...                                      | What happens                     |
| ----------------------------------------------- | -------------------------------- |
| “Analyze this video and tell me what happens”   | Full video analysis pipeline     |
| “Extract the key moments from this recording”   | Keyframe extraction and analysis |
| “Find the part where they discuss pricing”      | Query-based video search         |
| “Generate a 30-second clip of the product demo” | Video clip extraction            |
| “Transcribe and analyze this podcast episode”   | Audio processing                 |

## Configuration

- Three-phase pipeline: preprocess (ingest, deduplicate), map (Gemini-powered visual analysis), reduce (Claude-powered reasoning)
- Supports keyframe extraction, dead time detection, and cost tracking
- Resumable if interrupted

## Tips & gotchas

- **Automatic chunking.** Large media files are handled automatically — video is split into keyframes or chunks.
- **Cost tracking.** Shows you how much API usage each analysis requires.
- **Resumable.** If processing is interrupted, it picks up where it left off.
- **Simple transcription?** For transcription without visual analysis, use the Transcribe skill instead.