Audio Extract Skill | Agent Skills

Audio Extract

Extracts the audio track from a video file. This is a local operation using the bundled ffmpeg binary - no API keys or external services required.

Command

npx agent-media@latest audio extract --in <path> [options]

Inputs

| Option | Required | Description | |--------|----------|-------------| | --in | Yes | Input video file path or URL (supports mp4, webm, mkv, avi, mov) | | --format | No | Output audio format: mp3 (default) or wav | | --out | No | Output path, filename or directory (default: ./) |

Output

Returns a JSON object with the extracted audio file:

{
  "ok": true,
  "media_type": "audio",
  "action": "extract",
  "provider": "local",
  "output_path": "extracted_123_abc.mp3",
  "mime": "audio/mpeg",
  "bytes": 24779
}

Examples

Extract audio as MP3 (default):

npx agent-media@latest audio extract --in video.mp4

Extract audio as WAV:

npx agent-media@latest audio extract --in video.mp4 --format wav

Custom output directory:

npx agent-media@latest audio extract --in video.mp4 --out ./audio-files

Use Case: Video Transcription Workflow

Since transcription services work best with audio files (smaller uploads, faster processing), use this workflow:

# Step 1: Extract audio from video (local, instant)
npx agent-media@latest audio extract --in interview.mp4 --format mp3
# Output: extracted_xxx.mp3

# Step 2: Transcribe the audio (cloud API)
npx agent-media@latest audio transcribe --in extracted_xxx.mp3 --provider fal

Provider

This action uses the local provider with bundled ffmpeg (via ffmpeg-static). No API keys required.

Agent Skills: Audio Extract

Install this agent skill to your local