Extracts audio track from a video file. Use when you need to get audio from video, prepare audio for transcription, or separate audio from video content. Runs locally with no API key required.
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.
agent-media audio extract --in <path> [options]
| 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: ./) |
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
}
Extract audio as MP3 (default):
agent-media audio extract --in video.mp4
Extract audio as WAV:
agent-media audio extract --in video.mp4 --format wav
Custom output directory:
agent-media audio extract --in video.mp4 --out ./audio-files
Since transcription services work best with audio files (smaller uploads, faster processing), use this workflow:
# Step 1: Extract audio from video (local, instant)
agent-media audio extract --in interview.mp4 --format mp3
# Output: extracted_xxx.mp3
# Step 2: Transcribe the audio (cloud API)
agent-media audio transcribe --in extracted_xxx.mp3 --provider fal
This action uses the local provider with bundled ffmpeg (via ffmpeg-static). No API keys required.