yt-music-id

Identify music from any segment of a YouTube video. Point it at a URL and a time range — it'll pull the audio, slice the clip, and Shazam it.

Install

pip install yt-music-id

Prerequisites: ffmpeg must be installed and on your PATH.

Usage

CLI

# By start + end time
yt-music-id "https://youtube.com/watch?v=xyz" --start 1:23 --end 1:45

# By start + duration (seconds)
yt-music-id "https://youtube.com/watch?v=xyz" --start 0:30 --duration 15

# Just start time (defaults to 10s clip)
yt-music-id "https://youtube.com/watch?v=xyz" --start 2:00

# Auto-split long segments (recommended for anything >15s)
yt-music-id "https://youtube.com/watch?v=xyz" --start 7:00 --end 8:00 --multi

# Raw JSON output
yt-music-id "https://youtube.com/watch?v=xyz" --start 1:00 --end 1:20 --json

# Keep extracted audio files
yt-music-id "https://youtube.com/watch?v=xyz" --start 1:00 --keep-files

Python API

import asyncio
from yt_music_id import download_audio, extract_segment, identify_music, print_match

async def find_song():
    # Download audio
    path, title, duration = download_audio(
        "https://youtube.com/watch?v=xyz", "/tmp/raw"
    )

    # Extract a segment (7:00 to 7:15)
    extract_segment(path, "/tmp/clip.mp3", start=420, duration=15)

    # Identify
    result = await identify_music("/tmp/clip.mp3")

    # Print result
    if result.get("track"):
        print_match(result["track"], video_title=title)

asyncio.run(find_song())

Options

Tips

• 5-15 seconds is the sweet spot for Shazam recognition
• Use --multi for anything over 15 seconds — it auto-chops into overlapping 10s slices
• Works best when the music is prominent (less dialogue/SFX over it)
• Some montage videos use original/custom music that won't be in Shazam's database
• Requires ffmpeg on PATH — install via brew install ffmpeg, apt install ffmpeg, or ffmpeg.org

License

MIT