Make a Karaoke Version of a Song

Why this matters

Karaoke needs the instrumental plus timed lyrics. Source separation gives you the instrumental cleanly; OCR/speech timing (or manual sync) handles the lyrics.

Model-based separation beats old center-channel tricks — the backing track keeps its bass and stereo image.

How it works

Submit the song with stems=[vocals, instrumental]; the API returns both stems, and you build the karaoke session around the instrumental.

Submit a job with an input URL (S3, GCS or HTTPS), poll the job URL, and download the rendered output. No GPU, no queues, no ffmpeg builds to babysit.

Code

import requests, time

API = "https://api.mlslabs.io/v1/source-separation/jobs"
headers = {"X-API-Key": "YOUR_API_KEY"}
payload = {"input": "s3://bucket/song.wav", "stems": ["vocals", "instrumental"]}

resp = requests.post(API, json=payload, headers=headers)
job = resp.json()
while job["status"] not in ("succeeded", "failed"):
    time.sleep(3)
    job = requests.get(job["url"], headers=headers).json()

print("Output:", job["output_url"])

Pro tips

• Use the acapella stem to auto-generate lyric timing if your tool supports it.
• Normalize the instrumental to -14 LUFS for club/bar playback.
• For party use, export MP3 320kbps — WAV is for the editor, MP3 for the room.

Try the karaoke flow in Postman.