Finding an official instrumental for the song you actually want is the hard part. Popular tracks have them; anything regional, older, or slightly off the beaten path usually does not. Making your own takes a couple of minutes, and for a party or an event you will want to do a few more things beyond just removing the voice.
Quick Answer
- Get the song as an audio file — if it is in a video, pull the audio with video to MP3.
- Strip the vocals with the vocal remover.
- Trim the intro and outro with the audio cutter.
- Match levels across songs with the volume booster.
- Join the set into one file with the audio joiner.
Step 1: Get a Clean Source File
Start with the best audio you can find. Separation quality is limited by what you feed it — a low-bitrate rip has already lost high-frequency detail, and the model cannot restore it.
If your source is a video, extract the audio first using video to MP3 rather than screen-recording the sound.
Step 2: Remove the Vocals
Upload the file to the vocal remover. It runs Demucs, a deep learning separation model, and returns everything except the lead vocal as a WAV. A typical song takes about 30-60 seconds.
Two things worth knowing before you build a set around it:
Heavy reverb stays. If the original vocal was drenched in reverb, the tail tends to remain after the voice itself is gone — a faint shimmer in the gaps. It is usually inaudible over a PA in a noisy room, and very obvious on headphones.
Backing vocals are unpredictable. Stacked harmonies and vocal chops are often treated as instruments and kept. For a karaoke track that is frequently good — it leaves the singer something to lock onto — but do not assume the result is fully clean.
Listen to each track before the event rather than discovering it live.
Step 3: Trim to the Part You Need
Studio intros and long fade-outs are dead air at a karaoke night. Cut them with the audio cutter so the track starts a bar or two before the first vocal line.
If you are preparing several songs, trim them all to a consistent shape — a short lead-in, and a hard stop rather than a slow fade. It makes a set feel deliberate instead of accidental.
Step 4: Match the Levels
This is the step people skip, and it is the one the audience notices. Removing a stem removes its contribution to the overall level, so instrumentals come out quieter than the originals — and by different amounts, because a vocal-forward ballad loses more level than a dense rock mix.
Left alone, your set lurches between too quiet and too loud. Bring the quiet ones up with the volume booster so every track sits at roughly the same perceived loudness.
Step 5: Build the Set
For a continuous playlist, stitch the finished tracks together with the audio joiner. One file means no gaps, no fumbling between songs, and nothing to cue up mid-event.
If your playback system wants MP3 rather than WAV, convert at the end — not before — using the audio converter. Every lossy encode degrades the file slightly, so convert once, at the last step, after all the editing is done.
Practical Notes
Which songs work best. Sparse arrangements with one clear lead vocal separate most cleanly. Live recordings separate badly, because crowd noise and stage bleed are baked into the mix and there is no clean source to recover.
Keep the originals. If a track comes out poorly you will want to retry from the source rather than from a processed file. Each pass compounds artifacts.
Test on the actual system. Artifacts that are obvious on headphones often vanish over a PA in a room full of people — and occasionally the reverse. Check on the gear you will use.
Rights. Making an instrumental for private practice or a home party is one thing; playing it at a public or ticketed event is a licensing question, and separating a track yourself does not change who owns the recording.
If You Want the Vocals Too
Some setups use the original vocal as a guide track, faded low under the singer. For that you need the vocal as its own file, which the vocal remover does not produce — it returns the instrumental only.
Use the stem separator instead. It returns four files — vocals, drums, bass and other instruments — from the same underlying model, so you can rebuild the mix with the vocal at whatever level you want.
Related Tools
- Vocal Remover — the core step
- Stem Separator — four separate stems including isolated vocals
- Video to MP3 — pull audio out of a video source
- Audio Cutter — trim intros and outros
- Audio Joiner — combine a set into one file
- Audio Volume Booster — even out levels across tracks