You want the instrumental of one specific song, to sing over. Not a MIDI approximation, not a cover version — the actual backing track from the recording you already have.
The short path
- Trim the song to the section you need if it runs over five minutes.
- Open the vocal remover and drop the file in.
- Wait for it to finish — it runs on your own processor, so the time depends on your machine, not on a queue.
- Download the instrumental stem. Take the vocal stem too: it is the fastest way to hear what was removed.
Nothing is uploaded and no account is involved, so the whole thing takes about as long as the processing does.
What to expect
A neural separation gives a usable backing track on most material. It is not the studio instrumental, and it helps to know in advance where the seams show:
- Reverb tails. The vocal's reverb was printed into the mix and does not belong cleanly to either stem. You may hear a faint ghost of the voice's ambience over the backing track.
- Backing vocals and ad-libs are voices too. The model removes them along with the lead, which is usually what you want for karaoke and occasionally not.
- Dense, loud masters leave more artefacts than sparse arrangements. A crowded modern pop master is the hard case; a band recorded live is the easy one.
Getting a better result
- Start from the best source you have. A high-bitrate file separates more cleanly than a low-bitrate one, where the encoder has already discarded detail the model would have used.
- Trim to the part you will actually sing. Five minutes covers most whole songs, but cutting down to the verse and chorus you are practising is still faster and asks less of your device.
- Listen to the vocal stem. If the vocal came out clean, the instrumental is clean too — they are two halves of the same decision.
- Export as WAV if you are taking it into another editor, and MP3 only for the final file.
Why not Audacity's built-in effect
Audacity can remove a centred vocal by phase cancellation, and it is free and offline. But it removes everything in the centre of the stereo field — which is the kick, the bass and the snare as well as the voice — so the backing track comes back hollow. We wrote up how to do it and when it fails, including a 30-second test for whether your file will work at all.
Frequently asked questions
How do I make a karaoke version of a song?
Separate the recording into vocals and instrumental, and keep the instrumental. A browser-based separator does this in one step: open the vocal remover, drop in the file, download the instrumental stem. No account and no upload is required.
Can I use a karaoke track I made this way in a performance?
That depends on the rights in the original recording, which separating it does not change. Private practice is one thing; a public performance, stream or release of a derivative of a copyrighted master is another. See Terms, section 2.
Why does my instrumental still have a trace of the singer?
Usually reverb. The dry voice separates cleanly but its reverb tail was baked into the mix and spreads across the stereo field, so a faint ambience of the vocal can remain even when the voice itself is gone.
What if the song is longer than five minutes?
Trim it first. The five-minute cap exists because the work runs on your own device, not as a paywall — there is no paid tier to lift it.