How to Remove Vocals From a Song, Step by Step

Updated September 2026

A complete walkthrough, start to finish, for pulling a vocal-free (or vocal-only) track from a song using this tool — including the checks worth doing before and after, not just the click-through.

Step 1 — Get the best source file you can

Use the highest-quality, most unmodified stereo file of the song you can find — a standard mp3 or wav rip of the actual release, not a re-recording (like a phone recording of the song playing through speakers), a low-bitrate re-encode, or a version that's been through multiple lossy conversions. Each of those steps can subtly degrade the stereo image the separation math depends on, even before you've processed anything.

Step 2 — Upload it and check the file info

At the tool, upload your file. The info line that appears shows duration, channel count, and sample rate. The one to actually check: it needs to say stereo, not mono — if it says mono, separation can't proceed at all; see why mono files can't be processed for why, and what to do instead.

Step 3 — Separate, and listen critically to both outputs

Click "Separate vocals." Two players appear — the vocal isolate and the instrumental. Listen to both, not just the one you actually wanted, since checking both tells you how cleanly this particular song's mix was centered (see why some songs separate better than others): if the "wrong" output still has clear vocal or instrument bleed, that's a property of the source mix, not something a retry will fix.

Step 4 — Understand what you're hearing

Step 5 — Download the files you need

Each output downloads as a named .wav file (e.g. mysong-instrumental.wav) — see the output format page for exactly what's in that file and why it's WAV rather than MP3.

Step 6 — If the free preview isn't enough, know the current limit

The free tier processes the first 60 seconds of any track at full quality, with no limit on how many different songs you try — useful for judging a song's separability before committing further. Full-length processing is a paid tier planned for later; there's currently no way to process a different 60-second window of the same file (see the audio-joiner page for why chopping and rejoining isn't a real workaround for this today).

Troubleshooting checklist

SymptomLikely cause
"This file is mono" message, separation blocked entirelySource file has only one channel — try a genuinely stereo file of the same song
Both outputs sound almost identical to the original, barely separatedThe file's left and right channels may be nearly identical (a near-mono stereo file) — see mono vs. stereo
Weak separation on an otherwise normal-sounding stereo songThe vocal likely wasn't mixed tightly centered in that specific recording — a property of the mix, not fixable by retrying
Couldn't upload the file at allFormat not supported by your browser's decoder — try a standard mp3 or wav

FAQ

Do I need to sign up or install anything first?
No — no account, no download, no browser extension. Everything happens on this page.
Will this work the same on mobile as on desktop?
It should, since it relies on standard Web Audio API features supported by modern mobile browsers — the same upload-and-process flow applies either way.

→ Try it free