How to Remove Vocals in Audacity

How to Remove Vocals in Audacity (Two Methods That Actually Work)

Updated June 2026

Audacity can remove vocals free with its built-in Vocal Reduction and Isolation effect, which cancels audio panned to the center of a stereo mix. It works well when the lead vocal is centered and dry; it fails on reverb tails, stereo-spread vocals, and mono files. Here is exactly how to do both methods — and what to use when they are not enough.

Method 1: the Vocal Reduction and Isolation effect

Audacity ships with a purpose-built effect for this, and it is the first thing to try (steps below reflect Audacity 3.x as of June 2026 — menu locations occasionally move between versions):

Open your song (File → Open). It must be a stereo track — the effect compares the left and right channels, so a mono file gives it nothing to work with.
Select the whole track (Ctrl/Cmd+A).
Go to Effect → Special → Vocal Reduction and Isolation.
Set Action to Remove Vocals and apply. If too much music disappears with the voice, undo and try again with the Strength value lowered.
Use the Low Cut and High Cut frequency settings to protect the rest of the mix: raising the low cut (say, to 120 Hz) keeps the bass and kick from being thinned out, since the effect then leaves everything below that frequency alone.

The same dialog can do the reverse: set Action to Isolate Vocals to attempt an acapella. Quality varies a lot by song — isolation is a harder problem than reduction.

Method 2: the split-stereo invert trick

This is the classic manual technique the built-in effect automates. Knowing it is worth it — it works in any audio editor, not just Audacity:

Click the track’s name (or the dropdown arrow next to it) and choose Split Stereo to Mono. You now have two mono tracks: left and right.
Select only the lower track, then run Effect → Special → Invert (or just search the effect list for “Invert”).
Play both tracks together. Everything that was identical in both channels — typically the lead vocal — cancels itself out.

The catch: the bass and kick drum are almost always centered too, so they vanish along with the voice. The common fix is to duplicate the original track, low-pass that copy at around 100–120 Hz so it contains only the low end, and mix it back in under your vocal-cancelled version.

Why this works — and exactly when it fails

Both methods are phase cancellation: a sound recorded identically in the left and right channels disappears when one channel is inverted and the two are summed. That is why the results are predictable:

Works: studio mixes with a single, dry, center-panned lead vocal.
Fails — reverb and delay tails. The vocal itself is centered, but its reverb is stereo and different in each channel, so a ghostly “singing in the next room” echo survives the cancellation.
Fails — doubled or wide vocals. Modern pop choruses pan vocal doubles left and right; those layers are not identical across channels and stay in the mix.
Fails — mono sources. One channel means nothing to subtract.
Collateral damage. Anything else in the center — bass, kick, snare, often the lead synth — gets removed or thinned along with the voice.

The AI fallback: source separation

Phase cancellation removes sounds by where they sit in the stereo field. AI source separation removes them by what they sound like: a model trained on a large catalog of mixed and isolated tracks recognizes the voice itself — including its reverb tail and its panned doubles — and rebuilds the instrumental without it. That is why it succeeds on exactly the material where the Audacity methods fail.

The trade-off is honest. Audacity is completely free, runs offline, and gives you full-length output; on a centered, dry vocal it can be all you need. Trackpeel’s AI route gives a free 30-second preview (and 30-second MP3 samples with a free account), but the full-length lossless WAV is paid, from a $5 Day Pass. A practical workflow is to combine them: run the separation on Trackpeel, then bring the WAV stems back into Audacity to trim, fade, and master — the editor is still the right tool for everything after the split.

Which method should you use?

Centered, dry lead vocal, zero budget: Audacity’s Vocal Reduction and Isolation effect. Start here.

Learning the craft, or no internet: the split-stereo invert trick — it teaches you how stereo mixes work, and it travels to any editor.

Reverb-heavy, doubled, or wide-mixed vocals — or you need a result clean enough to perform with: AI separation. Preview it free first; if the phase trick already got you there, you have lost nothing.

One more option worth knowing about: community AI plugins for Audacity exist (Intel’s OpenVINO effects suite includes music separation, as of June 2026). Setup is more involved and hardware-dependent than either method above, but if you want AI separation entirely inside Audacity, it is out there.

Frequently asked questions

Does Audacity have a built-in vocal remover?

Yes. Audacity 3.x includes Vocal Reduction and Isolation under Effect → Special. It reduces center-panned vocals via phase cancellation and is free. It needs a stereo file, and results depend heavily on how the song was mixed.

Why can I still hear vocals after removing them in Audacity?

Usually one of three reasons: the vocal has a stereo reverb or delay tail that phase cancellation cannot touch, the chorus uses doubled vocals panned left and right, or the file is effectively mono. Those leftovers are a limitation of the technique itself, not a settings problem — AI source separation is the approach that handles them.

Does removing vocals in Audacity ruin the bass?

It can. Bass and kick are usually panned center like the vocal, so naive phase cancellation removes them too. Use the Low Cut setting in the Vocal Reduction and Isolation effect (around 100–120 Hz) to protect the low end, or mix a low-passed copy of the original back in.

Can Audacity isolate vocals instead of removing them?

Yes — the same Vocal Reduction and Isolation effect has an Isolate Vocals action. Expect rougher results than removal: isolation is the harder direction for phase-based tricks. For a cleaner acapella, an AI separator like Trackpeel extracts the vocal as its own stem.

Is Audacity good enough for a karaoke track?

For some songs, genuinely yes — try it first, it costs nothing. If the result has a ghost vocal or a hollow, phasey sound, run the same song through an AI vocal remover and compare the free preview before paying for anything.