If you searched for how to remove vocals in Audacity and the menus in your copy of Audacity do not match the tutorial you found, you are probably not doing anything wrong.
A lot of Audacity tutorials are simply old.
For years, the usual advice was to use Vocal Reduction and Isolation or some form of stereo phase cancellation. That is still useful background knowledge, but it is no longer the best default workflow for a current Audacity installation.
Today, the first method worth trying is OpenVINO Music Separation. Instead of cancelling whatever happens to sit in the center of the stereo image, it tries to separate the mix into musical sources such as vocals and instrumental.
This guide covers all three approaches:
- OpenVINO AI Music Separation;
- manual stereo inversion;
- the older Vocal Reduction and Isolation plugin.
You do not need screenshots to follow along. The menu paths are written out directly so the guide stays useful even if Audacity changes small parts of the interface later.
Which method should you use?
For most people, the answer is simple:
Try OpenVINO first.
Use stereo inversion only when you want a quick built-in method or you are working with a mix that has a very strongly centered lead vocal.
Use the legacy Vocal Reduction and Isolation plugin only if you specifically need the older workflow.
| What you want to do | Best starting point |
|---|---|
| Remove the singer and keep the backing track | OpenVINO 2-stem separation |
| Isolate the singer | OpenVINO 2-stem separation |
| Separate vocals, drums, bass, and other parts | OpenVINO 4-stem separation |
| Avoid installing an AI plugin | Stereo inversion |
| Follow an old Audacity tutorial | Legacy Vocal Reduction and Isolation plugin |
| Get a quick vocal/instrumental split without editing | Vocal Remover |
The important thing is that these methods are not three versions of the same trick. They work in fundamentally different ways.
Why older Audacity tutorials often look wrong now
This is where most of the confusion starts.
Vocal Reduction and Isolation is no longer bundled by default
Starting with Audacity 3.5.0, the old Vocal Reduction and Isolation effect stopped being included in the default installation.
So if a tutorial says:
Effect → Vocal Reduction and Isolation
and you cannot find that option, your Audacity installation is not necessarily broken.
The tutorial may simply predate Audacity 3.5.
The old effect still exists as a separately downloadable Nyquist plugin, but for a fresh installation it is no longer the obvious place to start.
OpenVINO support changed too
Early OpenVINO tutorials were often written with Windows in mind.
That changed when Audacity 3.7.4 added official macOS support for the OpenVINO AI plugins.
So there are two kinds of outdated Audacity tutorials floating around:
- tutorials that expect the old Vocal Reduction effect to be installed;
- tutorials that say OpenVINO is unavailable on Mac.
Before troubleshooting anything, check which version of Audacity the tutorial was written for.
Method 1: Remove vocals with OpenVINO Music Separation
If your goal is a usable instrumental or isolated vocal, this is the method I would try first.
OpenVINO is not just doing a smarter form of EQ. It is attempting source separation: given a finished stereo mix, it estimates which parts of the sound belong to vocals and which belong to the accompaniment.
That does not mean the result is identical to the original studio stems. It is still a reconstruction from the final mix.
But in practice, it handles many modern songs much better than old center-cancellation methods.
Step 1: Install the OpenVINO AI effects
Audacity and the OpenVINO AI effects are separate packages.
Install Audacity first, then install the OpenVINO package for your operating system.
After installation, close Audacity completely and reopen it.
This matters because users often install the plugin, return to the already-open Audacity window, and assume the installation failed when the new effects do not appear immediately.
Step 2: Import the song
Open Audacity and choose:
File → Import → Audio
You can also drag the audio file into the project.
If you have the original WAV or FLAC, use that instead of a repeatedly converted MP3. A better source will not guarantee perfect separation, but it gives the model more information to work with.
Step 3: Open Music Separation
Select the track or the section you want to process.
Then open:
Effect → OpenVINO AI Effects → OpenVINO Music Separation
If you do not see OpenVINO AI Effects, skip ahead to the troubleshooting section below. Do not start hunting for Vocal Reduction and Isolation unless you deliberately want the old plugin.
Step 4: Choose two stems or four stems
For ordinary vocal removal, choose the option that separates the recording into:
Vocals + Instrumental
That is the cleanest workflow when your goal is karaoke, singing practice, or an instrumental backing track.
If you are remixing or studying the arrangement, use the four-stem option instead. That gives you more control over elements such as vocals, drums, bass, and the remaining instruments.
For karaoke: two stems are usually enough.
For production or arrangement study: use four stems or the Stem Splitter.
Step 5: Choose a processing device
Depending on your system, OpenVINO may offer CPU, GPU, or NPU processing.
If you do not know which one to choose, start with CPU.
It may take longer, but it is usually the least confusing option.
AI separation is much heavier than a normal gain, EQ, or fade effect, so do not expect it to finish instantly on a long song.
Step 6: Process the track
Run the separation and let Audacity create the new outputs.
When it finishes, listen to both stems.
Do not only listen to the first 20 seconds.
The difficult parts are usually:
- loud choruses;
- stacked harmonies;
- long vocal reverb tails;
- sections with distorted guitars;
- parts where synths and vocals occupy a similar range;
- quiet intros with lots of ambience.
A track can sound clean in the verse and fall apart in the chorus.
That does not automatically mean the model is poor. It often means the original mix is difficult to untangle.
Step 7: Export what you actually need
If you want an instrumental, mute the vocal stem.
If you want the isolated singer, mute the instrumental.
Then use:
File → Export Audio
If you plan to keep editing the result, export WAV. If you only need a smaller file for listening or rehearsal, MP3 may be more convenient.
Why OpenVINO usually beats old center cancellation
To understand the difference, imagine a normal pop mix.
The lead vocal may be in the center, but so are the kick, snare, bass, and sometimes piano or lead instruments. The vocal may also have stereo reverb, doubled takes, harmonies, and widening effects.
A phase-cancellation method does not know which sound is the singer.
It only knows which information is similar between the left and right channels.
That is why the old method can remove some of the vocal and some of the bass at the same time.
OpenVINO is asking a different question:
Which parts of this sound are most likely to belong to the vocal source?
That is why it tends to behave better on modern productions.
It can still make mistakes. But it is solving the right problem.
Method 2: Remove centered vocals with stereo inversion
Stereo inversion is still useful because it is built into Audacity and takes almost no setup.
It is also a good way to understand why older vocal removers often sounded hollow.
Step 1: Start with a stereo recording
Import the song and confirm that it is stereo.
This technique relies on the relationship between the left and right channels. A mono file does not contain the stereo information the method needs.
Step 2: Split the stereo track to mono
Open the track menu and choose:
Split Stereo to Mono
You should now have two mono tracks.
One represents the original left channel and the other represents the right.
Step 3: Invert one channel
Select one of the two mono tracks and apply:
Invert
Only invert one channel.
If you invert both, you have not created the cancellation relationship you need.
Step 4: Play both tracks together
Now play the two mono tracks at the same time.
Information that was very similar in the left and right channels can cancel.
If the lead vocal was mixed tightly in the center, it may become much quieter.
Sometimes the result is surprisingly usable.
Sometimes it sounds terrible.
Both outcomes are normal.
Why stereo inversion leaves “ghost vocals”
This is the part that old tutorials often skip.
A vocal is rarely one simple centered signal.
A modern lead vocal can include:
- the dry lead;
- doubles;
- harmonies;
- stereo delay;
- long reverb;
- chorus or widening;
- distortion;
- chopped vocal effects.
The dry voice might be centered while the reverb spreads across the stereo image.
So the main vocal disappears, but its reverb remains.
That is where the strange “ghost singer” effect comes from.
You can also lose instruments you wanted to keep, because the method cancels centered information rather than vocals specifically.
If the backing track suddenly sounds thin or loses punch, the bass or drums may have been affected too.
Method 3: Use the legacy Vocal Reduction and Isolation plugin
There are still situations where the old plugin is useful.
Maybe you are following a course built around an older Audacity version. Maybe you want to reproduce an old project. Maybe you are comparing classic methods for learning purposes.
In those cases, you can still install Vocal Reduction and Isolation separately as a Nyquist plugin.
But for a new project, I would not go out of my way to install it before trying OpenVINO.
That would be choosing the older workflow first simply because an old tutorial ranked well in Google.
OpenVINO vs stereo inversion vs the old plugin
| Feature | OpenVINO | Stereo inversion | Vocal Reduction and Isolation |
|---|---|---|---|
| Tries to identify musical sources | Yes | No | No |
| Creates vocal and instrumental stems | Yes | No | Limited |
| Can separate more than two stems | Yes | No | No |
| Requires an extra install | Yes | No | Yes on current Audacity |
| Handles off-center vocals better | Usually | Poorly | Limited |
| Can work on mono material | Yes | No | Limited |
| Can accidentally remove centered instruments | Less directly | Yes | Yes |
| Best first choice today | Yes | Situational | Legacy use |
The difference is easiest to remember like this:
Stereo inversion removes location.
OpenVINO separates sources.
That one distinction explains most of the quality gap.
What happens with three common types of songs?
The mix matters almost as much as the tool.
Pop song with a centered lead and wide reverb
This is extremely common.
The lead vocal sounds centered, but the reverb and delay are wide.
With stereo inversion, the dry vocal may drop dramatically while the reverb remains obvious.
OpenVINO usually gives you a more useful result because it is trying to group vocal-related information together.
Best first choice: OpenVINO.
Older recording with a simple stereo image
Some older mixes have a very obvious center vocal and instruments placed clearly to the sides.
This is the kind of track where phase cancellation can still be fun to try.
You may get a cleaner result than expected.
Best first choice: OpenVINO, but stereo inversion is worth comparing.
Dense rock or metal mix
This is difficult for every separator.
Distorted guitars cover a lot of the same spectrum as the vocal. Cymbals spread across a huge frequency range. Vocals may be doubled and heavily compressed.
OpenVINO will usually give you more control than center cancellation, but artifacts are normal.
Best first choice: AI separation, followed by careful listening.
Do not expect the original studio stems
This is worth saying clearly because “AI stem separation” can sound more magical than it really is.
A separator does not reopen the original Pro Tools, Logic, or Ableton session.
It only sees the final mix.
If the producer originally had 50 tracks, compressed them, added effects, sent everything through buses, mastered the song, and exported one stereo file, the separator is working backward from that finished audio.
So a separated vocal may contain:
- faint cymbals;
- pieces of synth;
- reverb from the instrumental;
- watery artifacts.
And the instrumental may still contain:
- backing vocals;
- reverb tails;
- vocal distortion;
- breaths or consonants.
That is normal.
The useful question is not:
“Is it mathematically perfect?”
It is:
“Is it clean enough for what I need?”
For karaoke practice, a tiny amount of vocal bleed may not matter at all.
For a remix where the isolated vocal plays alone for eight bars, it matters a lot.
OpenVINO missing in Audacity? Check these first
Before reinstalling Audacity, check the simple things.
You installed Audacity but not the OpenVINO package
They are separate.
A fresh Audacity installation does not automatically mean the OpenVINO AI effects are ready to use.
Audacity was open during installation
Close it fully and reopen it.
This sounds obvious, but it solves a surprising number of plugin issues.
The tutorial you are following is old
If the tutorial tells you to use Vocal Reduction and Isolation, it may predate Audacity 3.5.
If it says Mac does not support OpenVINO, it may predate Audacity 3.7.4.
Your plugin build does not match your setup
Use the current OpenVINO package intended for your operating system and Audacity version.
Plugin compatibility changes faster than most blog posts are updated, so official Audacity documentation is the safest place to check when something does not match.
Audacity or an online vocal remover?
Audacity is the better choice when vocal removal is only one step in a larger editing job.
After separation, you can stay in the same project and:
- trim the song;
- cut sections;
- change levels;
- add effects;
- record new parts;
- rearrange the track;
- export a finished edit.
OpenVINO also runs locally, which is useful if you prefer keeping the processing on your computer.
But sometimes all you want is an instrumental.
In that case, opening an editor, installing a plugin, learning the menu, waiting for processing, and exporting the result may be more work than necessary.
A browser workflow is simpler:
Upload → Separate → Preview → Download
That is where Tunii Vocal Remover makes more sense.
If you want individual drums, bass, vocals, and other parts, use the Stem Splitter.
Which workflow makes sense for your goal?
Karaoke
Separate vocals and instrumental → listen to the chorus → export the instrumental.
If that is all you need, the Karaoke Maker is the shortest route.
Isolated vocal
Run a two-stem separation → keep the vocal → check harmonies and reverb-heavy sections.
You can also go directly to the Vocal Isolator.
Remixing or production
Use multi-stem separation → listen to each stem → keep only the parts you need.
For that workflow, the Stem Splitter is usually more useful than a simple vocal remover.
Learning Audacity
Try OpenVINO and stereo inversion on the same track.
The difference between the two results will teach you more about vocal removal than another page of theory.
A few things that genuinely improve the result
There is no hidden “perfect separation” setting, but a few choices help consistently.
Start with a clean source
If you have WAV or FLAC, use it.
If all you have is MP3, that is fine too. Just avoid repeatedly converting the same lossy file before separation.
Try OpenVINO before destructive tricks
Stereo inversion can remove useful centered instruments.
If OpenVINO is available, try that first and keep the manual method as a comparison.
Listen to the chorus
The easiest part of a song is not a good test.
Check the loudest, most layered section before deciding whether the result is good enough.
Judge the output by the job
A karaoke backing track does not need the same level of isolation as a professional remix stem.
Use the result where it will actually be used before deciding it failed.
Final thoughts
If Audacity vocal-removal tutorials have been confusing, the main reason is version drift.
Older guides often assume Vocal Reduction and Isolation is still part of the default installation. Others were written before OpenVINO became available on macOS.
For a current Audacity setup, start with OpenVINO Music Separation.
Use stereo inversion when you want a built-in experiment on a simple stereo mix.
Use the legacy plugin only when you specifically need the older method.
And if your real goal is not “learn Audacity” but simply “give me the instrumental,” use Tunii Vocal Remover and skip the editor setup altogether.
Editorial sources used for this guide
- Audacity Support — Isolating or removing vocals from a song — supports the current Audacity guidance to use Intel OpenVINO Music Separation for separating vocals from music.
- Audacity 3.7.4 release notes — confirms that OpenVINO effects became available on macOS in Audacity 3.7.4.
- Intel OpenVINO AI Plugins for Audacity — documents the OpenVINO plugin project used by Audacity for local AI effects.
- Audacity Manual — Vocal Reduction and Isolation — documents the legacy center-panned vocal reduction and isolation effect.
- Adobe Audition Help — Stereo Imagery effects — provides additional context on center-channel extraction and why centered vocals, bass, and lead instruments can be affected together.
Frequently Asked Questions
How do I remove vocals in Audacity?
For a current Audacity installation, the best starting point is OpenVINO Music Separation. Install the OpenVINO AI plugins, import your audio, open Effect > OpenVINO AI Effects > OpenVINO Music Separation, and choose the two-stem vocal and instrumental mode.
Why can't I find Vocal Reduction and Isolation in Audacity?
Audacity stopped shipping Vocal Reduction and Isolation by default starting with version 3.5.0. The legacy effect is still available separately as a Nyquist plugin, while Audacity recommends OpenVINO AI separation as the newer approach.
Does Audacity OpenVINO work on Mac?
Yes. Audacity added official macOS support for its OpenVINO AI plugins with Audacity 3.7.4 in June 2025. Install a compatible current Audacity version and the macOS OpenVINO package.
Can Audacity remove vocals without a plugin?
Yes. You can split a stereo track into two mono tracks and invert one channel. This can reduce center-panned audio, but it may also remove centered instruments and often leaves vocal reverb or backing vocals.
Can Audacity isolate vocals instead of removing them?
Yes. OpenVINO Music Separation can create separate vocal and instrumental stems, so you can keep the isolated vocal instead of the instrumental.
Why can I still hear vocals after removing them in Audacity?
Reverb, delay, backing vocals, stereo widening, and overlap between vocals and instruments can leave audible residue. Manual phase cancellation is especially limited because it removes center information rather than identifying the vocal source.
Tunii