StemVocal
Menu

How to remove vocals in Audacity

Two ways to take the vocals out of a song in Audacity, or keep only the vocals: Intel's AI Music Separation plugin and the older center-channel effect.

Which method to use

Audacity no longer comes with a vocal remover. Its manual explains that the Vocal Reduction and Isolation effect has not shipped with Audacity since version 3.5.0, because the OpenVINO AI plugins yield a better result. That leaves two routes:

For most songs, start with Music Separation. Use the older effect when the plugin will not install on your version of Audacity, or for a quick check of a stereo track.

Method 1: OpenVINO Music Separation

  1. Install the plugins. Audacity's AI plugins page links the downloads: installers for Windows, with the release notes and instructions on Intel's GitHub page, and a macOS download. That page also warns that the plugins are not yet compatible with the latest Audacity version, so pick the Audacity version the release names.
  2. Open your song with File > Open, then select the track: click in the track and press Ctrl+A (Cmd+A on a Mac), or drag over a section.
  3. Run the effect. Choose Effect > OpenVINO Music Separation. Under Separation Mode, choose 2-Stem for an instrumental and a vocal track, or 4-Stem for drums, bass, other instruments and vocals. OpenVINO Inference Device chooses whether your CPU, graphics chip or NPU does the work.
  4. Wait for the first run. Intel's documentation says the model is compiled for your device the first time, which usually takes 10 to 30 seconds; later runs start faster because the compiled model is cached.
  5. Use the new tracks. They appear under your song with -Vocals, -Instrumental, -Drums and so on added to the name. Mute the vocal track to hear the instrumental, or solo it to hear the acapella.
  6. Export with File > Export Audio. Only unmuted tracks are mixed into the file, so mute or delete what you do not want first.

Under Advanced, Shifts runs the separation several times with the audio shifted and combines the results. Intel notes that more shifts may sound better, and that processing time grows in proportion.

Method 2: Vocal Reduction and Isolation

This effect works on stereo songs where the lead vocal is mixed equally into both channels, which the manual calls center-panned. Download it as a Nyquist plugin from the Muse support site that the manual links, add it through Audacity's plugin manager, then:

  1. Select the stereo track and choose Effect > Special > Vocal Reduction and Isolation.
  2. Choose Analyze first and click Apply. It reports the average pan position and a correlation between the two channels. The manual calls a value around +50 ideal; near 0, the channels have little in common and removal will not work.
  3. Choose an Action:
    • Remove Vocals removes the center within the vocal frequency range and keeps the track in stereo.
    • Remove Vocals: to mono does the same by the invert-and-mix method, which is quicker but returns mono.
    • Isolate Vocals keeps only the center within that range, as a dual mono track.
  4. Adjust Low Cut and High Cut. The manual suggests 120 Hz for most lead vocals, about 170 Hz for female voices and about 230 Hz for children's voices, and warns that sounds like S and Z reach 5,000 to 8,000 Hz.
  5. Raise Strength above 1.0 to remove more of the area around the center, then preview and apply.

Anything panned to the center goes with the vocal, often the bass and kick drum, which is why the Low Cut matters. Stereo reverb and backing vocals spread wide stay behind.

Isolating the vocals instead

To keep only the singing, use Music Separation and delete every track except the one ending in -Vocals. The older effect's Isolate Vocals action also extracts the center, but it returns dual mono and brings along anything else panned there.

Doing it without Audacity

If the plugin will not install, or you would rather not edit tracks, StemVocal's vocal remover does the same job in a web browser. Choose the song, pick Instrumental for a karaoke track or Vocals for the acapella, and download WAV or MP3 files. The stem splitter gives drums, bass and other instruments separately. Two songs a day are free, with credits for more.

Once you have the instrumental, the key finder tells you the key to sing it in.

Sources

Checked September 30, 2026:

Frequently asked questions

Does Audacity have a vocal remover built in?

Not any more. Audacity's manual says Vocal Reduction and Isolation stopped shipping with Audacity from version 3.5.0, because the OpenVINO AI plugins give a better result. The older effect is still available as a separate Nyquist plugin download.

Why does Vocal Reduction and Isolation leave the vocals in?

It only removes sound that is common to the left and right channels, so it needs a stereo track with the lead vocal panned to the center. Reverb, doubled vocals spread across the stereo field and instruments panned to the center behave differently, and a mono file cannot be processed at all.

How do I isolate vocals in Audacity?

With Music Separation, the new track ending in -Vocals is the vocal on its own. With Vocal Reduction and Isolation, choose Isolate Vocals; the manual notes that the isolate options return a dual mono track.

Is the OpenVINO plugin free?

Yes. Intel publishes it for Windows on GitHub, Audacity's AI plugins page also lists a macOS download, and the models run on your own computer without an internet connection.

Can I remove vocals without installing anything?

Yes. StemVocal's vocal remover works in a web browser: choose a song, pick Instrumental or Vocals, and download the result. See the last section of this guide.