Background music removers compared: which ones keep the voice and the room

Most tools sold as a background music remover are built for one of two jobs: pulling the singer out of a song, or cleaning up speech. Removing the music from a video is a third job, and the difference shows up in what the tool throws away along with the song.

The short version: to fix a song YouTube already claimed, use Studio’s free Erase song. For a talking head, voice isolation in your editor (CapCut Pro, DaVinci Resolve Studio, Adobe Enhance Speech) is enough. To keep the room sound and get the whole video back, use a tool that separates dialogue, music and effects, such as MusicWipe, AudioPod or Adobe Podcast’s music remover.

I make MusicWipe, so I’m not neutral. I’ve tried to describe each option by what it’s designed to do and where it falls short, including mine.

What to look for

A video soundtrack has three parts: the voices, the music, and everything else (footsteps, applause, traffic, the hum of the room). A good result removes only the music. So the questions that matter are:

• Does it keep the sound effects and ambience, or only the voice? Voice only sounds like a dubbed film.
• Does it give you a video back, or an audio file you have to line up again in an editor?
• Can you hear the result before you pay or spend credits?
• Where does it run? Browser, phone, desktop, or inside an editor you already use.

To hear the difference, play the three versions of this test clip. “Voice only” keeps the speech and nothing else, the way a voice-first tool aims to. “Ambience 60%” puts some of the room back.

The options

Vocal removers and stem splitters

Tools like LALAL.AI, Moises and the many free “vocal remover” sites split a song into vocals and instrumental, and some go further into drums, bass and other instruments. That’s the job they are built for.

On a video, the speech lands in the vocals stem, so you keep the voice. The catch is everything else. These splitters are built for songs, so effects and room sound tend to go with the instrumental and leave with the song. Most hand back audio files, not your video, though LALAL.AI can return a video file, and it also has a voice cleaner for video that keeps the voice and removes the rest, room sound included. That can be enough for a talking head with music under it. It’s a poor fit for a vlog, a sports clip or anything where the scene’s sound matters, because those sounds aren’t in the vocals stem.

Voice isolation inside your editor

• CapCut has a voice isolator that lifts speech out of background noise and music. You can try it free, but exporting the result needs CapCut Pro.
• DaVinci Resolve Studio (the paid version) has Voice Isolation in the audio Inspector and on the Fairlight page, with an amount control. The free version doesn’t include it.
• Adobe Enhance Speech, in Premiere Pro and on Adobe Podcast, cleans up speech, with a mix control to blend the original back in. (Adobe also has a separate music remover, below.)

These are convenient if you already edit there, and you never leave the timeline. They are voice-first by design: they keep speech and treat the rest as noise to remove, so the room goes with the music. Turning isolation down brings the room back, and the music with it.

YouTube Studio’s Erase song

If YouTube has already claimed a song in an uploaded video, Studio offers Erase song, which removes only the claimed song and leaves dialogue and other audio. It’s free and needs no re-upload. It only works on claimed songs in videos already on YouTube, it may fail on a song that’s hard to separate, and since June 2025 saved edits can’t be reverted. More in what to do about a YouTube copyright claim .

Dialogue, music and effects separators

A newer group of tools splits a soundtrack the way film studios do: dialogue, music, and effects as three separate parts. AudioPod’s dialogue tool returns the three as separate files on its Creator plan and up, and Adobe Podcast’s music remover gives separate speech, music and ambience tracks on Adobe’s Firefly Pro plan, with a 7-day trial. Pro audio suites like iZotope RX can get there too, with more skill and time.

MusicWipe is in this group. It separates the three parts, removes the music, and gives you back the same video with a new soundtrack, without re-encoding the picture once you’ve bought minutes or a plan. A slider sets how much of the effects and room sound goes back in. Every video starts with a free 30-second preview, in the browser, on Android or on the Mac.

The weaknesses: it isn’t free beyond the preview and 3 free minutes once, whose exports carry a small watermark (plans start at $9.99 a month, or a one-time $2.99 for 10 minutes on the web), steady rhythmic sounds such as chopping can be mistaken for music, and in our own tests on 95 commercial songs mixed under dialogue, an open-source fingerprinting tool (not YouTube’s) still identified the song in 3 of the cleaned clips, against 87 of the originals. The details are in how we tested it .

Side by side

Which one to use

• The song is the problem only on YouTube, and it’s already claimed: try Erase song first. It’s free.
• You already pay for CapCut Pro or DaVinci Resolve Studio, and it’s mostly speech: use their voice isolation.
• You want the instrumental or a karaoke track: use a vocal remover. That’s the job it’s built for.
• The scene’s sound matters (vlogs, weddings, events, sports, street footage) or you want the finished video back without an editor: use a dialogue, music and effects separator.

Whatever you pick, listen to the loudest part of the song with headphones before you publish. That’s where traces survive.