# Remove Background Music from Video but Keep the Voice

Choose the right way to remove music while keeping dialogue. Compare separate-track editing, speech separation, and noise cleanup.

Canonical: https://soundclean.app/blog/remove-background-music-from-video-keep-voice
Author: Carl Brownson
Published: 2026-10-18
Updated: 2026-10-18

To remove background music from a video and keep the voice, first check whether you still have the editing project. If music is on its own track, mute or lower that track and export again. If you only have the finished video, speech and music may already be mixed together. That needs a separation test, not a full-track mute.

A noise cleaner and a music separator solve different problems. [SoundClean](https://soundclean.app) offers a speech-cleanup preview for noisy recordings. Do not assume that it will deliver separate music and dialogue tracks.

## Quick Answer

Use the original editing project when possible and mute only the music track. For a finished video with music and speech mixed together, test a tool that explicitly separates them. Listen through the words spoken under the loudest music before accepting the result.

## Choose the route that matches your file

| Source available | First action | What to check |
| --- | --- | --- |
| Editing project with a music track | Mute or lower that track | Dialogue and useful effects remain |
| Original camera file before music was added | Return to that source | It contains the complete performance |
| Finished video with mixed speech and music | Test speech/music separation | Words survive during the loudest music |
| Speech with fan or traffic noise | Test speech noise cleanup | Voice remains natural |
| Video where no sound is needed | Mute all audio | Dialogue is intentionally removed |

Ask the editor for the original voice track before spending time on separation. A clean export from the project can be better than a repair of the final mix.

## When music is on a separate track

Duplicate the project before editing. Find the music layer by listening to tracks individually. Lower or mute only that layer, then play the full sequence.

Check transitions. An introduction may use one music clip while the middle section uses another. Check any nested sequences or grouped clips too. Muting one visible item does not prove that all music is gone.

Export a short section first. Confirm that the voice remains synchronized with the picture and that its first and last words are complete. If it passes, export the full edit.

## A separation workflow for one mixed video

Keep the source and make a test copy. Choose a passage where the speaker talks over music, including a quiet sentence. A music-only pause is an easy test that says little about speech preservation.

Use a tool that explicitly supports speech/music separation. Adobe's [music-removal page](https://podcast.adobe.com/en/remove-music-from-audio-or-video-with-adobe-podcast) describes separating speech, music, and ambience from a mixed file and exporting individual tracks. Check its current plan terms before using it.

This is a capability reference, not a claim that a specific recording will separate perfectly. Singing, instruments, and speech can overlap. Listen for faint melodies, missing consonants, and a watery voice.

If your goal is ordinary background-noise reduction after separation, return to a short speech sample and test cleanup separately. See [Remove Background Noise from Video](/blog/remove-background-noise-from-video-online).

## A practical acceptance test

Imagine a training video with a narrator over a piano track. Check three sections:

1. A quiet instruction while the piano is loud.
2. A pause between sentences.
3. A word ending as the music changes.

Keep the separated version only when the instruction remains complete. A clean pause does not compensate for a missing word. It can be better to leave a little music than to remove the last part of a sentence.

For an interview, verify quotations against the original. Use the [interview cleanup guide](/blog/clean-interview-audio) for a source-preserving workflow.

## Where SoundClean can fit

If what you hear is mainly room noise, fan sound, or another background disturbance around speech, try SoundClean from the original recording. Compare its preview before processing the full file.

For music-heavy material, test a short excerpt first and keep expectations limited. SoundClean should not be treated as a guaranteed music-removal service. If you need isolated tracks, use a tool that offers them.

If you compare several tools, give each the same original excerpt. Avoid repeatedly processing one tool's output through another just because music remains. Each pass can alter the voice.

## Put repaired audio back into the video

When a separator returns audio only, keep the original video in your editor. Add the repaired audio at the same starting time. Mute the old mixed track so both copies do not play together.

Check a visible consonant or hand clap near the start and near the end. Matching only the beginning can miss a timing error later. Keep the frame rate and sequence timing unchanged during this audio replacement.

For a screen demonstration, see [Clean Screen Recording Audio](/blog/clean-screen-recording-audio-online).

## FAQ

### Can I remove music without removing the voice?

If music and speech are separate tracks, mute the music track. For a mixed file, separation may help, but the result can contain music remnants or damaged speech.

### Does muting a video keep the dialogue?

No. Muting the complete audio track also removes dialogue on that track. You need separate tracks or a separation process to preserve speech.

### Is SoundClean a music stem separator?

SoundClean provides speech noise cleanup and a preview. Do not assume it will produce separate music and dialogue tracks or fully remove music from a mixed file.

## Test the right repair for your video

Return to the original project for separate music tracks. If the main problem is background noise around usable speech, upload a short source recording to [SoundClean](https://soundclean.app) and compare its preview. For a mixed music track, start with a separation tool and check the dialogue before export.