How to Remove Background Music from a Video (Free, No Editing Skills)
The short answer: extract the audio track from your video (one ffmpeg command or any free converter), run it through an AI voice extractor, and you'll have a clean voice-only track in under a minute. If you need it back in the video, mux it in with one more command. Full steps below.
Step 1: Get the audio out of the video
Any of these free options works:
- ffmpeg (Windows/Mac/Linux, the pro option):
ffmpeg -i video.mp4 -vn -c:a libmp3lame -b:a 320k audio.mp3 - VLC: Media → Convert/Save → pick your video → Convert → Audio-MP3.
- Online converters work too, but for private videos prefer a local tool — your file never leaves your machine.
Use the highest audio bitrate available (320 kbps): separation quality depends heavily on source quality.
Step 2: Strip the music with AI
Upload the extracted audio to the free voice extractor. The AI — Demucs, the open-source model that tops academic separation benchmarks — isolates every human voice and discards music and most background sound. A 10-minute audio track processes in well under a minute on our GPUs.
Download the result: a voice-only MP3.
Step 3 (optional): Put the clean audio back in the video
One ffmpeg command swaps the audio track without re-encoding the video:
ffmpeg -i video.mp4 -i vocals.mp3 -c:v copy -map 0:v:0 -map 1:a:0 output.mp4
The video streams are copied bit-for-bit, so this takes seconds and loses no quality.
What kinds of videos does this work on?
| Video type | Result quality |
|---|---|
| Talk/lecture with intro or bed music | Excellent — speech separates cleanly |
| Vlog with licensed background track | Excellent |
| Interview in a café with music playing | Good — music removed, some room noise kept |
| Concert/performance footage | The voice is kept; whether that's useful depends on your goal |
| Very loud music, quiet distant speech | Fair — intelligible but thinner voice |
Why not just use a "music removal" filter in an editor?
Video editors' built-in noise reduction is designed for hiss and hum — broadband, constant noise. Music is neither: it moves, it has melody, and it overlaps the voice's frequencies. Simple filters either barely touch the music or hollow out the speech with it. Neural source separation is the only approach that reliably distinguishes what a voice is from everything else, which is why every serious tool built since 2020 uses it.
Copyright note
Removing a copyrighted music track from your own footage for your own use is unproblematic in practice. Re-publishing someone else's video, even cleaned up, still needs their permission — see our rights guide for details.