How to remove audio from video

Sometimes the audio has to go. A product demo where your cat meows at a critical moment. A screencast where you forgot your mic was off. A clip you want to autoplay on social media where unwanted sound would spook viewers. There are two ways to kill the audio — muting (keeping the track but silencing it) and stripping (deleting the track entirely) — and the right choice depends on whether you plan to add new audio later, whether you care about file size, and what platform you’re publishing to. This guide covers the mute-vs-strip distinction, platform autoplay policies that make silent video the norm, the file-size savings from stripping, replacement audio workflows, and captions as a frequently better alternative.

Mute vs strip

Muting replaces the audio with silence but keeps the audio track intact. Players still show an audio indicator (a muted speaker icon), and the file size is barely different from the original — you’ve swapped one compressed audio stream for another compressed audio stream of silence.

Stripping removes the audio track from the container entirely. Players show no audio indicator; downloaders save a video-only file. File size drops by whatever the audio track contributed — typically 5–15% for H.264+AAC videos, sometimes 20–30% for high-quality audio tracks.

# Strip audio entirely (removes the track)
ffmpeg -i input.mp4 -c:v copy -an output.mp4

# Mute audio (keeps track, replaces with silence)
ffmpeg -i input.mp4 -c:v copy -f lavfi -i anullsrc \
  -shortest -c:a aac -map 0:v -map 1:a output.mp4

When to mute

Keep the track and silence it when you plan to replace the audio later (post narration, background music, sound design). Muting preserves the audio channel layout and sample rate, so your editor sees the expected stereo/48kHz track and you’re not reconfiguring every time.

Also mute when your publishing platform expects an audio track. Some older workflows (broadcast, certain uploaders, some ad networks) reject video-only files. Silent track satisfies the check.

When to strip

Strip when the final deliverable is genuinely silent — a looping UI demo on a landing page, a background hero video, a silent GIF-style product card. You get the file-size savings and there’s no chance of a player glitching and briefly unmuting the audio.

File size savings

Typical audio track sizes (at common bitrates):
AAC 128 kbps stereo   = 960 KB per minute
AAC 192 kbps stereo   = 1.44 MB per minute
AAC 256 kbps stereo   = 1.92 MB per minute
Opus 96 kbps stereo   = 720 KB per minute
Uncompressed 48kHz 16-bit stereo = 11.5 MB per minute

For a 5-minute video at 2Mbps video + 192kbps audio, total is ~80MB; stripping audio drops it to ~73MB (~9% savings). For a 30-second autoplay loop at 4Mbps, total is ~16MB; stripping audio drops it to ~15MB (~6%). Not huge, but free.

Autoplay policies and silent video

Every major browser and platform has tightened autoplay over the past few years. Chrome requires videos to be muted to autoplay with sound. iOS Safari requires the muted attribute plus playsinline. Instagram and TikTok autoplay muted with sound toggled on tap. The result: any video that appears on a page without user interaction must be silent.

<video autoplay muted playsinline loop>
  <source src="/hero.mp4" type="video/mp4">
</video>

For videos that exist solely to autoplay (hero sections, product loops, decorative background), strip the audio at the source. The muted attribute works, but stripping guarantees no edge case where a browser update or user setting unmutes the video unexpectedly.

Replacing with new audio

A common workflow: strip the original audio, then attach narration or background music.

# Strip original audio, add new audio track
ffmpeg -i video.mp4 -i narration.mp3 \
  -c:v copy -c:a aac -b:a 192k \
  -map 0:v:0 -map 1:a:0 -shortest output.mp4

# The -shortest flag cuts the output to the shorter of the two streams
# If narration is longer than video, drop -shortest and pad video instead

Match the new audio’s length to the video explicitly. Misaligned audio beyond the video causes player glitches on some platforms (Twitter has been known to freeze on audio-longer-than-video files).

Captions as an alternative

Before stripping audio, ask whether captions solve the problem. If the original video has narration that conveys information, strip the audio and add captions or burn-in text — you keep the content accessible, meet autoplay silent policies, and accommodate viewers in quiet environments.

Adding captions is almost always a better user experience than silent video. Instagram data shows 85% of feed video is watched on mute by default; captioned video retains viewers, uncaptioned video loses them.

Audio track selection in multi-track files

Some MKV and MP4 files have multiple audio tracks (dubbing, commentary). Stripping all audio removes them all. To drop specific tracks:

# Keep only the first audio track, drop the rest
ffmpeg -i input.mkv -c copy -map 0:v -map 0:a:0 output.mkv

# Drop a specific track (track index 2), keep others
ffmpeg -i input.mkv -c copy -map 0 -map -0:a:2 output.mkv

Preserving metadata when stripping

Video metadata (timestamps, GPS, camera make) lives on the container and the video stream. Stripping the audio track with -an preserves video and container metadata but drops anything tied to the audio track (audio language tag, audio codec info). For most uses that’s fine. For archival, document what you stripped in a separate metadata field.

Lossless vs re-encoded stripping

Both -an and map-based stripping should be paired with -c:v copy to avoid re-encoding the video. If you omit -c:v copy, FFmpeg re-encodes with its default settings, which might downgrade quality or change bitrate. For a simple audio strip, there’s no reason to touch the video stream.

Normalizing volume vs stripping

If the original audio is just too loud or too quiet, don’t strip — normalize. Loudness normalization to -14 LUFS (the YouTube standard) or -16 LUFS (podcast standard) fixes the level without losing the content. Stripping throws away potentially useful information; normalize first, strip only as a last resort.

# Normalize to YouTube loudness (-14 LUFS)
ffmpeg -i input.mp4 -af loudnorm=I=-14:TP=-1.5:LRA=11 \
  -c:v copy output.mp4

Common mistakes

Re-encoding video unnecessarily. Stripping audio should never touch the video stream. Always use -c:v copy.

Confusing muted with stripped in the browser. A muted video still has the audio track; users can unmute it. A stripped video cannot be unmuted because there’s nothing there.

Forgetting platform autoplay rules. Any video that needs to autoplay on the web must be silent — either muted or audio-stripped.

Stripping audio when captions would serve better. Narration removed without text replacement loses the content entirely.

Leaving multiple audio tracks when replacing. If the file has original dialogue plus a dub, stripping with -an removes both; mapping preserves what you want.

Mismatched audio length when replacing. New audio longer than video causes glitches on some platforms. Use -shortest or trim audio first.

Skipping the shortest flag for short videos. For 6-second loops, a 30-second audio track plays beyond the visible video; viewers hear silence inexplicably.

Run the numbers

Remove or mute audio from a video file without installing editing software using the video mute tool. Pair with the video trimmer to cut the clip before stripping, and the audio trimmer when you plan to replace the stripped track with narration or music that needs its own pre-cut.