Change Video Speed

Speed a clip up or slow it down, with the audio re-timed instead of pitch-shifted. Runs entirely in your browser.

Re-timing, not resampling

Speeding up video is easy: each frame carries a presentation timestamp saying when to show it, and scaling those timestamps down makes the whole thing play faster. Nothing about the picture changes.

Audio is where tools go wrong. The naive approach is to play the existing samples faster, which is exactly what happens when you spin a record too fast — the pitch rises with the tempo and everybody sounds like a cartoon. The right approach stretches the audio in time while holding the pitch, which is what ffmpeg’s atempo filter does. A double-speed interview still sounds like the same two people, just in a hurry.

The 0.5 to 2.0 limit, and why the range stops at 4x

atempo will not accept a factor outside 0.5 to 2.0 in one instance. Ask it for 4 and the filter graph fails to build, which surfaces to the user as a conversion error that says nothing about audio. The fix is to chain instances — 4x is two doublings — and this page does that automatically, telling you when it happens.

The range stops at 4x because each link in that chain is another pass over the audio, and the quality cost is real. Two links is a sensible ceiling. If you genuinely need more, running the tool twice is both honest about what it is doing and no slower.

When to just drop the sound

At high speeds the audio is often not worth keeping — a timelapse of a build, a long screen recording skimmed at 4x. Turning it off skips the filter chain entirely, which makes the encode faster and avoids stacking resamples on audio nobody will listen to.

Change Video Speed FAQ

Does the audio end up sounding like a chipmunk?

No. There are two ways to make audio play faster, and they sound completely different. Resampling it — playing the same samples at a higher rate — speeds it up and raises the pitch, which is the chipmunk effect. Re-timing it with ffmpeg's atempo filter stretches or compresses the timing while leaving the pitch where it was, so a sped-up voice still sounds like the same person talking quickly. This uses the second one.

Why is the range limited to 0.25x and 4x?

Because of a real constraint in the audio filter. atempo accepts a factor between 0.5 and 2.0 in a single instance and refuses anything outside it, so faster or slower speeds have to be built by chaining instances — 4x is atempo=2.0,atempo=2.0. Each pass is another resample of the audio, so the chain is kept to two links. Beyond that you would be trading real audio quality for a number, and you can always run the tool twice if you genuinely need 16x.

Is my video uploaded?

No. The whole re-encode happens in your browser via WebAssembly. Nothing is sent anywhere, which also means the speed of the conversion depends on your machine rather than on someone else's queue.

Why does this take longer than muting or trimming a video?

Because it genuinely has to re-encode. Changing speed means rewriting every frame's presentation timestamp, and a video's frames are compressed relative to one another, so the picture has to be decoded and encoded again. Trimming on a keyframe boundary or dropping an audio track can copy the existing data untouched; retiming cannot.

What format do I get back?

H.264 video in an MP4 container, whatever went in. Since the file is being re-encoded anyway, it comes out in the one combination that plays on every phone, browser and messaging app without anyone having to think about codecs.