Audio Speed Changer

How to use

1. Drop the song you want to speed up or slow down onto the Track box. Its waveform appears with a player for the original.
2. Drag Playback speed, or tap Slowed, Faster or Nightcore. The readout tracks it to two decimals in 0.05 steps, and the strip under the controls shows the new length as you go.
3. Choose an output format. M4A is the default; WAV and FLAC are lossless and come out several times larger, and the strip states the WAV size before you press.
4. Press Change speed. A bar fills while it resamples and encodes, and Cancel stops it if you picked the wrong number.
5. Switch between the Original and Sped takes to check it, then press Download. The copy keeps its old name with the speed added, so mix.mp3 at 1.5× saves as mix-1.5x.m4a.

About this tool

Two effects come out of the same control here. Push a track above 1x and it gets faster and higher, which is the nightcore sound. Pull it below 1x and it slows and drops in pitch, which is what a slowed edit means. Both happen because changing audio speed by resampling drags pitch along with it, the way a record played at the wrong rpm does. The slider runs 0.5× to 2× in steps of 0.05, and three presets sit beside it: Slowed 0.75×, Faster 1.25× and Nightcore 1.5×, each lit while the slider rests on its speed.

Away from edits, I use it to drag a fast passage down far enough to work out what's being played, and to get through a recorded lecture in less time. Length moves with speed, because resampling is the entire mechanism. A 4-minute track at 0.75× runs 5:20, and the same track at 1.5× comes in at 2:40; the page shows that figure as you move the slider. At 1.00× nothing would change, so the button waits until the speed moves. M4A is selected the first time and is the right answer for most things you'd post, and whichever format you pick is remembered.

A 52-minute lecture, listened to in 34:40

Load lecture.m4a, drag the slider to 1.50× or tap Nightcore, and keep M4A. The strip says 34:40 before you press anything, because the new length is just the old one divided by the speed; the bar's job is the resample and the encode, not the math.

The price is pitch: at 1.5× the lecturer squeaks. If that grates over a full hour, step down to 1.25× and the same file runs 41:36 with the voice still recognisably human. Switch to the sped take for thirty seconds before you download either one; the copy saves as lecture-1.5x.m4a.

Rough edges

• Pitch is welded to speed. There's no time-stretch mode that holds the original key, so a lecture at 1.5x sounds like a chipmunk rather than a fast person.
• The slider stops at 0.5x and 2x. Running the output back through gets you further, but resampling twice compounds the artifacts and it turns ugly fast.
• The resampling is linear interpolation between neighboring samples, which is cheap and quick. On bright material at 0.5x you can sometimes hear that as a fizz on cymbals and sibilance. Fine for a vibe edit. Not what you'd master with.
• The whole file gets one rate; there is no way to speed a single passage from here. Cut the section out with the Audio Cutter, change its speed alone, and join the pieces back afterwards.