Audio Speed Changer
Speed it up or slow it down. Keep the pitch, or don't.
or paste one from your clipboard
How to use it
- Drop an audio file onto the page, or click to browse. The file is decoded by your browser and never leaves your device.
- Set the speed with the slider, or tap a preset. The new running time updates as you move it, so you can aim for a target length.
- Decide what happens to the pitch. Leave "Keep original pitch" off and the whole recording shifts up or down like a record played at the wrong speed; switch it on and the pitch stays put.
- Press play to hear the result. Resampled speed changes preview instantly; the pitch-preserving mode is rendered when you download, because it is far heavier to compute.
- Pick a format and click Download. The filename records the speed you chose, so 1.5x saves as song-1-5x.mp3.
Two different ways to change speed
A speed control looks like one setting, but underneath there are two unrelated operations and they sound nothing alike.
The first is resampling. The samples are read out at a different rate than they were recorded at, exactly like spinning a record at 45 rpm when it was cut for 33. Everything in the file moves together: play at 2x and every frequency doubles, which is a rise of exactly one octave. Play at 1.5x and the pitch climbs about seven semitones, a perfect fifth. Nothing is invented or thrown away, so the result is mathematically clean at any ratio. It simply sounds like a different recording.
The second is time-stretching. The audio is cut into overlapping windows of about 46 milliseconds and laid back down at a new spacing. Closer together to speed up, further apart to slow down. Because each window still contains the original waveform at its original frequency, the pitch never moves. The hard part is joining the windows: butt them together at the wrong phase and you get a rasp on every join. The fix is to search a small range around each target position for the offset where the waveform best continues what has already been written, which is why this mode takes real computation rather than a single pass.
Which mode you want
Leave pitch preservation off when the pitch change is the point: sped-up meme audio, slowed edits, deep or squeaky voice effects, sound design where you want a hit to sit lower. It is also the honest choice for anything short and percussive, because there are no stretching artefacts to worry about at all.
Turn it on for anything you intend to listen to for information: lectures, podcasts, audiobooks, interview recordings you are transcribing, or a backing track you are playing along to. Speech at 1.4x with the pitch held is comfortable for most people after a minute of adjustment. The same 1.4x resampled is comprehensible but tiring, because your brain is simultaneously correcting for a voice that has moved into the wrong register.
Choosing a number
For speech, 1.25x is nearly free. Most listeners stop noticing within a sentence. 1.5x needs attention but works for familiar material. Past 2x intelligibility falls off quickly, because consonants get shorter than the ear needs to tell them apart. Going the other way, 0.75x is the usual setting for transcribing an unclear recording, and 0.5x is about as slow as speech stays natural before it turns into a drawl.
For music, remember that halving the speed drops the pitch a full octave and doubling raises it one. If you want a smaller move than that with the timing untouched, the pitch shifter is the right tool; if you want the timing changed with the key untouched, use the tempo changer, which is the same stretching engine expressed as a percentage.