Account

What new tool do you want?

Video & Audio

Audio Speed Changer (Keep Pitch)

Speed up or slow down audio from 0.5× to 2× — keep the original pitch (time-stretch) or change it like a tape — for podcasts, lessons and practice.

How to use

  1. 1Add audio and choose the speed.
  2. 2Keep 'Preserve pitch' on for natural voices, preview, then download.

How it's calculated

Pitch-preserving speed changes use WSOLA time-stretching (overlap-add of best-matching windows). Without it, audio is resampled, which changes pitch too.

Frequently asked questions

Will voices sound natural?

Between 0.75× and 1.5× speech sounds natural with pitch preservation on; extreme settings add a slight echo.

How do I slow down a song to learn it?

Choose 0.5×–0.8× with pitch preserved, so notes stay in tune.

Can I do the 'nightcore' effect?

Turn pitch preservation off and speed up to about 1.25×.

A worked example of the duration math

A 10-minute podcast episode sped up to 1.5× becomes exactly 6.7 minutes (10 ÷ 1.5), and the tool displays this conversion directly as a note before you download — useful for quickly checking whether a sped-up lecture will actually fit your commute. Slow the same file to 0.75× instead and it stretches to 13.3 minutes (10 ÷ 0.75) — the duration always scales by the inverse of the speed factor, which is worth knowing when you're trying to hit a specific target length rather than just a target speed.

Why 'preserve pitch' and 'don't preserve pitch' are fundamentally different operations, not just a toggle

With pitch preservation off, the audio is simply resampled — read back at a different rate, exactly like speeding up a tape or vinyl record, which raises the pitch when sped up and lowers it when slowed down (the classic chipmunk-voice or nightcore effect at 1.25×+). With pitch preservation on, the tool uses WSOLA time-stretching instead: it finds the best-matching overlapping windows of audio and reassembles them at the new tempo while keeping each window's original pitch content intact — a genuinely different algorithm, not a filter applied afterward, which is why it costs more processing time but produces a natural-sounding voice at 1.5× instead of a sped-up cartoon voice.

Why WSOLA still adds a slight echo at extreme settings

WSOLA works by overlapping and blending windows of audio, and that overlap-and-blend process is itself an approximation — it works very well within a moderate range (the FAQ specifically cites 0.75×-1.5× as the natural-sounding zone) but the artifacts of stitching together overlapping windows become more audible as the speed change grows more extreme in either direction, producing the slight watery or echo-like quality you can hear pushing toward 2× or down toward 0.5×. This is an inherent tradeoff of time-stretching algorithms generally, not a bug specific to this implementation — every pitch-preserving speed changer makes the same compromise at extreme ratios.

With pitch kept, 1.25× speed works in 40-millisecond pieces

Preserve pitch starts checked. At the 44,100 Hz rate files are read at, each piece of audio is 40 milliseconds long, faded smoothly so the joins stay quiet, and the next piece starts 20 milliseconds later. At the default 1.25× speed, each step reaches about 25 milliseconds further into the original, and the match is chosen from roughly 12 milliseconds either side of the ideal spot. The finished length is the original length divided by the speed. Set the slider to exactly 1× and that search is skipped: you get a copy of the audio instead of a processed one.

With pitch preservation off, 2× is exactly one octave higher

Unchecking Preserve pitch changes the length by keeping or adding samples, rounded down, and never leaves you with an empty file. At 2× you keep half the samples. At 0.5× you get twice as many. Each new sample is a smooth blend of its neighbors, and the ends of the file stay on the first and last samples instead of inventing sound past them. Because the result still plays at 44,100 Hz, the pitch moves by the same factor as the speed. 2× is one octave up, 0.5× is one octave down, and 1.25× is about 3.86 semitones — twelve times the base-2 logarithm of 1.25. The shortcut buttons are 0.5, 0.75, 0.9, 1.1, 1.25, 1.5, and 2, and the slider can still sit on 1.0×.

Related tools

More: Video & Audio