Will a stretched voice still sound like a voice?

Within reasonable limits, yes, because the pitch is untouched and the grains are aligned to the waveform. Large stretches can still sound unnatural on speech, because the spacing between the harmonics stays fixed while the events around them spread out, which no time-domain method fully resolves.

Which material stretches best?

Sustained material: pads, room tone, drones, strings and reverb tails, where there are no sharp events to place wrongly. Percussive loops are the hardest case, because a drum hit is a very short event and the grain grid can only land near it rather than on it.

Why does stretching take longer than a speed change?

Because every output grain searches a small range of positions for the one that continues the waveform best, and that search costs roughly a second of processor time per minute of audio. Resampling has no search at all and is effectively instant.

Should I stretch in one big step or several small ones?

Several small steps usually hold up better. Each pass only has to move grains a short distance, so less misalignment accumulates, and the total change is the product of the factors. It costs more processing time, which is the trade.

Does the length come out exactly as the multiplier says?

The output length is driven by the factor applied to the sample count, so the duration follows the arithmetic closely. Reading the reported before and after figures is the reliable way to confirm what a particular file actually produced.

Time Stretcher

You Might Also Need

Tempo Changer

Adjust tempo while preserving pitch

Playback Speed Calculator

Calculate speed change ratios

Audio Speed Changer

Change playback speed, with or without pitch change

Sample Rate Calculator

Calculate sample rate requirements

Time Stretcher makes a recording longer or shorter without touching its pitch. The control is a length multiplier, so the job is stated in the units a sound editor usually thinks in.

What is Time Stretcher?

Choose a multiplier, drop a file, and process. The report gives the factor, the duration before and after, and confirmation that the pitch came through unchanged.

The controls:

A single slider from half length to double, in steps of a twentieth, defaulting to one and a quarter.
A report line with the multiplier, the two durations, and the statement that the pitch was preserved.
No mode switch and no secondary settings, because a length multiplier fully describes the operation.
Up to two channels, so a stereo file stays a stereo file.

Stretching works by overlapping grains. The audio is cut into short windows that overlap each other, and the output rebuilds the signal by placing those windows at new positions — closer together to shorten, further apart to lengthen. The naive version of that produces a hollow, phasey result, because grains that were never adjacent get crossfaded into each other. The method used here searches, for every output grain, a small range of positions for the one whose onset best continues what has already been written, which keeps successive grains on the same part of the waveform. That search is the entire reason a stretched voice still sounds like the same voice rather than like a chorus of it, and it is also why a long stretch takes noticeably longer than a resample would. A multiplier below one removes time and a multiplier above one inserts it, and the reported duration is the source duration multiplied by the factor. The limit is material-dependent rather than numerical: pads, ambiences and sustained tones stretch beautifully because they have no sharp events to misplace, while percussive loops are happiest within twenty percent or so, since a drum hit is a very short event that a grain grid can only approximate. Nothing about the pitch moves in either direction, which is what separates this from simply playing the file faster.

What it cannot fix:

Extreme factors sound smeared on transient-rich material.
No pitch control on top, so a stretch and a shift are two separate passes.
One factor for the whole file, with no gradual change.
No tempo or beat information is read, so the result is proportional rather than bar-aligned.
Formants are not handled, so a large stretch on speech can sound unnatural even with the pitch intact.
Output is sixteen bit WAV.

Use it for fitting a piece of audio to a fixed slot in an edit, slowing a phrase down to hear it in detail, or extending an ambience that was recorded slightly too short. When the change should be described as a tempo in beats per minute, the tempo changer takes the same ratio from the other direction, and when the pitch needs to move as well, the pitch shifter and speed changer cover that ground. Two things are worth knowing before a stretch. The output length follows the multiplier applied to the sample count, so the duration you get is the duration the arithmetic implies rather than an approximation. And because grains are placed by search rather than by rounding, the cost of a long stretch is processor time, not quality, which means a file that takes a while to process has not been damaged by the wait.

How to use Time Stretcher

Drop an audio file onto the panel and wait for the duration to appear.
Move the slider to the multiplier you need, above one to lengthen and below one to shorten.
Press process and read the reported before and after durations.
Play the result and download it if the pitch and transients hold up.

When to use Time Stretcher vs related tools

Time Stretcher is the right shape when a piece of audio has to fit an exact duration, such as a cue that runs two seconds long. Tempo Changer states the same change as a beat rate instead, and Audio Speed Changer can move speed and pitch together when that is what the material wants. To work out what a proposed change does to the running time, Playback Speed Calculator does the arithmetic without processing any audio.

Privacy & Security

This tool runs entirely in your browser — no data ever leaves your device. There is no server round-trip, no upload, no logging, and no account required. Your input is processed locally using client-side JavaScript and is never stored, transmitted, or accessible to anyone else. When you close the tab, everything disappears.