Voice & files · 0.25× – 4× · pitch preserved
Time Stretcher
Change the speed of a recording without changing its pitch. Load a file, choose a tempo, and the engine rebuilds the audio with overlap-add resynthesis so a 3-minute solo at 0.5× lasts 6 minutes and still sits in the same key. Loop a hard four bars and drill them, or push a podcast to 1.5×.
Processing happens entirely on your device - nothing is uploaded. Be realistic about the extremes: below 0.5× and above 2× the resynthesis smears transients and adds a faint warble, and exported files keep that. Turn the header level down before exporting long clips.
File
-
Duration
-
Sample rate
-
Channels
-
- / -
Ready. Choose an audio file to begin.
On - overlap-add resynthesis changes the length only; the pitch stays put.
Loop - → - · length -
Source length
-
Result length
-
Pitch ratio
1.0000
Tempo ratio
1.0000
With Preserve pitch off, the clip is resampled like a tape played faster - exactly the chipmunk effect, and the length still changes. It is the same maths; only the pitch handling differs.
Slowing down to learn, speeding up to listen
Practising an instrument from a recording is impossible at full speed if the passage is faster than you can play it. Slowing it to 0.5× or 0.75× while the pitch stays put lets you hear every note in the original key and play along; a loop over two bars makes the drill. Transcribers work the same way - half speed, small loops, no key change to confuse the ear. In the other direction, 1.25×–2× is the standard for spoken audio: podcasts, lectures and audiobooks, where the pitch can take a little roughness in exchange for finishing sooner.
Why a naive speed change shifts the pitch
Sound is a pressure wave, and pitch is how fast it repeats. Play a recording faster by reading its samples more quickly and every cycle finishes sooner: the frequency rises in direct proportion. At 1.5× a 440 Hz note becomes 660 Hz - a fifth up - and at 2× it is 880 Hz, an octave up. That is why speed-adjusted video so often sounds like chipmunks. Pitch and duration are tied together in a resample, because resampling changes the time axis itself. To move one without the other you have to rebuild the waveform at a new length rather than read it faster, which is exactly what overlap-add resynthesis does.
Tempo, duration and pitch are three settings
- tempoRatio sets the output length: sourceLength / tempoRatio.
- pitchRatio sets the frequency scale, independently.
- Preserve pitch on → pitchRatio = 1. Off → pitchRatio = tempoRatio (a resample).
| Tempo | 3-min clip becomes | Typical use |
|---|---|---|
| 0.25× | 12:00 | Note-by-note transcription of very fast playing. |
| 0.5× | 6:00 | Slow practice; the usual first pass at a solo. |
| 0.75× | 4:00 | Learning a part just under tempo. |
| 1× | 3:00 | Original - no processing. |
| 1.25× | 2:24 | Comfortable podcast speed-up. |
| 1.5× | 2:00 | Fast listening; still intelligible. |
| 2× | 1:30 | Skimming; consonants start to blur. |
| 4× | 0:45 | Extreme - heavy smearing, use only to locate a passage. |
What is time stretching actually used for?
Three things, mostly. Transcribing: slow a solo or a fast passage to half speed so you can write it down or play along, without the pitch dropping an octave. Practising: loop four bars at 0.75× and creep the tempo up as you learn them. And listening: speed a podcast or a lecture up to 1.5× or 2× to get through it faster while the voices still sound like voices.
Why does speeding audio up normally make it sound like a chipmunk?
Because a plain speed change is a resampling. If you play a waveform back faster, you are reading its samples more quickly, so every cycle of the waveform takes less time - the frequency rises. Frequency and duration are two sides of one number: play it twice as fast and everything moves up an octave. The only way to change the duration without touching the pitch is to rebuild the signal at a new length instead of just reading it faster, which is what overlap-add resynthesis does.
How does this tool keep the pitch when I slow it down?
With WSOLA - waveform-similarity overlap-add. The signal is cut into ~50 ms frames and re-laid onto a longer (or shorter) timeline; because each frame is a copy of the original waveform, its frequency content is unchanged. The synthesis hop sets the new duration while the frame contents keep the pitch. The processor also searches nearby material so consecutive frames stay phase-aligned, which is what keeps the result from turning into a metallic buzz.
What does the loop feature do, and how do I set the loop points?
Drag the two handles on the waveform to mark a section, then switch Loop on and playback repeats that section instead of running to the end - the usual way to drill a difficult bar. The default loop region is the whole file. Exports always render the whole file at the chosen tempo, not the loop.
Instrument Tuning Tones
Reference pitches for guitar, bass, ukulele, violin, cello, mandolin, banjo and viola.
Online Metronome
Accurate metronome from 20 to 300 BPM with time signatures, accents and tap tempo.
Voice Recorder
Record from your microphone, watch the live waveform and download the take as a WAV file.
Pitch Shifter
Shift any audio file by semitones, with or without tempo correction, then export the result.