Audio Speed Changer
Change audio speed from 0.50× to 2.00× in your browser with pitch-preserved or tape-style retiming, then preview the result and save a WAV.{{ summaryTitle }}
{{ summaryLine }}
Audio preview
The chart renderer is unavailable. The same values remain in the processing ledger.
| Signal | Value | Meaning | Copy |
|---|---|---|---|
| {{ row.label }} | {{ row.display }} | {{ row.detail }} |
Introduction:
Audio speed changes two closely related things: how long the recording lasts and how quickly its waveform is read. At 2× speed, a 30-second source becomes about 15 seconds. At 0.5×, it becomes about 60 seconds. The difficult choice is whether pitch should move with playback speed or stay near the original.
Tape-style retiming changes speed and pitch together. Faster playback raises pitch; slower playback lowers it. This can suit deliberate effects, sound-design experiments, or material where the linked change is expected. Speech, lessons, rehearsals, and music practice often need time-scale modification instead, where duration changes while vocal and musical pitch is kept approximately stable.
Pitch-preserving retiming works by dividing the waveform into short overlapping segments, aligning similar waveform regions, and blending the segments at a new spacing. It avoids the direct pitch shift of simple resampling, but large changes can still produce smearing, flutter, doubled attacks, or rough transitions. Dense music, percussion, reverberation, and mixed voices are harder to stretch cleanly than a steady single voice.
A short edge fade only changes the start and end of the result. It can reduce clicks caused by a cut that begins or ends away from a waveform zero crossing, but it cannot repair artifacts created throughout the recording. Listen to transients, sibilants, sustained notes, and the final boundary before using the file.
How to Use This Tool:
Choose one short audio file, then decide whether duration alone or both duration and pitch should change.
- Select one WAV, MP3, M4A, AAC, OGG, FLAC, or WebM file no larger than 12 MB and no longer than 60 seconds. Browser decoding must support the source format.
- Set Speed from 0.50× to 2.00×. Use Exact speed when the multiplier must be repeatable.
- Keep Preserve pitch on for speech or music that should stay near its original key. Switch it off for tape-style retiming where pitch should rise or fall with speed.
- Choose an Edge fade of 10, 25, or 50 ms only when the start or end clicks. No fade leaves the boundary samples unchanged.
- Preview the retimed audio and compare the source and output durations. Download the WAV only after checking speech clarity, attacks, sustained tones, stereo balance, and the first and last moments.
Interpreting Results:
The duration comparison should follow the inverse of the speed multiplier: higher speed gives a shorter file, and lower speed gives a longer one. Small frame-rounding differences are normal because audio duration is represented by a whole number of sample frames.
Pitch preservation is an audio-quality policy, not a guarantee that every note or voice remains perfectly unchanged. Compare the preview with the source. If consonants, percussion, or sustained notes sound unstable, move the speed closer to 1× or try tape-style retiming when a linked pitch change is acceptable.
Technical Details:
Decoded mono or stereo samples are processed at 44,100 Hz. The output length is the source frame count divided by the selected speed, rounded to a whole frame. The result is encoded as 16-bit pulse-code modulation (PCM) in a WAV container.
Formula Core:
For source duration T and speed multiplier s, the expected output duration is T divided by s.
At 1.5×, a 30-second source becomes about 20 seconds. The approximation symbol reflects whole-frame rounding rather than a different speed rule.
Transformation Core:
| Path | Transformation | Audible consequence |
|---|---|---|
| 1× | Retain the decoded sample sequence | Duration and pitch remain unchanged apart from an optional edge fade and WAV encoding |
| Tape style | Read source positions at the selected rate and interpolate between sample frames | Duration changes and pitch moves with speed |
| Preserve pitch | Use waveform-similarity overlap-add with short weighted frames and correlation-based alignment | Duration changes while pitch stays approximately stable, with possible stretch artifacts |
| Edge fade | Apply a linear gain ramp to the first and last 10, 25, or 50 ms | Boundary clicks may soften; the middle of the recording is unchanged by the fade |
The pitch-preserving path uses roughly 20 ms analysis windows with 50% synthesis overlap and searches a nearby region for a similar waveform before overlap-add. A Hamming-style window weights each frame, and overlapping contributions are normalized. If a region lacks enough usable overlap, interpolated samples provide a bounded fallback.
Output samples are limited to the PCM range before 16-bit encoding. The finished WAV is decoded again and its frame count is checked against the reviewed result, so preview and download use the same locally generated audio.
Privacy and Quality Notes:
Selected audio is decoded, retimed, previewed, and encoded in the browser tab; it is not uploaded for processing. Local processing does not remove normal risks from clipboard use, downloaded files, browser extensions, or a shared device.
- Format support depends on the browser's audio decoder even when the file extension is accepted.
- The output is 44.1 kHz, 16-bit PCM WAV, so it does not preserve the source codec, bitrate, metadata, or exact sample format.
- Avoid repeated retiming passes; process the original source at the final multiplier when possible.
- Keep an untouched source copy and listen through the complete output before publishing, transcribing, or synchronizing it.
References:
- Waveform similarity based overlap-add for time-scale modification of speech, Roelands and Verhelst, Eurospeech 1993.
- Web Audio API, W3C Recommendation, 2021.
- Resource Interchange File Format and WAVE, Microsoft Learn, 2021.