An audio pitch changer raises or lowers the perceived musical pitch of an audio file by whole semitone steps, and the browser-based Audio Pitch Changer applies a shift from -12 semitones (one octave down) through +12 semitones (one octave up) using the playback rate 2^(n/12). The shift happens entirely inside your current browser tab — no file upload is involved, no account is required, and nothing leaves your device. After you set a shift value, the tool re-encodes the file as a fresh uncompressed PCM16 RIFF/WAVE file that you download directly. Because the shift is implemented through playback-rate resampling rather than a time-stretching algorithm, raising the pitch also shortens the audio and lowering it lengthens it. A +12-semitone shift runs at 2× speed and produces roughly half the original duration; a -12-semitone shift runs at 0.5× and produces roughly twice the original. That trade-off is the central thing to understand before you start, since it determines whether this kind of pitch change is right for your material.
What the Audio Pitch Changer actually does
This kind of tool is the right fit when you want a quick, self-contained way to transpose music, voice clips, or sound effects through a fixed range without installing a full editor. The full reachable range covers one octave down (-12) and one octave up (+12), with 0 leaving the source at its original pitch. Whole semitones only — fractional cent values are not exposed. Twelve-tone equal temperament applies to every shift: each step changes pitch by the same ratio, and a twelve-step jump covers exactly one octave. A semitone itself contains 100 cents, so a +12 step is twelve equal steps of 100 cents each, totaling 1,200 cents — which lines up with the W3C Web Audio definition that detuning contributes a multiplicative factor of 2^(detune/1200) (see W3C Web Audio: AudioBufferSourceNode). Pitch shifts apply uniformly across the whole file; there is no per-region control and no key/scale preset, just one integer value applied to everything.
If you want a structured check on whether pitch-shifting is the right move for the job you have in mind, see Do I Need an Audio Pitch Changer? Here's How to Tell.
What you'll need before you start
Three things make a first run go smoothly: a file your browser can actually decode, awareness of the size ceiling, and a clear expectation that duration moves with pitch.
| Format | Typical extension | Decode note |
|---|---|---|
| MP3 | .mp3 | Widely supported across browsers |
| WAV | .wav | Always accepted when PCM-based |
| M4A / AAC | .m4a, .aac | Browser-dependent codec profile |
| Ogg | .ogg | Most modern browsers |
| WebM audio | .webm | Browser-dependent |
| FLAC | .flac | Browser-dependent |
The exact codec profile still depends on your current browser; a recognized extension does not guarantee every unusual variant will decode. If the file refuses to load, re-exporting it as a plain MP3 or WAV from any standard audio editor usually fixes it. The single-file cap is 50 MiB, measured on the encoded bytes — so a typical three-minute MP3 or WAV is comfortably under the limit, while a 60-minute lossless recording will not fit.
Before clicking render, check that your input clears the decoding budget: decoded length at most five minutes, at most eight channels, at a sample rate between 8,000 and 192,000 Hz. Decoded input is capped at 30 million channel samples. The pitch-shifted output is budgeted separately at 30 million channel samples — and lowering pitch grows the output, so an input that fits inside the input budget can still fail the output budget after a strongly negative shift.
Shift your first clip in three steps
The whole workflow reduces to file, value, download. There is no inline audio element on the page because the site media policy disallows a blob-backed preview; you hear your result by opening the downloaded WAV in any local audio player.
- Choose one browser-decodable audio file no larger than 50 MiB. Click the file picker and select a single MP3, WAV, FLAC, Ogg, WebM, or M4A/AAC file. The page never accepts more than one file and never batches multiple clips into one render.
- Set a whole-number pitch shift from -12 through +12 semitones. Type the value directly or use the control to step it. Remember the formula: playback rate is 2^(semitones/12). Each whole integer maps to one ratio in twelve-tone equal temperament, ranging from 2× at +12 down to 0.5× at -12. Setting 0 returns an output that matches the original pitch and roughly the original duration.
- Create the complete local render, review the stated duration change, and download the PCM16 WAV. Read the duration delta the tool reports before saving. Every change to the input file or the semitone value invalidates the previous render, so re-render after each edit. Save the resulting .wav locally.
Worked example with the formula: for a +12-semitone shift on a 10-second clip, the playback rate is 2^(12/12) = 2^1 = 2.0, so the rendered output runs at twice the speed and lands near 10 ÷ 2.0 = 5 seconds. For a -12-semitone shift on the same clip, the rate is 2^(-12/12) = 2^-1 = 0.5, producing roughly 10 ÷ 0.5 = 20 seconds. The relationship is purely multiplicative, the standard equal-tempered behavior, where an octave contains twelve equal semitones and each semitone equals 100 cents (1,200 cents per octave).
Limits the render will refuse
Validation happens before the expensive offline render begins, so a refused file never produces a half-written WAV. Inputs over 50 MiB are rejected at the file-reading step. Empty files and unsupported MIME types fail at decode. Decoded audio outside five minutes, eight channels, or 8,000–192,000 Hz fails with an explicit message and no output is created. Decoded input that exceeds the 30 million channel-sample budget is rejected before the render starts.
The most surprising limit for first-time users is the output-side budget. Lowering pitch multiplies the required output frame count by roughly 2^(−n/12), so very negative semitone values on already long files can push the computed output past 30 million channel samples. The error message in that case states the requested shift is over budget and confirms nothing was truncated. The tool never partially writes a clipped WAV and never swaps in a fallback — a failed render simply creates no download. If the browser cannot allocate the full offline buffer, the job reports an error and no file is produced.
MDN's reference on AudioBufferSourceNode.playbackRate describes playbackRate as a proportion of the original sampling rate, with values below 1 slowing the audio and values above 1 speeding it up, which is exactly the resampling behavior this tool applies to the whole offline buffer. The result is accepted only when the browser returns the expected channel count, sample rate, and full frame count.
What comes out of the download
The downloadable file is a brand-new .wav, encoded as uncompressed, interleaved, little-endian PCM16 inside a standard RIFF/WAVE container. Floating-point samples are clipped to the legal -1 through +1 range before conversion to 16-bit integer PCM. The output does not inherit any of the source's codec, bitrate, ID3/Vorbis tags, embedded artwork, chapter markers, cue points, loop metadata, or container metadata. PCM16 quantization is also distinct from the floating-point source, so the downloaded bytes are not bit-identical to a hypothetical lossless re-encode of the decoded buffer. Browser resampling quality is implementation-defined, and very large upward shifts can expose aliasing or codec artifacts that were masked in the original.
The result is intended for creative experiments, practice backing tracks, sound effects, and rough editing — not as a transparent mastering-grade pitch process. Because the whole render happens locally inside the same browser tab, you can disconnect from the internet at any point after the page is loaded and the tool still works on every supported file. The output replaces any prior preview placeholder: download Object URLs are revoked when a new render takes their place, so a successful download never points at stale audio data.
First-time mistakes that ruin the result
The single most common expectation mismatch comes from a different family of tools. Tools that keep speech cadence, beat positions, and exact duration typically use a phase-vocoder or other time-stretching workflow — this implementation does not. If your project requires pitch change without duration change, the right move is to use that dedicated workflow; playback-rate resampling cannot satisfy it. For a focused walkthrough of where playback-rate resampling fits among other options, see Audio Pitch Changer: Practical Tips and Common Mistakes.
A second common mistake is assuming any filename with a recognized extension will decode. Containers that look ordinary on disk may rely on codec profiles your specific browser does not support; the reliable fix is to re-export the source as a vanilla PCM WAV or constant-bitrate MP3 first. A third mistake is expecting the export to carry the source's title, artist, or chapter data — every byte of the downloaded WAV is freshly synthesized, and none of the source's metadata survives. Finally, treating the page like a player will trip you up: there is no inline blob-backed preview by design, so the loop is file in, pick a shift, download, then open the WAV in any trusted local player to verify the result.