Choose Local Audio
Drop a mono or stereo file or use Choose File. After decoding, check its displayed name, duration, sample rate, and waveform.
Ready — sound is off
WAV or MP3 recommended. AAC/M4A and OGG depend on your browser. Up to 100 MB; decoded memory limits also apply.
Processed locally in your browser0:00 / 0:00Source waveform · click to seek
1.0000× frequency · original duration
Pitch changes keep the file duration. Processing can introduce artifacts. To change tempo while retaining pitch, use the Time Stretcher.
Browser audio signal toolkit
This pitch shifter changes the pitch of a local recording while retaining its duration. Choose semitones and cents, compare Original and Shifted audio, then export the processed samples as a WAV file.
A pitch shifter moves the frequency relationships in recorded sound upward or downward. A shift of +12 semitones raises a steady note by an octave; −12 lowers it by an octave. Ordinary playback-rate changes also change duration. This pitch shifter instead combines resampling with time-domain stretching so the output contains changed pitches across the original number of sample frames.
The file stays on your device. Decoding, waveform reduction, processing, preview, and WAV encoding use browser APIs and a local worker. The tool contains no audio-upload endpoint or conversion-service fallback. It starts sound only when you press Play, and it uses a conservative preview gain. Processing can color the sound, particularly at large intervals; keeping the timeline stable does not imply a transparent or studio-mastering result.
Three simple steps
Choose a short recording, set the musical interval, and compare the processed result before exporting.
Drop a mono or stereo file or use Choose File. After decoding, check its displayed name, duration, sample rate, and waveform.
Choose −12 to +12 semitones and optional −100 to +100 cents. Presets set whole musical intervals; Reset returns both controls to zero.
Play Shifted to process the result, switch to Original for comparison, seek to a passage, then export the complete processed WAV.
The pitch shifter keeps duration and changes pitch. The Time Stretcher changes duration while aiming to keep pitch stable. Suppose a passage lasts 20 seconds and contains A4 at 440 Hz. A +12 shift requests approximately 880 Hz while the passage still lasts 20 seconds. A 0.5× time stretch instead requests a 40-second passage with its tonal pitch approximately unchanged.
These are separate practice tasks. Choose a pitch shifter when a backing track needs a different key; choose time stretching when a fast passage needs a manageable tempo. Neither tool identifies the recording’s original key or reconstructs individual instruments from a mix. The waveform is a timeline overview, not a pitch spectrum or an automatic musical transcription.
For slower or faster playback with approximately stable pitch, choose the Time Stretcher.
To check your instrument against the shifted rehearsal track, use the Instrument Tuner for a clear single note.
In twelve-tone equal temperament, a semitone is one-twelfth of an octave and 100 cents equal one semitone. The pitch shifter combines both controls with the ratio 2^(semitones/12 + cents/1200). Thus +7 semitones is about 1.4983 times the source frequency, +100 cents matches +1 semitone, and −12 gives a ratio of 0.5. The displayed ratio describes the request, not a calibrated acoustic measurement.
Use cents for a small reference adjustment or deliberate detuning. A4 at 440 Hz shifted to 432 Hz needs about −31.77 cents; −32 cents is a practical rounded entry. That changes every frequency proportionally, not only the A notes. The pitch shifter accepts a fine-tuning range of one semitone in either direction in addition to its octave-range main control.
Choose a positive semitone value to raise the recording and a negative value to lower it. The +2 preset moves a part up a whole tone; −5 moves it down a fourth; +12 moves it up an octave. A pitch shifter is useful for matching a rehearsal backing track to a comfortable register, exploring sample textures, or comparing the same melody in several keys.
The presets are intervals rather than key labels because the source key is unknown. If you know a song is in C and want D, choose +2; if it is in E and you want D, choose −2. Verify the result with your instrument or a reference note. Do not assume that changing pitch grants permission to reuse or redistribute the recording.
A background worker first resamples the source with windowed-sinc interpolation. For upward shifts it applies an anti-alias cutoff to reduce frequencies that would exceed the representable range. The pitch shifter then applies the site’s WSOLA processor at the reciprocal tempo ratio. Short overlapping windows are aligned with nearby source material, which adjusts the timeline without simply undoing the new pitch.
The final buffer has the original frame count at the decoded sample rate. This preserves total duration exactly at that rate, while local transients can move slightly within alignment windows. The pitch shifter uses the same processed buffer for preview and export, so a WAV is actual transformed audio rather than an unchanged file with a playback-rate instruction. Small edge fades and peak protection reduce boundary clicks and clipping.
Select Original to hear the decoded source or Shifted to hear the current processed settings. A/B switching pauses the old source and starts the selected version at the same timeline position. When a new shift needs processing, there is a preparation pause. The pitch shifter does not silently reset the passage to zero on each comparison, so short musical details are easier to judge.
Play, Pause, Stop, and the seek slider refer to the original-duration timeline. Stop returns to the beginning. Changing pitch pauses playback and invalidates the earlier processed version, which prevents exporting stale settings. Process Shifted Audio prepares the result without starting sound. Cancel stops the worker, and the tool remains ready for another attempt.
For practice, begin with a modest interval such as −2 or +2 and listen to attacks and sustained tones. For sound design, larger intervals can create useful textures even when the algorithm’s coloration is audible. A pitch shifter moves formants along with the audio; this version has no independent formant control, so voices may acquire a thinner or heavier character.
Dense mixes, percussion, reverberation, and extreme pitch changes can expose repeated texture, softened attacks, or rough joins. Upward shifts can discard high-frequency information, and downward shifts cannot invent detail absent from the source. The pitch shifter is a practical practice and prototyping tool. Audition the exported file, especially before placing it in a finished project.
The pitch shifter accepts common audio files but relies on your browser’s decoder. WAV and MP3 are useful starting choices; AAC/M4A, OGG, and FLAC depend on browser and operating-system support. Corrupt or unsupported files show a readable error. Mono and stereo are supported; multichannel files are rejected rather than silently discarding channels.
The compressed-file guard is 100 MB and decoded audio is limited to 30 minutes and 256 MB. Pitch processing uses a stricter estimated 256 MB working budget, allowing for source copies, resampled buffers, overlap accumulators, preview, and WAV encoding. In practice this requires much shorter files, especially at high sample rates or on phones. A small compressed file can still decode into substantial memory. Close other heavy tabs or process a shorter excerpt if allocation fails.
Export Processed WAV prepares the current shift if necessary, then encodes the complete result as 16-bit PCM WAV at the decoded sample rate. The output retains mono or stereo channels and has a cleaned filename with semitone and cents values. It does not export only the visible passage. Zero shift exports decoded original samples without an intentional musical change.
The pitch shifter releases buffers when you replace or remove a file, terminates workers on cancellation or navigation, and revokes temporary download URLs. Hiding the page pauses preview. No file names, waveform samples, or audio contents are attached to product analytics. The privacy policy explains the site’s optional general analytics separately from local audio processing.
Clear answers
The exported buffer retains the original frame count and total duration. WSOLA alignment can move individual attacks slightly; tempo preservation does not guarantee artifact-free music.
Yes. Both request the frequency ratio 2^(1/12). Semitone and cents controls add before the ratio is calculated.
No. File reading, decoding, processing, playback, and WAV encoding take place locally. There is no remote audio-processing request.
Some algorithms offer that separately. This implementation does not preserve formants independently, so shifting a voice changes its apparent vocal character.
Compressed size does not show decoded memory use. A long or high-rate file can exceed the working-memory budget even below 100 MB. Try a short mono or stereo excerpt.