Change the pitch of a song or voice without changing its length

How to use the pitch shifter

1. Add an audio file: MP3, M4A, WAV, OGG, FLAC, WebM, or the audio of a video file.
2. Drag Shift from −12 to +12 semitones. It moves in 0.5 steps and starts at +2.
3. Pick Save as (16-bit WAV unless you change it) and a Bitrate for lossy formats.
4. Press Shift pitch. A shift of 0 returns “Choose a shift other than 0.”
5. Listen to the result and download it. The Frequency ratio tells you how much every pitch was multiplied.

What it does and when to use it

Singers use it to move a backing track into a comfortable key. Musicians use it to practise along with a recording that was tuned a little off, or to hear a part an octave lower. Video makers use small shifts to make a voice harder to recognise, or a large downward shift for a monster effect. Because the length does not change, the shifted audio still lines up with the original video or any other track.

For a speed change, use the audio speed changer. For the backwards effect, try the audio reverser. To check which note a recording sits on before and after, play it through the audio spectrum analyser, which shows the strongest frequency and the nearest note name.

How it works

The ratio is r = 2^(semitones ÷ 12). The tool first time-stretches the audio to r times its length while keeping pitch, using WSOLA: overlapping 46 ms frames, each lined up with the one before by searching a few milliseconds either side. It then resamples that longer version back to the original length with a band-limited (windowed-sinc) filter. Squeezing it back raises every frequency by r; for a negative shift the steps work the other way round. The result is padded or cut to the exact original sample count. All processing happens in a Web Worker on your device.

Worked examples

From the tool’s output on a 4-second recording:

Each run also shows a note that full mixes and shifts beyond ±5 semitones can sound phasey.

Limits and tips

• Voices shifted far up sound thin (“chipmunk”) because the formants, the resonances that give a voice its character, move too. That is expected with this method.
• Drums and other sharp attacks can soften slightly.
• Shift the original file rather than one that has already been compressed and re-saved.
• Compressed formats other than WAV need WebCodecs in your browser. AIFF, AMR, WMA and CAF input, and MP3 output, use the optional ffmpeg fallback, about 30 MB downloaded once from cdn.jsdelivr.net; the audio itself never leaves your device.