Try another backing-track key
Transpose an existing instrumental to try a different range. For a mixed song, follow the karaoke preparation guide first. Pitch shifting does not remove vocals or recreate missing instrumental detail.
Free tools for your next audio edit
Move a song excerpt, backing track or sound up or down in whole semitones without intentionally changing its tempo. Compare the actual export, then download WAV or MP3. Your selected audio stays in this browser.
Four seconds of an original synthesized tone, not a song or AI result. Loading replaces the selected file and keeps your pitch setting. Try +2 semitones for about 493.88 Hz, or +12 for 880 Hz; then prepare and compare the result. It is not a speech or music quality test. Download the original WAV · Source and CC0 license notes.
Choose the tonic names you already know. This does not detect the recording's key or change major to minor.
C to D: +2 semitones by the nearest route. A six-semitone tie chooses upward; choose −6 manually if you prefer downward.
This sets the pitch control and clears a previous result. Prepare a new export to apply it. Use the slider for another octave or direction, and keep the same major/minor mode when interpreting the target key.
No file selected. Audio stays in this browser.
Pitch shifting can change attacks, texture and voice character. Formants are not preserved. All sounds move together; this does not detect a song's key or separate a singer. The original nominal duration and nonzero processed tail are retained, so actual duration can extend slightly. MP3 may add codec delay or padding. At zero semitones, pitch processing is bypassed but decoding and export still apply. Keep your original.
Transpose a working copy
Use mono or stereo audio up to five minutes and 50 MiB. Preview it first. The file is decoded locally rather than uploaded.
Negative values lower pitch; positive values raise it. Start with a small change, choose WAV or MP3, and prepare the actual export. Zero bypasses pitch processing.
Compare opening attacks, sustained notes and the ending with your original. Read the actual duration and file size. Save a useful result before changing settings or leaving.
Pitch and tempo are separate choices
A semitone is one step in the twelve-step equal-tempered octave. Moving up twelve semitones doubles a tone's frequency; moving down twelve halves it. The frequency ratio is 2 raised to the power of semitones divided by twelve. This tool applies that change to the complete selected recording, rather than recognizing notes or deciding which key suits your voice.
For example, moving a known C-major backing track up two semitones transposes it to D major. Moving it down two gives B-flat major. This is transposition, not conversion between major and minor. If the track contains singing, that singing moves with the instruments.
Use the separate audio speed changer when you want a slower practice copy with its notes unchanged. This pitch tool combines resampling and time stretching to keep nominal tempo. It retains the source's nominal duration and any nonzero processed tail; it does not promise sample-accurate alignment with the original.
Prepare the source you actually need
Transpose an existing instrumental to try a different range. For a mixed song, follow the karaoke preparation guide first. Pitch shifting does not remove vocals or recreate missing instrumental detail.
Use the trimmer to keep a representative phrase. Compare a small shift with the original; large shifts can make attacks grainy or voices unnatural. No automatic key detection or tuning is applied.
Shift an already selected sound for a design experiment. All audible sources change together. If the effect is buried in a mix, the separate AI effect extractor has an account and credit flow.
Keep the original
The browser decodes at 48 kHz. WAV writes 16-bit PCM; MP3 is re-encoded at 128, 192 or 320 kbps. Tags, artwork and original bit depth are not copied. MP3 can introduce additional codec delay or padding. Compare the actual downloaded file in your editor, including the beginning and ending.
Pitch processing can alter texture, transient shape, phase and peaks. Vocal formants are not preserved. No normalization, limiter or edge fade is applied. Out-of-range processed samples clip during integer PCM preparation; lower excessive source gain in an editor before retrying. Use the loudness meter to measure a new copy when levels matter.
Processed output is limited to five minutes; WAV output must fit 50 MiB. A retained tail that exceeds a limit causes an error instead of being silently cut. Cancel preparation to discard pending work. Each attempt starts from the selected original, and changing settings clears the previous result.
Processing uses pinned SoundTouchJS 2.1.1 with Lanczos interpolation. License, modification notes and corresponding source are available. Synthetic native checks establish frequency and duration behavior, not perceived quality on your recording.
MP3 encoder credits
The optional MP3 download uses lamejs 1.2.1 by Alex Zhukov, a JavaScript port of the LAME MP3 encoder. Its unmodified script is served separately. See the attribution and source notes, upstream license notice, LGPL-3.0 and GPL-3.0 texts, and complete upstream source package.
A few useful answers
Yes. Local pitch processing, preview and downloads need no account or credits. The separate AI isolation service has its own account and download-credit requirements.
The processor combines resampling and time stretching to change pitch while keeping nominal tempo and source duration. A nonzero processed tail is retained and reported, so the actual duration may extend slightly. MP3 can add codec delay or padding.
Choose a whole number from −12 to +12. Twelve semitones is one octave. This does not detect your song's current key. The optional known-key calculator finds the nearest tonic interval; apply it to the control, then prepare and listen to a new export. It does not convert major to minor.
Every sound in the selected file moves together. For a separate vocal or backing track, prepare that source first. This tool does not identify voices, preserve vocal formants, or extract stems.
Mono or stereo MP3/WAV up to 50 MiB and five minutes. The browser decodes at 48 kHz. Download 16-bit PCM WAV or a newly encoded MP3 at 128, 192 or 320 kbps. WAV output must fit 50 MiB and all processed output must fit five minutes.
It can change attacks, texture, phase and voice character, especially at larger shifts. No formant preservation, automatic gain, limiter or lossless processing is promised. Keep the original and listen to the actual result.
After preparing your audio