Online Vocal Pitch Correction
Open a local mono or stereo file, select the time range and apply your change. The rendered download is 16-bit PCM WAV. Supported input codecs depend on the browser.
File limits are 25 MiB, 60 seconds and 16 MiB decoded PCM. Creative pitch and time edits can change the sound of transients. Audio is processed locally.
Use an isolated mono vocal. Listen to the result before keeping it. Complex accompaniment, fast changes and extreme settings can create artifacts.
| Clip seconds | Target | Decision |
|---|
Ready.
Related tools
Use this online vocal pitch correction tool to bring accepted notes in an isolated vocal recording closer to a chosen scale. Open a local file, select the line you want to work on, and choose Chromatic, Major or Natural minor. Set a root for the keyed scales and choose how strongly to pull detected pitches toward their targets. Listen to the original and corrected previews before downloading a 16-bit PCM WAV. The best first test is a short solo line with a mistake you can already hear.
The editor works with one prominent voice. Pitch decisions depend on whether the recording gives the analyzer enough periodic information. The target-note table shows which parts of the selected clip produced correction decisions. Uncertain and unvoiced passages can remain dry. A silence or breath that receives no target is an appropriate outcome.
Correct a short vocal phrase first
- Choose an isolated monochannel vocal recording. A dry practice take or exported solo vocal stem gives the tracker a clear signal.
- Enter start and end times in source seconds. Select the phrase plus a little space on both sides so you can hear transitions.
- Choose a target scale. Set the correct root when using Major or Natural minor. If you only need smaller adjustments to the nearest equally tempered notes, start with Chromatic.
- Choose a correction strength and render the selection. Try a moderate setting for the first version and compare it with the source.
- Check the targets at the places you heard a problem, then listen all the way through. Download the WAV version you prefer.
Each file has limits of 25 MiB, 60 seconds and 16 MiB of decoded PCM. The first applicable limit controls. The file’s compressed size alone cannot tell you how much decoded memory it needs. Input decoding depends on the browser. Try a short exported WAV from your recorder if an otherwise valid file cannot be opened.
Choose the musical targets
Chromatic correction makes all twelve pitch classes available. It is a useful baseline for an unaccompanied melody where each intended note is already reasonably close. Major and Natural minor constrain the available choices to their seven-note scales. That restriction can help a slightly drifting note stay within the key. It can also pull an intentional outside note in the wrong direction.
For example, C major contains C, D, E, F, G, A and B. C natural minor contains C, D, E flat, F, G, A flat and B flat. The word C alone cannot resolve whether the intended third is E or E flat. A phrase also can borrow an outside note temporarily. First confirm the intended melody against the accompaniment or a reference instrument, then use a scale that fits that passage.
A note that landed very far from its target can sit closer to another allowed target. If the render repeatedly corrects a missed D toward C, inspect that passage in the original and confirm what you sang. Re-recording the line closer to the melody may give a better result than raising the correction strength.
| Recording goal | Setting to try | What to listen for |
|---|---|---|
| Keep a gently drifting held note centered | Chromatic with moderate strength | Whether the steady part settles while the entrance still sounds like the take |
| Tune a line in a known minor key | The intended root and Natural minor | Whether the notes agree with the actual chord progression |
| Retain an intentional slide or bend | Compare a lower-strength render with the original | Whether the slide remains audible and continuous |
| Move a line into another arrangement key | Choose the Vocal Pitch Changer instead | Whether the whole saved phrase is shifted by the intended interval |
Set strength by listening to the musical result
Strength controls how much correction is applied toward the target. It describes the render setting. Treat a high value as an experiment for a pronounced tuning effect, rather than a measure of singing ability. Start from the same source clip at each strength, keep the other controls fixed, and audition the last two notes, the first entrance and the transition across the trickiest interval.
Vibrato and expressive slides create real pitch movement. That variation can be part of the style. A singer may also deliberately approach an in-tune note from below. When the original attack feels musical, use a conservative strength and assess the sustained portion separately. If the tail of a note sounds pulled toward a neighbor as the singer releases it, shorten the selected clip or try a version with gentler correction.
For a practical comparison, render a verse line once at a gentle setting and once at a stronger setting. Name each download by the settings you used. Listen in the arrangement with the same backing track level each time. Pay special attention to whether the start of the line joins the previous phrase cleanly once you place the selected audio back into the session.
Read the target-note decisions
The table is displayed for the selected clip. Its time column is measured in clip seconds. If the selection starts at source second 12, an entry three seconds in belongs around source second 15. Use that relationship to find the affected moment in the original file and place a selected render correctly in an editor.
A target is a chosen comparison note under the current settings. A breath, consonant, silence or noisy fragment may provide too little evidence for a stable pitch. Strong backing instruments can also pull the detector toward a different prominent frequency. Inspect the table in conjunction with the original recording, particularly when the indicated note seems disconnected from the singer’s audible line.
Play across boundaries between accepted and uncertain sections. The label of a stable region can help you find the location, while your ears tell you whether a syllable or transition sounded natural. If the table fails to identify much of a clearly sung melody, try a recording made with less room noise or a cleaner stem. Keeping the source file available makes that comparison straightforward.
Distinguish correction, transposition and formant shifting
Scale-aware correction is for nudging local pitches toward selected notes. Uniform transposition moves the entire selected recording by the same interval. For a melody you already sang accurately in a different key, use the Vocal Pitch Changer and choose the required semitone offset. Retuning the old take to the destination scale alone will not preserve the same melody intervals.
The Formant Shifter changes the spectral envelope to alter perceived vocal color. A voice can be measured at the same fundamental and still sound darker or brighter due to changes in its resonances. The formant editor gives a creative control plus an estimated input/output pitch comparison when it can make one. Use that effect after you understand the pitch changes you intend.
Use the Vocal Pitch Monitor for a live view before you record. Sing the same note twice at a comfortable volume, watch the steady section of the trace and identify the place where it drifts. The monitor’s numeric CSV describes its analysis. Keep the recording you intend to tune as a separate playable audio file.
Review the WAV in its original musical context
The downloadable result is PCM WAV from the selected material. Keep the source take and add the render on a separate track. Line up the start with the same source position you selected. A render that begins with a part of the original verse belongs at that verse’s time; dropping it at the project origin makes it sound out of time.
Listen for doubled vocals if both the original and corrected versions are audible simultaneously. Mute one, compare at similar playback levels, then bring in the accompaniment. A pitch adjustment that sounds pronounced in isolation can blend well in context. Conversely, an apparently smooth solo correction can clash with the chord it lands on. Check both settings before replacing a section.
To compare renders with a bandmate, send them the source start and end times, the selected scale and root, and the strength. They can place the excerpt consistently and tell you which syllable needs more work. Retain the dry take so returning to the previous phrasing is straightforward.
Fix a correction that sounds wrong
- The target is consistently a different note. Check the key and mode, then listen to the input with a reference tone in a separate track. An inaccurate take may be closer to an unintended target.
- The vowels sound unstable. Try a smaller correction and select a shorter phrase. Fast movement, strong vibrato and very large changes can make artifacts more noticeable.
- A long break is unchanged. Compare its row state and the source audio. Unvoiced material may have no defensible pitch target.
- Backing instruments show up in the decisions. Export the vocal track on its own. Lowering volume on an already mixed file changes overall level without separating voices.
- The preview plays the wrong section. Recheck start and end times in the source. Listening to the same section in the raw recording can catch an offset before another render.
Use the correction as feedback during practice
Begin with a single line that you can repeat comfortably. Record it dry, write down the measure where you hear a pitch problem and compare the original with a modest render. If the difference occurs throughout a held vowel, practice sustaining that note against a quiet reference before recording a second take. If the main difference occurs on a jump between notes, practice the transition as a two-note phrase first.
Use a new recording for the comparison after practice. Correcting an already corrected file hides the connection to the way you originally sang it. If the new dry take already fits the phrase, keep it. That choice retains the natural timing and phrasing you recorded and makes any remaining editing decisions easier to hear.
For a line with two singers, process each isolated track separately only if those stems are available. Confirm the target for each voice using the harmony parts, then audition the two against the original accompaniment. A target that makes sense for the lead may be unsuitable for the harmony, especially when their lines move through different chord tones.
Plan around local processing and browser limits
The tool renders in the browser from the file you pick. It does not send that audio or the pitch decisions to the site for rendering. A WAV you save lives in your device downloads according to your browser settings. Retain anything you need before reloading or closing the page. The website can still load ordinary page assets, and your browser or machine may have its own history and storage policies.
On a phone, let the render finish before playing the preview, then use headphones to hear the same detail you listened for on the source. Keep the clips short when the device struggles to load or decode them. Preparing shorter mono excerpts in the original editor solves both duration and decoded memory limits without requiring you to discard any of the full track.
For a very quiet take, compare the source with headphones at a reasonable level before assuming the processor received silence. A distant microphone and noise floor can obscure periodic sound. Try a clean recording with stable level. If the original recording is distorted, a new take provides clearer pitch information than applying another round of processing to the damaged one.
Frequently asked questions
Can I use this to tune singing without a microphone prompt?
Yes. The correction editor works on an existing local audio file. Record or export the passage first, then select it using the file control. A live microphone session in the pitch monitor is a separate workflow.
Which setting is best if I do not know the key?
Chromatic lets the tool consider all twelve pitches. Listen carefully to the results, especially if a sung note landed between intended pitches. Work out the melody or key separately before attempting restricted correction.
Does a 100 percent strength setting mean my singing was exactly on pitch?
Strength sets how much the processor attempts to move accepted pitches toward selected targets. It does not certify a recording or give a grade. The target choice, input quality and musical context all matter when you decide to keep the edit.
Can I correct the vocals in a full song?
Use a solo vocal stem for best results. The tool operates on one prominent pitch stream and cannot reliably separate the lead vocal from accompaniment or harmonies in a finished mix.
What file do I share after correction?
Download the 16-bit PCM WAV. Share the WAV when your collaborator needs to hear or edit the result. Include its source start time if you rendered part of a longer take.