Autotune is pitch correction. Software detects the pitch of a vocal, compares it against a chosen key and scale, then moves it to the nearest allowed note. One control, retune speed, decides how it sounds. Slow settings correct a take transparently. Near-zero settings snap every note into place, which is the hard-tuned sound.
Ask what autotune is in a studio and you get two answers, because the word does two jobs. Auto-Tune is a specific product, a trademark of Antares Audio Technologies. Autotune, lowercase, is the general technique of correcting vocal pitch, and several companies build tools for it. The distinction matters the moment you open a manual or argue about a vocal in a mix.
This guide covers the mechanism instead of the mythology. What the software measures, which single control produces the sound people recognise from Cher and T-Pain records, how that differs from note editing in Melodyne, and where correction stops being repair and becomes an instrument. If you are still putting a production workflow together, Music Production Basics covers the wider picture from DAW setup through mastering.
Auto-Tune the Product, Autotune the Technique
Auto-Tune is Antares' trademark, not a category. Antares' current pages, read on August 26, 2026, describe AutoTune Pro 11 as its flagship tuner; its release notes list version 11.6.0, while the online manual remains labeled Version 11.0.0. The same pages list AutoTune 2026, AutoTune Hybrid, AutoTune EFX+ 10, AutoKey 2, and an AutoTune Unlimited subscription. Antares now styles the name without the hyphen in its current marketing, while the manuals still carry the hyphenated form.
Calling every pitch corrector an Auto-Tune is the same move as calling every vacuum a Hoover. It is understood, and it is wrong in the places where being wrong costs you. Melodyne belongs to Celemony, VariAudio is Steinberg's, and Flex Pitch is Apple's. These are not rebadged versions of one engine, and they do not behave the same way on the same take.
Two practical consequences follow. First, a setting someone recommends online may not exist in your plugin under that name, because the naming is per product and it changes between versions. Second, the tools split into two families that solve different problems, which is the distinction the rest of this guide is built on.
How Pitch Correction Works, Step by Step
Strip away the branding and the process is four steps.
1. Detect the pitch. The software tracks the fundamental frequency of the vocal as it moves, frame by frame. This is why pitch correction wants a clean monophonic source. One voice, minimal bleed, no reverb printed to the track.
2. Pick a target. You set a key and a scale. That choice defines which notes count as legal destinations. Everything sung between those notes is treated as something to be moved.
3. Move the pitch. The detected pitch gets shifted toward the target note. How fast that move happens is the whole ballgame, and it is a control you set.
4. Preserve the voice. Formant controls keep the timbre from shifting with the pitch, which is what stops a corrected vocal sounding like a chipmunk.
Step two is where most bad tuning starts. If the scale is wrong, the plugin will confidently drag a correct note to an incorrect one, and it will repeat that every time the note appears.
The parallel with quantization is exact. Grid quantizing fixes timing against a grid you chose, and pitch correction fixes pitch against a scale you chose. Choose the reference badly and the tool makes the problem worse, faster.
Retune Speed: The Control Behind the Hard-Tuned Sound
Antares defines retune speed in milliseconds, as how quickly the correction travels from where the singer performed to the target pitch. At zero, that move is instant, so each pitch snaps to the nearest scale tone as soon as the detector resolves it. The stepping you hear on a hard-tuned record is not a distortion effect bolted on top. It is the sound of a pitch moving with no transition time.
Two of the neighbouring controls decide how much of the performance survives. Antares describes Flex-Tune as letting notes that are already close to correct pass through with less correction, so slides, bends, and vibrato stay intact while genuinely wrong notes get pulled. Humanize is narrower: it relaxes correction on sustained notes specifically, so a long held note keeps its natural drift rather than sitting frozen.
Antares publishes settings for both destinations, and they are further apart than most people assume.
Control | What it does | Transparent correction | Hard-tuned effect |
|---|---|---|---|
Retune Speed | How fast correction moves from the performed pitch to the target | 20 to 80 ms | 0 to 10 ms |
Flex-Tune | Lets notes already near the target pass with less correction | On | Off |
Humanize | Relaxes correction on sustained notes to protect vibrato | Moderate | Zero |
Key and Scale | Defines which notes the corrector is allowed to aim at | The song's key | A tighter scale, so the steps land harder |
The first three rows are Antares' own current guidance, from its Auto-Tune Pro 11 documentation and its pitch-correction guide, read in August 2026. The fourth row is craft, not vendor instruction. Naming is version-specific, so check the control list in your own plugin rather than assuming a tutorial from three versions ago still applies.
Where the Sound Came From, and What Gets Misreported
Cher's "Believe" is where the hard-tuned vocal reached the charts, in 1998. The production story attached to it is misreported constantly, and the reason is documented rather than speculative.
Sound On Sound ran an interview with producer Mark Taylor in February 1999, written by Sue Sillitoe, in which the vocal effect was credited to a Digitech Talker, a then-new vocoder-style pedal. That article now carries a footnote from the magazine's own editors, who say the producers appear to have been protecting a trade secret and were willing to let the pedal take the credit. Taylor later described the technique differently, telling The South Bank Show he had been "playing around with the Auto-Tune," a quote MusicRadar revisited in October 2023.
Be precise about what that does and does not settle. Auto-Tune producing the signature stepped artifact is supported by Taylor himself. The footnote's reading of the producers' motive is the magazine's inference, not a confession, and the 1999 account was detailed enough about its vocoder rig that no source rules out a vocoder being in the chain too.
What is safe to say is that a pedal was credited for something Auto-Tune did. The frequently repeated detail that retune speed was set to zero on the session is not sourced to either producer in anything on the record.
The software existed well before the record, which is worth stating because the timeline gets doubted. Sound On Sound reviewed the original Antares Auto-Tune in August 1997, as a Pro Tools plugin, more than a year before "Believe" charted.
T-Pain is the second landmark; he also kept the technique secret for a time, but later described publicly how he found and used it. He arrived in 2005 with "I'm Sprung," and told Complex in 2014 that he chased the sound after hearing it on a Jennifer Lopez remix, then worked through every plugin and preset he could find until he located it. He has consistently placed himself in the lineage of Roger Troutman's talk box rather than claiming an invention. Keep that lineage straight: a talk box and a vocoder are different machines, and neither corrects pitch.
The trap era turned the setting into a working default rather than a special effect. Engineers interviewed by Sound On Sound in 2017 described tracking artists through Auto-Tune as standard practice, with the vocalist hearing the corrected version in their headphones as they perform. Kesha Lee, engineering for Lil Uzi Vert, described the retune speed drifting down from around 20 to somewhere between 12 and 5 as the artist asked for more of the effect. Seth Firkins, who engineered Future, said he switched it on by default because it was the sound, and printed it rather than leaving it as a plugin decision for later.
Autotune, Graph Editing, and Melodyne Are Three Different Jobs
Melodyne is a pitch-correction and note-editing product, but its user-directed note workflow differs from Auto-Tune's real-time Auto Mode. The tools divide by whether they retune in real time or hand you an editable representation of the recording.
Approach | How it works | Real time | Best for |
|---|---|---|---|
Auto mode correction | Detects pitch and moves it to the scale at the retune speed you set | Yes | Tracking, live use, the hard-tuned effect |
Graph or pitch-curve editing | You edit the detected pitch line by hand, note by note | No | Fixing a few problem notes without touching the rest |
Melodyne | Represents detected audio as editable notes; Melodyne editor and studio add DNA editing of individual notes within polyphonic material | No | Detailed vocal work, and polyphonic material in editor and studio |
Celemony's distinguishing technology is DNA Direct Note Access, available in Melodyne editor and studio, which identifies and edits individual notes inside polyphonic recordings, including notes within a chord on a piano or guitar part. No retune-speed control is involved, because nothing is being snapped on the fly. You are editing detected notes and committing the result, usually through ARA integration with your DAW. Antares lists ARA2 support on AutoTune Pro 11 as well, which is why the two can sit in similar places in a session and still be doing unrelated work.
The short version for a session: reach for real-time correction when you want a sound, or a fast fix across a whole take. Reach for note editing when you want four specific words repaired and the rest of the performance untouched.
Choosing Settings That Fit the Song
Most tuning problems are decisions, not settings. Five steps, in order.
1. Fix the take before the plugin. Correction quality tracks input quality, and a badly captured vocal gives the detector less to work with. Recording vocals properly at home removes more tuning work than any preset.
2. Set key and scale first. The scale defines every target the plugin can aim at. Getting this wrong is the single most common cause of a vocal that sounds tuned and still sounds wrong.
3. Pick correction or effect, and commit. Transparent tuning and the hard-tuned sound live at opposite ends of the retune speed range. Settings halfway between them tend to sound like a mistake rather than a choice.
4. Tune before you mix. Pitch correction changes the harmonic material the rest of your chain reacts to. Doing it after you have set compression and EQ means redoing that work, which is part of why mixing and mastering sit in that order.
5. Keep the untuned take. Print the tuned vocal to a new track and archive the original. Taste changes, hard-tuned production dates faster than transparent production, and the raw take is the only version you cannot recreate.
None of this makes tuning dishonest. Correction is a production tool, the same as compression, and artists who treat it as one get better results than those who avoid it on principle and release a flat vocal instead.
Frequently Asked Questions
What retune speed gives the T-Pain effect?
Antares puts the hard-tuned sound at a retune speed of 0 to 10 milliseconds, with Flex-Tune off and Humanize at zero. Slower retune speeds make transitions more gradual and can reduce the stepped quality; the result also depends on the performance, key and scale, Flex-Tune, and Humanize.
Is Melodyne the same as Auto-Tune?
No. Auto-Tune corrects pitch toward a scale in real time using retune speed. Melodyne represents recordings as editable notes; Melodyne editor and studio can also edit individual notes inside polyphonic chords.
Did Cher's "Believe" really use Auto-Tune?
Yes. The 1999 Sound On Sound interview credited a Digitech Talker, but the magazine later footnoted that credit as apparent trade-secret protection, and producer Mark Taylor afterwards described using Auto-Tune himself.
Do I need to buy Auto-Tune to correct pitch?
No. Auto-Tune is a separate Antares purchase, and most DAWs ship some form of pitch correction or pitch editing. Check what your DAW edition already includes first.
Read Next
Once the vocal is tuned and the track is finished, the next problem is getting it out. How to Plan a Music Release is the step-by-step version of that, working backward from your release date.