How to use Auto-Tune as a Pro

Starting points, not rules. Retune Speed is in milliseconds. The ranges follow Antares’ own guidance: 0 ms is instant, 25–50 ms reads as natural support, 80 ms and above barely touches a typical vocal. Flex-Tune and Humanize are 0–100 controls, and these are the values we start from before listening. Set the key and scale first: with the wrong key, no setting helps.

What is Auto-Tune?

Auto-Tune is real-time pitch correction software made by Antares Audio Technologies. Dr. Andy Hildebrand developed it, and Antares released it in 1997. The plugin detects the pitch of a voice, or of an instrument playing one note at a time, and shifts it toward the nearest note allowed by the key and scale you set. How fast it does that decides whether you hear it.

The word also means a sound. When people say a vocal has “auto tune” on it, they usually mean the hard-tuned effect: notes that jump from pitch to pitch with no glide in between. Cher’s “Believe” (1998) made that sound famous, and T-Pain made it the signature of a wave of hip-hop and R&B in the late 2000s. Most tuning on modern records is the opposite, gentle enough that you can’t point to it.

Antares now writes the name as one word, AutoTune. The current editions on antarestech.com are:

• AutoTune (formerly AutoTune 2026): streamlined automatic tuning with Retune Speed, Humanize, Flex Tune, and Classic and Modern modes.
• AutoTune Advanced: automatic tuning plus Natural Vibrato and a Live low-latency mode for tracking and stage.
• AutoTune Pro 11: the full set, including Graph Mode for note-by-note editing, a 4-part Harmony Player and ARA2 support.

All three load as VST3, AU or AAX plugins, so they run in any major DAW.

How to use Auto-Tune in any DAW

1. Insert it first on the vocal track. Put Auto-Tune before compression, EQ and effects, so it reads the raw pitch. Use one instance per vocal track. It follows one note at a time, so it won’t work on chords or on a bus of stacked harmonies.
2. Set the vocal range (Input Type in older versions): Soprano, Alto/Tenor or a low voice setting. Choose by the singer’s register, not by genre.
3. Set the key and scale to the song’s key. If you’re not sure, AutoKey 2 on the master bus detects it and syncs every AutoTune instance in the session.
4. Choose the mode. Modern for natural correction, with Flex-Tune and Humanize available. Classic for the old AutoTune 5 character. Classic disables Flex-Tune, and in the base AutoTune it disables Humanize too.
5. Set Retune Speed from the table above, then Flex-Tune, then Humanize.
6. Tracking or mixing? Use the low-latency setting while recording, so the singer hears no delay. Switch to HQ (High Quality) for the mix, where your edition has it.
7. Listen with the beat, then solo. Artifacts that sound obvious solo can vanish in the mix, and the other way round.

Auto-Tune settings explained

The screenshots below are from an earlier Auto-Tune Pro. In Pro 11 and the newer editions the layout is different and a few names have changed, but the controls do the same job.

1. Input type (vocal range)

This tells the pitch detector what range to expect. Choose Soprano for a high voice, Alto/Tenor for a medium male or female voice, and the low setting for a deep male voice. There are also instrument options. Whatever the source, it has to play one note at a time.

2. Key and scale

Set the key of the song, so Auto-Tune only pulls notes to pitches that belong in it. This one setting prevents most tuning mistakes.

If you can’t work out the key, AutoKey 2 (sold by Antares) listens to the mix or the vocal and sets it for you. It isn’t always right, so check the result by ear. Chromatic lets Auto-Tune pull to any semitone. Use it only when the melody really uses notes outside the key, because a flat note can then land on the wrong neighbour. You can also build the scale note by note, see step 12.

3. Auto or Graph mode

Auto mode corrects in real time as the track plays. Graph mode, in AutoTune Pro 11, shows each note on a pitch graph so you can move, reshape or leave alone individual notes, much like Melodyne. We compare the two approaches in Melodyne vs Auto-Tune. Graph mode is slow, detailed work, and it’s the part of the job our engineers do by hand in manual vocal editing. This guide covers Auto mode.

4. Formant and Throat

Formant correction keeps the character of the voice when the pitch moves a long way, so a big correction or transposition doesn’t sound like a chipmunk. Throat changes the modelled length of the vocal tract, which makes the voice sound bigger or smaller at the same pitch. For ordinary tuning, leave Throat at its default. It’s a sound-design control.

5. Transpose

Shifts the output up or down in semitones. Leave it at zero for tuning work. If a part needs a different key, a new recording almost always sounds better than a big transposition.

6. Detune

Sets the reference pitch. The standard is A = 440 Hz, and you’ll rarely change it. If the beat or the track was tuned to something else, for example 432 Hz, match Detune to it. Otherwise Auto-Tune corrects every note slightly away from the music.

7. Tracking

Tracking sets how strict the pitch detection is. The default suits most vocals. If Auto-Tune glitches on breaths, or misses notes in a noisy or very breathy take, move it a little and listen again. Not every edition shows this control.

8. Retune Speed

This is the main control. It sets how quickly the pitch is pulled to the target note, in milliseconds.

At 0–5 ms the pitch snaps to each note instantly and jumps between notes without gliding. That’s the Cher and T-Pain effect. Around 25–50 ms the correction follows the singer and nudges each note toward center, which sounds natural. Slower than about 80 ms, it barely touches a typical vocal and may miss short notes altogether.

9. Flex-Tune

At zero, Auto-Tune corrects every moment of the vocal to the nearest note. Above zero, it leaves alone pitch movement that is far from the target note and only corrects notes that are already close to it. In practice, scoops, slides and expressive bends pass through, and the held part of each note still gets tuned.

Raise Flex-Tune before you slow Retune Speed down. It’s also essential on guitar or any part with deliberate slides. Flex-Tune works in Modern mode only.

10. Natural Vibrato

Positive values deepen the vibrato that’s already in the performance, negative values flatten it. Leave it at zero unless the vibrato really needs help.

11. Humanize

Humanize relaxes the correction on sustained notes only. Short notes stay tight at your Retune Speed, and long notes get a slower retune so they keep some natural drift. Keep it at zero while you find the Retune Speed, then raise it if held notes sound frozen.

12. Create Vibrato

These controls add vibrato that wasn’t in the performance. Shape sets the waveform, Rate the speed, Onset Delay the time from the start of the note to the vibrato, Onset Rate how long it takes to reach full depth, and Variation adds randomness so it sounds less mechanical. Pitch, Amplitude and Formant Amount set the depth of each part. It’s an effect, not a correction tool.

13. Edit the scale note by note

On the on-screen keyboard you can change the scale one note at a time with two switches.

Bypass leaves a note uncorrected: anything the singer sings near it passes through untouched. Remove takes a note out of the scale, so pitches near it are tuned to the next allowed note instead.

Use them to:

• Fix only a few notes. Bypass every note except the two or three that are out of tune.
• Add a note outside the key. If the melody uses a borrowed note, put it back in the scale.
• Stop a wrong pull. Remove a note that keeps catching a pitch meant for its neighbour.
• Build an effect scale. Keep a few notes far apart for a more dramatic hard-tuned jump.

Auto-Tune tips from our sessions

Comp first, tune second. Build the best take from the best phrases, then tune it. If a whole take needs heavy correction everywhere, another vocal pass will sound better than any setting.

For the hard effect, let the singer hear it while recording. The effect changes how people sing into it, and the take should be sung for the effect. Leaving one headphone off helps the singer hear their real pitch too, so they don’t drift further off as the song goes on.

Tune each vocal track separately. Lead, doubles and harmonies each get their own instance and their own range setting. Doubles tuned with exactly the same settings can sound too perfect next to each other. A little difference keeps them wide.

Check the key in every section. If the song changes key, automate the key or split the vocal across tracks with separate instances.

How Auto-Tune changed music

Auto-Tune started as a correction tool that nobody was meant to hear. After “Believe” it also became an effect, and from the late 2000s hip-hop and R&B artists built whole sounds around it. Today it’s both: a hidden finishing tool on many vocals, and an instrument in its own right in rap, pop and afrobeats.

More on vocals: Melodyne vs Auto-Tune, Auto-Tune in Logic Pro and how to mix a vocal.

Basic pitch correction is part of every mix we do. For note-by-note pitch and timing work by hand, add Vocal Rescue ($100 per song) to a mixing order, or order manual vocal editing on its own from $57 per song (up to 4 vocal tracks). Mixing and mastering starts at $133 per song, 2–3 days per song, 6 revision rounds included.