The Winning Pitch

I used to be a cynic on the subject of pitch correction; I thought that those who needed it shouldn't be singing. I also worried that it was increasingly

I used to be a cynic on the subject of pitch correction; I thought that those who needed it shouldn't be singing. I also worried that it was increasingly being used as a tool to further the careers of Britney Spears knockoffs who can dance and look great in slinky outfits but couldn't sing if their lives depended on it.

Well, I've changed my tune. I now believe that pitch correction is a perfectly valid and important studio weapon, and one worth having in every arsenal. In fact, pitch correction has saved my tail in the studio many times. That said, it should be used sparingly; you should always strive to obtain a performance on tape that needs no correction.

Although pitch is a critical factor when recording vocals, it's the emotional qualities of a vocal performance that make a track really sing. Sometimes the feel of a take is perfect, but the intonation is a little shaky. When that happens, your options are to have the singer try again, in hopes of duplicating the feel with better intonation, or to try to fix the pitch problems on that initial take with a little electronic magic. In the past I would lean toward the retake route, which sometimes resulted in more harm to the vocal track than would have occurred from the sonic degradation of pitch correction. (I have yet to hear pitch correction that doesn't compromise fidelity, which has always been the main reason I have avoided using it.)

A few years ago, I engineered an album in which the singers had a tough time singing in tune. In an effort to get the notes where they needed to be, we did countless takes and punches. In hindsight, I wish that I had kept the first-take energy of their performances and used the miracles of modern science to do the rest. Nonetheless, I learned my lesson and have come to appreciate pitch correction, and I've even become pretty adept at using it.


Whether you're running a project studio, a production house, or a basement special, sooner or later you'll have need for pitch correction — especially if you record projects in which time is of the essence. Just to be clear, in this column I am discussing the use of intelligent pitch correction as opposed to fixed pitch shifting. The latter is simply adjusting an entire track by raising or lowering the pitch by a fixed amount (every note will be affected equally).

Fixed pitch correction is useful if an instrument is slightly but consistently out of tune. Consistent flatness or sharpness can also be dealt with by adjusting the tape speed while recording (when working with analog technology). For example, you record tracks at one studio and the tape machine's speed control is a hair off, and you go to another studio to overdub the piano and it's out of tune because of the different speed of the second machine. At that point you can either pitch-shift the piano track or varispeed the machine to get the tracks and the piano matched up.

Intelligent pitch correction, on the other hand, is a much more complicated technology. A box or plug-in will continuously sample every note, determine the actual pitch, and if it doesn't match a prescribed note, it will change the pitch. This requires the box to know what pitch the singer meant to hit, and this usually means you need to tell the pitch corrector a thing or two.

In terms of devices to choose from, there are quite a few options on the market these days. There are several outboard models ranging in price from home-studio affordable to “if you have to ask.” In addition, there are a handful of plug-ins available for computer-based recording systems. I have used Antares Auto-Tune (probably the best-known pitch-correction product) for most of my correction. I have yet to try the latest TC-Helicon boxes, but by all accounts they sound great and do some amazing things in addition to correcting pitch. Regardless, when purchasing your system, find the product that pleases your ears as well as your wallet.


The first thing you need to decide when using pitch correction is whether to use it as you're recording or during mixdown. The former can be immediately pleasing because you hear things in tune right off the bat. But I never use pitch correction during the recording process, and I doubt I ever will. There are several reasons why.

First, you often need very little of it. Sometimes a vocal track might have only one or two notes out of tune. In that case, why subject an entire vocal to a process that could harm the sound quality and also introduce artifacts — glitches introduced by improper settings, or a loss of fidelity from mediocre A/D/A. I find it better to spot-correct by running the offending passages through a pitch corrector and recording onto another track (or automating the process, but more on that later). Then I can comp the tracks or punch in the phrases by recording over the original bits with the pitch-corrected version from the other track. That will preserve the integrity of the whole track and minimize the loss in sound quality.

The second reason that I wait until mixdown to use pitch correction is that it's not an exact science. No matter how sophisticated your pitch corrector is, and no matter how carefully you use it, it will not always achieve perfect results. It can miss notes, correct too slowly, correct too quickly, or even correct to a wrong note. Then you'll have to re-record or use pitch correction again to fix those mistakes. Because of the complex balance of parameters involved, it's a far safer and smarter approach to use it after the fact, instead of as tracks are going down.

The third reason is courtesy. It's much better to give yourself the luxury of tweaking the settings while mixing rather than annoying the singer (and possibly distracting his or her concentration and ability to perform) by trying to adjust pitch-correction parameters during the vocal session. In addition, there are times that you don't want the singer to know you're using pitch correction. It can be a distraction for singers to hear the box doing its thing, because they will be hearing a clash between the note they are singing and the corrected note they are hearing in the headphones. In addition, some egos don't handle it well when you tell them you have to use a pitch-correction device. All the more reason to do it after the singers have left the studio.

One important rule as an engineer is to keep confidences. I make a point never to mention the names of the artists whose vocals I've pitch-corrected. Anyone who was not in the room when the technology was used will not know about it.


Whether you're using a plug-in on a computer or a hardware-based pitch corrector, there are a few parameters you'll need to tweak. The first is the scale and the key. The simplest way to go is to set the corrector to a chromatic scale. That means the unit will shift any wrong note that triggers it (based on a user-selectable threshold — more on this in a minute) to the nearest half step. Most singers don't miss by much, so a nudge to the nearest note will usually do the trick. However, this can actually create problems if the singer's pitch is particularly bad.

On a recent project that I engineered, the singer was going sharp on a regular basis. He was so sharp that on a chromatic setting the pitch corrector would often correct to a half step above the intended note (he was singing more than 50 cents sharp). The result was a perfectly intonated wrong note (see Fig. 1). In that case, you have to set the device to the key and scale of the melody.

If the song is in E major, and the entire vocal conforms to that scale, set the unit to “E Maj” (see Fig. 2). That means that any misplaced note will be shifted to the closest note within that key — pretty smart. The more sophisticated the pitch-correction device, the more scale options you will have, as well as the option to create your own scales or to exempt notes from correction. You can even set most units to detect and correct a particular pitch if a singer consistently misses only that note. Obviously, it is crucial to know the key of the song, and it's helpful to know the melody so that you can set up the unit or the plug-in effectively.

On the aforementioned project, I mixed all the analog tracks down to two stereo tracks, kept the vocals on a separate track, and ran it all into Digi-design Pro Tools. I then used the Antares Auto-Tune plug-in on the vocals; I preset it to the appropriate key for each song and then modified the scale to fit the melody. That way, there was no chance of the pitch corrector spitting out a wrong note.


In addition to setting the scale, pitch correctors have various parameter controls that determine how exact the automatic pitch-correction effect will be. Finding the optimal settings for a given situation can be tricky and can frequently require trial and error.

The terminology and functions differ from product to product. For example, the main determinant of correction on Antares's Auto-Tune plug-in is the Retune Speed parameter. On the TC-Helicon VoiceOne 2.0, the key parameter is called Window; it adjusts how choosy the correction will be.

You'll find plenty of occasions when you don't want to set your device or plug-in for maximum correction. One reason is that singers often intentionally slide into a note by approaching it just from below (aka scooping). In other cases, being just slightly off pitch might be a cool sound. But if the pitch corrector is adjusted too high or too fast, it will treat any pitch that's the slightest bit off the grid as one that needs fixing. That could end up canceling out those intentional microtonal effects and making the vocal sound unnatural. (Many people now use these types of extreme settings as deliberate effects thanks to singers like Cher.)

Whereas overly high settings can sound unnatural, settings that are too low or too slow can cause off-key notes to sail by uncorrected. I start out with the settings in the middle and work from there. Getting it right takes precision listening and subtle tweaking. I never want the effect to be heard, so as soon as I hear a note bending too quickly or sounding at all like an effect, I back off the settings ever so slightly. Depending on the singer, it can really be a compromise. Sometimes it's not possible to cleanly correct every bad note using automatic settings, and you have to decide which you prefer — a note that's a little off key or a note that sounds slightly artificial.

If you're just starting out using pitch correction, I recommend you spend some time playing with the parameters on your unit to get a sense for how they interact. Run each control through its range and listen to hear what happens. You will gradually find what works best. I still end up with settings in the central range of the parameters because I am always shooting for a natural, transparent sound. Your demands may vary.


One additional factor to contend with is vibrato. A forcefully pitch-corrected note will inherently be stripped of any natural vibrato (because vibrato is a rhythmic pitch fluctuation, any pitch correction device will want to correct the wavering and smooth out the pitch of the note), and sometimes that's a bad thing. So, most devices will allow you to dial some (artificially produced) vibrato back in. Standard parameters include delay, speed, and depth.

Again, be careful because it can sound very mechanical since few people have a rhythmically perfect vibrato. If I dial any in, I use a shallow depth and adjust the speed to fit with the tempo of the tune. If you want vibrato that is more accurate sounding, you might consider the TC-Helicon VoiceOne 2.0 (see Fig. 3), which offers vibrato based on models of a variety of actual human voices.


Some pitch correctors offer graphic or manual modes. Rather than setting the parameters as discussed above and letting the pitch corrector loose in hopes that it will do what you want it to, you can take matters into your own hands and remove the element of chance.

In Antares Auto-Tune, Graphical mode tracks the notes in a selected region and then draws them on a grid (see Fig. 4). You can see the notes and how close they are to the correct pitch. You can manually redraw any notes that missed the mark, and then adjust the speed setting to control how quickly the note will go from the original pitch to your drawn pitch.

This mode of operation is more time-consuming but ultimately more precise, because you're telling the processor exactly what to do. It takes a little practice to draw with the mouse, but anyone adept at freehand computer graphics will have no problems.

Pitch correctors can also be controlled through MIDI. Devices typically give you control of functions ranging from basic automation (triggering the device in and out) to more advanced melody setting. Some pitch correctors can alter the vocal using a melody you play on a keyboard either to define a scale or to correct pitch in real time based on the MIDI notes played.


All of the above parameters and approaches can be used for any instrument, not just vocals. Some devices have an input setting that is designed to help the sampling software adapt to the input signal. For example, there might be presets for high voice, low voice, instrument, as well as other sources. If a saxophone player lays down a track but goes sharp here and there, pitch correction will fix things up just as it will on a vocal. And you can store all of your settings, so if you routinely record the same musicians and instruments, you are able to keep the settings you have already created that work so magically.

Again, I always caution people to use pitch-correction devices or plug-ins only when necessary. There is no question that they are invaluable in production studios when time, rather than sound quality, is the driving dimension, and they are often a lifesaver in the professional studio. Nevertheless, I'll take an off-key, emotionally charged vocal by Robert Plant over a technically correct, but soulless Britney vocal any day. Of course, with the proper use of your pitch corrector, you can have your emotion and your intonation, too.

Sean D. Carberry,a freelance producer in Boston who loves to complain about new technology, can no longer live without hard-disk recording and intelligent pitch correction.


Fix it in the mix: When recording to multitrack, it's best to save the pitch correction for the mixing stage.

Keep it natural: Be careful not to use settings so high that they eliminate microtonal effects such as scoops and vibrato. Unless you're going for an artificial effect, make sure that the settings are subtle enough to leave the vocal sounding natural and unaffected.

Correct only where it is absolutely needed: If there are only a couple of bad notes on an entire vocal track, selectively pitch-correct those rather than running the entire track through. Otherwise, you could cause more problems, both in pitch and fidelity.

Don't make it worse: Be careful that far-out-of-tune notes don't get corrected to the wrong note.

Watch what you say: If the vocalist has a sensitive ego, it's better not to tell him or her that you're going to use pitch correction. And never tell those who weren't at the session that you used pitch correction on a particular artist.

Get graphic: For the most precise control, use a corrector that features a “graphic” or “manual” mode.


Antares Audio Technologies
tel. (888) 332-2636 or (831) 461-7800
e-mail Web

tel. (805) 373-1828