top of page

Vocal Recording: Pitch and Timing Correction

Updated: Apr 11, 2023

In my experience, most talented vocalists desire to maintain the most natural elements of their performance, the human factor.

This isn’t to say Pitch and Timing Correction are not desired. But to autotune a vocal is usually not even in the cards unless the vocalist calls for it. Similarly, fixing every single note to the grid can also lead to the same clinical result.

But where pitch and timing variance can really reveal themselves is when you double a lead track or add harmonies. And if the singer is not spot-on, now you need some help!

PITCH CORRECTION IS AN ART

I use Melodyne, and from my experience, I have yet to see any problem with their algorithms. The detection is usually on the money! Each word is represented like clouds, I call them blobs. Within each blob is a line that represents the pitch. Variations in pitch are easy to identify like slides or vibrato because they appear like a heartbeat reading within each blob.

Upon importing a region into Melodyne, you will also notice the blobs aren’t necessarily represented tightly on each note row. Instead, they can sometimes be 10-50% high or low. This is the “average” note of each word. So if I am a little flat, my D may be 25 points lower than the prime D.

If you exceed greater than 50%, you end—up at the next higher or lower note on the chromatic scale. So if I sing a D and it is sounding to Melodyne like a sharp C#, I may see the blob appear on the C# row, albeit sharp as well.

So if I take a vocal performance and simply snap everything to the grid, that sharp C# may end-up a perfect C# instead of the D I actually thought I was singing. Long story short, you rarely can snap a performance to the grid and expect perfect results. This is where the art of pitch correction comes into play.

NOTE BY NOTE

Typically I will take a vocal performance and snap everything to the grid first to see how it sounds. This won’t ruin your track, but it’s a great way to identify weaknesses. Once snapped, I will check the key against the notes identified. Generally speaking the notes should all be in key. If not, this is the first place I look for anomalies.

When I pitch correct a vocal, I will usually not solo it. Instead, I will keep at least all of the melodic instruments in the mix so I can tell if a note is standing-out. Off key notes are not necessarily a bad thing in the context of a song, but since I snapped everything, I may have created some problems. And these should stand-out easily.

Then I listen back to each section on a loop. This allows me to absorb the section and consider the pitch changes that resulted from me snapping everything to the grid. If something sounds off or too clinical, I will undo the pitch changes for a word or all words, and start to examine each, note by note.

I can move blobs so they sound more natural. For example, this blob slides between notes. If I snap it to the grid, the slide may not be in tune. Instead, I will pitch the blob up or down slightly to keep the natural feel of the performance and balance the pitch between the notes instead of sacrificing one note for a perfect other.

Generally speaking, if you are doubling or adding backing tracks, you are going to need a solid baseline, snapping everything to the grid is usually a good place to start.

For Choruses especially, I will usually ensure there is little variance in pitch between the primary and accompanying tracks. Just like a guitar that is out of tune, a single string or backing track that is flat or sharp will cause havoc.

EDITING TIMING

Before I discovered VocAlign, I would literally spend hours aligning doubles or backing tracks to the primary vocal. If you’ve tried this yourself, you will quickly learn it is an unforgiving exercise. And simply put, there are few tools like VocAlign that actually work. The $150 I paid saves me about $150 per track!

But again, the only reason for using VocAlign is to align tracks. You would not use it on a single track, there is nothing to align it to. But if you are dealing with harmonies, doubles, or gang vocals, nothing beats it.

That said, Melodyne is also a very powerful tool for making minor timing changes to a vocal. Which is where I usually start on the primary track.

Say the vocalist comes-in a few milliseconds early on a word, or late. This kind of problem is super easy to fix. You can stretch whole words or simply move the word to its appropriate position.

You can also modify the emphasis on a word by moving it forward or backward on the grid. This is useful if a slide occurs too soon or a sustained vowel starts too late.

LIMITATIONS OF TOOLS

But don’t expect to make giant changes to a vocal performance with any tool. Stretching a word too far and you will start to hear phase issues.

Similarly, I have found simply copying a performance and pitching it up or down an octave or even a fifth or third, in an effort to manufacture harmonies, results in either a Chipmunk voice when pitched-up or a really unrealistic low pitched copy.

You can make harmonies from a single track, but you will probably want to modify the formants up or down to account for the pitch adjustment. A higher harmony will likely benefit from lowering the formants, while a lower one would need higher formants. This offsets the “manufactured” sound dramatically in some cases.

WORKFLOW

I try not to touch any vocal editing until the mixing stage is ready to begin. We will typically track everything and get it close, and then when all of the content is ready, we start editing.

I will first start with my primary vocal. Get it in shape and then use it as a guide for the rest of the doubles and harmonies. I do all of this work with Melodyne.

Once the primary vocal is clean, I will VocAlign everything to it. This saves me from using Melodyne for timing adjustments. VocAlign provides the ability for tight or loose alignment; in most cases, a tighter alignment works well.

Then I will treat each accompanying track with Melodyne. And solo just the Primary track with it. I do this for every track. Then I will solo all the doubles or harmonies with the primary track and listen for phase anomalies. Many times you will find phase issues with vibrato, which is why I will always encourage the vocalist to limit vibrato to only the Primary track. Competing vibratos will turn a good double or harmony into garbage really quick.

If I encounter competing vibrato, I can simply dial it out with Melodyne. Be careful here though, squashing vibrato too much may result in artifacts that will be obvious to the listener. I usually reduce vibrato between 25-50% and that seems to take care of anything that pops-out.

Just remember, at the end of your editing sessions, ALWAYS bounce your track and remove the Melodyne plugin. I have found leaving Melodyne running for too long, it will start to change the timing of my tracks. I have even seen it revert my changes. Melodyne is a temporary tool. It was never meant to run long-term on a track.

Comments


bottom of page