Vocal Production and Perfection
By Giles Reaves
So you finally finished recording all your vocal tracks, but unfortunately you didn’t get one take that was perfect all the way through. You’re also wondering what to do about some excessive sibilance, a few popped “P”s, more than a few pitchy lines and some words that are all but too soft to even be heard – don’t worry, there’s hope! And hey, welcome to the world of vocal editing.
A Little History…
Since the beginning of musical performance, singers (and instrumentalists) have craved the possibility of re-singing that one “if only” note or line. You know the one: “if only I had hit that pitch, if only I had held that note out long enough, if only my voice hadn’t cracked”, etc. With the advent of early recording technologies, these ‘if only’ moments were now being captured, and performers were forced to face reliving those ‘if only’ moments forever! One ‘if only’ moment could ruin an entire take.
With the popularity of analog tape recording in the mid 20th century also comes the popularity of splice editing. Now you can record the same song two different times, and choose the first half of one take and the second half of another. Next comes multi track recording, where you don’t even have to sing the vocal with the band!
Multi track recording introduced punching in and out, which allowed re-recording of just the “if only” moments on an individual track. But more importantly as it relates to the subject at hand, multi-track recording also introduced the idea of recording more than one pass or ‘take’ of a lead vocal, leading to what is now known as “vocal comping”. More on that in just a bit.
But before we get into the nitty-gritty, here’s a brief outline of the typical vocal editing process for lead and background vocals. Of course, much of this is subject to change according to production direction, or the vocalist’s skills and past experience.
- Recording: This ranges from getting the first take, to punching in on a take, to recording multiple takes for comping.
- Comping: Combining various takes into one final track, tweaking edits to fit, crossfading if needed.
- Basic Cleaning: Listen in solo one time through. Typical tasks include removing the obvious things like talking, coughing, mouth ‘noises’ etc., checking all edits/crossfades, fading in/out where necessary.
- Performance Correction: Timing and pitch correction takes place after you have a solid final comp track to work with.
- Final Prep: this includes everything from basic compression/EQ, to de-essing, reducing breaths, filtering out Low Frequencies, etc.
- Leveling: During the final mix, automating the vocal level (if needed) to sit correctly in the mix throughout the song.
Note that at many stages along the way you will be generating a new ‘master vocal’ file (while still holding on to the the original files, just in case!). For example, let’s say you record 4 vocal ‘takes’ which become the current ‘masters’. The you comp those takes together to create a new “Comp Master” vocal track, and then you tune/time the Comp Master and sometimes create a “Tuned Vocal Master” track (which is then EQ’d and compressed to within an inch of its life while simultaneously being drowned in thick, gooey FX, all before being unceremoniously dumped into what we like to call the mix).
Recording Vocals for Comping
In order to comp a vocal, you must first have multiple vocal tracks to choose from. Recording comp tracks can be slightly different from recording a ‘single take’ vocal. For one thing, you don’t have to stop when you make a mistake — in fact, many times a performer gets some great lines shortly after making a mistake!
I tend to ask for multiple ‘full top to bottom’ takes from the vocalist, to preserve the performance aspects and to help keep things from getting over-analytical. Then I use comping to work around any mistakes and ‘lesser’ takes, choosing the best take for each line. Often the vocalist will be involved with the comping choices, so be prepared to be a good diplomat (and don’t be too hard on yourself if you’re comping your own vocals)!
How many tracks?
This will be different for every singer, but for comping I generally suggest recording around three to five tracks. Any less and I don’t feel that I have enough choices when auditioning takes — any more and it becomes difficult to remember how the first one sounded by the time you’ve heard the last take.
When recording tracks that I know will be comped, I usually let the singer warm up for a few takes (while setting levels and getting a good headphone mix) until we get a ‘keeper’ take that is good enough to be called ‘take one’. From there, simply continue recording new takes until you feel you have enough material to work with. If you find yourself on take seven or eight and you’re still not even getting close, it may be time to take a break!
In Reason, when tracking vocals for future comping, you simply record each ‘take’ on the same track. With ‘tape’ recording this would erase the previous take, but with ‘non-destructive’ recording you are always keeping everything (with the newest take laying on ‘top’ of the previous take). When you enter Comp Mode, you will see each take just below the Clip Overview area (with the newest take above the older takes). The ‘takes’ order can easily be rearranged by dragging them up or down. Double-click on any ‘take’ to make it the ‘active take’ (it will appear in color and in the Clip Overview, and this is the take you will hear if you hit play). Now comes the fun part.
Vocal Takes in Comp Mode
To combine or ‘comp’ different parts of different takes together, use the Razor tool as a ‘selector’ for the best lines/words. After creating cut lines with the Razor tool, you can easily move them earlier or later by dragging the ‘Cut Handles’ left or right. You can delete any edit by deleting the Cut Handle (click on it and hit the ‘delete’ key). Create a crossfade by clicking/dragging just above the Cut Handle. Silence can be inserted by using the Razor to make a selection in the “Silence” row, located below the Clip Overview and above the Comp Rows.
Comping (short for compositing): picking and choosing the best bits from among multiple takes, and assembling them into one continuos ‘super take’.
Now that you have your vocal tracks recorded, how do you know which parts to use? I’ve approached this process differently through the years. Previously, I’d listen to each take in its entirety, making arcane notes on a lyric sheet along the way — this was how others were doing it at the time that I was learning the ropes. More recently I’ve taken another approach that makes more sense to me and seems to produce quicker, smoother, and better comps.
Currently, my auditioning/selection process consists of listening to one line at a time, quickly switching between the different takes and not stopping for discussion or comments. This is the basic technique you will see me demonstrate in our first video (see below).
Now it’s time for a little thing I like to call a Video Detour. Enjoy De-tour (a-hem). Follow along in this ‘made for internet’ production as I comp the first verse of our demo song “It’s Tool Late” (by singer/songwriter Trevor Price).
Note: watch your playback volume – the music at the top comes in soft, but it gets louder when the vocals are being auditioned.
Comping a Vocal using Reason’s “Comp Mode”
The three most common issues with vocals are pitch, timing, and level/volume. All three are easy to correct with today’s digital tools and just a little bit of knowledge on your part.
After comping, I usually move on to correcting any timing issues. You may also jump straight into dealing with any tuning issues if you prefer. Often times there isn’t a lot of timing work that needs to be done on a lead vocal. But when you start stacking background vocals (BGVs) things can get ‘messy’ very quickly. User discretion is advised.
In our next video example (it’s coming, I promise), I will show you how to line up a harmony vocal track with the lead vocal. I will use the lead vocal as the timing reference, moving the harmony track to match the lead. Since you can only see one track at at time when editing, I use the playback curser (Song Position Pointer in Reason) to ‘mark’ the lead vocal’s timing, then when I edit the harmony track using this reference point to line it up with the lead vocal.
I will also use the following editing techniques:
Trim Edit, where you simply trim either end of a selected clip to be shorter or longer as desired, which will expose or hide more or less of the original recording that is inside the clip.
Time Stretch (called Tempo Scaling in Reason), where you use a modifier key [Ctrl](Win) or [Opt](Mac) when trimming an audio clip, allowing you to stretch or shrink any clip (audio, automation, or MIDI) which changes the actual length of the audio within the clip.
Clip Sliding (my term), where (in Comp Edit mode) you use the Razor to isolate a word or phrase, and you slide just that clip right or left to align it – using this technique allows you to slide audio forward or backwards in time without leaving any gaps between the clips!
OK, thanks for waiting – here’s the video:
Possibly an entire subject in itself, as everyone has their own take on vocal tuning. Of course, it’s always best to ‘get it right the first time’ if you can. But sometimes you are forced to choose between an initial performance that is emotionally awesome (but may have a few timing or pitch flaws), and one that was worked to death (but is perfect in regards to pitch and timing). If only you could use the first take with all its emotion and energy. Well now you can!
Neptune Pitch Adjuster on the lead vocal
In Reason, using Neptune to naturally correct minor pitch issues is about as simple as it gets. The following video demonstrates using Neptune for simple pitch correction, as well as using it in a few more advanced situations.
Vocal “Rides” (as they are called for ‘riding the fader/gain’), have been common from almost the beginning of recording itself. In rare cases, you may have to actually ride the vocal while recording the vocal(!) – this is the way it was done back with ‘direct to disk’ and ‘direct to two-track’ recordings. But luckily you can now do these ‘rides’ after the vocal is recorded, or you can even draw in these moves with a mouse (with painstaking detail, if you are so inclined). Most of the time I use a combination of both techniques.
The basic idea with vocal rides is to smooth out the overall vocal level by turning up the soft parts and turning down the loud parts (in relation to the overall mix). The end game is to get the vocal to sit ‘evenly’ at every point in the song, in a way that is meaningful to you. Or as I like to say, to get the vocal to ride ON the musical wave, occasionally getting some air but never diving too far under the musical water.
Great engineers learn the song line by line and ‘perform’ precision fader moves with the sensitivity and emotion of a concert violinist. It really can be a thing of beauty to watch, in an audio-geeky sort of way. For the rest of us, just use your ears, take your time, and do your best (you’ll get better!).
There’s no right or wrong way to edit vocal levels, only a few simple rules to follow: Obviously, you don’t want to ever make an abrupt level change during a vocal (but you can have somewhat abrupt automation changes between words/lines), and you don’t want to be able to actually hear any changes that are being made. All level rides should ideally sound natural in the end.
As for techniques, there are three approaches you can take in Reason. The most familiar is probably Fader Automation, which can be recorded in real-time as you ‘ride’ the fader. You can also draw in these moves by hand if you prefer. Additionally, you can do what I call “Clip Automation”, which involves using the Razor to create new clips on any word, breath or even an “S” that is too loud or too soft. Since each separate clip has it’s own level, you simply use the Clip Level control to make your vocal ‘ride’. Alternatively, you can use the clip inspector to enter a precise numeric value, increase/decrease level gradually in a ‘fine tune’ way, or simultaneously control a selection of clips (even forcing them all to the same level if desired).
The ‘pros’ to Clip Automation are that it is fast, you can see the waveform change with level changes, you can see the change in decibels, and you can adjust multiple clips at once. The main con is that you can’t draw a curve of any sort, so each clip will be at a static level. All I know is it’s good to have options, and there’s a time and place for each technique!
Using “Clip Automation” to reduce multiple “S”s on a Vocal Track
As a ‘fader jockey’ myself, I prefer to begin vocal rides with a fader (real or on-screen). From there I’ll go into the automation track and make some tweaks, or to perform more ‘surgical’ nips and tucks (if needed) on the vocal track. It’s these smaller/shorter duration level changes that are more often ideally created with a mouse rather than a fader. Reducing the level of a breath or an “S” sound come to mind as good examples of ‘precision’ level changes that benefit from being drawn by hand.
Leveling the vocal must ultimately be done in context, which means while listening to the final mix that the vocal supposed to be is ‘sitting’ in (or ‘bed’ it is supposed to ‘lay’ on, or choose your own analogy!). This is because you are ultimately trying to adjust the vocal level so that it ‘rides’ smoothly ‘on’ the music track at all times (ok, so I’m apparently going with a railroad analogy for now), which doesn’t necessarily imply that it should sit at a static level throughout the song.
You would think that a compressor would be great at totally leveling a vocal, but it can only go so far. A compressor can and will control the level of a vocal above a certain threshold, but this doesn’t necessarily translate into a vocal that will sit evenly throughout a dynamic mix. Speaking of compression, this is probably a good time to mention that all processing (especially dynamics) should be in place before beginning the vocal riding process, as changing any of these can change the overall vocal level (as well as the level of some lines in relation to others). Bottom line – do your final vocal rides (IF needed) last in the mixing process.
Let’s begin – set your monitors to a moderate level and prepare to focus on the vocal in the mix. Oftentimes I prefer smaller monitors or even mono monitoring for performing vocal rides – you gotta get into the vocal ‘vibe’ however you can.
Things to look for:
Before you get into any actual detail work, listen to the overall vocal level in the mix throughout the entire song. Sometimes you will have a first verse where the vocal may actually be too loud, or a final chorus that totally swallows up the vocal. Fix these ‘big picture’ issues first before moving on to riding individual lines and words.
When actually recording the fader moves (as in the video), I’ll push the fader up or down for a certain word and then I will want it to quickly jump back to the original level. In the “Levels” video, you will see me hit ‘Stop’ to get the fader level to jump back to where it was before punching in. The reason why I’m doing it this way is that if you simply punch out (without stopping) the fader won’t return to it’s original level (even though it’s recording it correctly). Long story short, it’s the quickest way I found to create my desired workflow, and it works for me (although it may look a bit weird at first)!
Often times you will find that it is the last word or two in a line that will need to be ridden up in level (sometimes the singer has run low on air by the end of a line). Also watch for the lowest notes in a vocal melody – low notes require more ‘air’ to make them as loud as the higher notes, so they can tend to be the quieter notes in a vocal track. Another thing to listen for are any louder instruments that may ‘mask’ the vocal at any time – sometimes the fix is to raise the vocal, other times you can get better results by lowering the conflicting instrument’s level momentarily. In extreme cases, a combination of both may be required!
Other problems that rear their heads from time to time are sibilance, plosives, and other ‘mouth noises’. These can all be addressed by using creative level automation, or by using a device more specifically for each issue – a ‘de-esser’ for sibilance, a High Pass Filter for plosives, for example.
Now, enjoy a short video interlude demonstrating the various techniques for vocal level correction, including the fader technique as well as automation techniques including break-point editing, individual clip level adjustments, and some basic dynamic level control concepts including de-essing and multi-band compression.
Controlling Vocal Levels in Reason.
Multi-bands for Multi Processes
I will leave you with one final tip; you can use a multi-band compressor on a vocal track to deal with multiple issues at once. The high band is good for a bit of ‘de-essing’, the mid band can be set as a ‘smoother’ to only ‘reduce’ gain when the singer gets overly harsh sounding or ‘edgy’, and the lower band can be used to simply smooth the overall level of the ‘body’ of the vocal. If there are four bands available, you can turn the level of the bottom-most band totally off, thus replicating a high pass filter for ‘de-popping’ etc. Additionally, adjusting the level of each band will act like a broad EQ!
Setting the crossover frequencies with this setup becomes more important than ever, so take care and take your time. Remember you are actually doing (at least) four different processes within a single device, so pay attention not only to each process on it’s own but to the overall process as a whole. When it works, this can be the only processor you may need on the vocal track.
Multi-band Compressor as ‘Multi Processor’
…all of the techniques in this article, however helpful they can be, are not always required – do I even need to remind you all to ‘use your ears’ at all times? Using vocal rides as an example, I’ve mixed two songs in a row (by the same artist), one where the vocal automation looked like a city skyline and the very next mix where the vocal needed no automation whatsoever!
As always; “listen twice, automate once”!
Annex Recording and Trevor Price (singer/songwriter) for the use of the audio tracks.
Giles Reaves is an Audio Illusionist and Musical Technologist currently splitting his time between the mountains of Salt Lake City and the valleys of Nashville. Info @http://web.mac.com/gilesreaves/Giles_Reaves_Music/Home.html and on AllMusic.com by searching for “Giles Reaves” and following the first FIVE entries (for spelling…).