The other day I was sent a track to mix and in the process I was asked if I minded choosing, or compositing, the lead vocal. Now I’ll put aside, for the time being at least, my feelings that this really should not be left to the mixing engineer, as it’s really the producer’s job, but in this case it was a friend and someone who was really asking for my help and so I agreed.
So you can imagine my shock when upon opening the session I discovered no less than16 complete vocal takes!
I won’t bore you with the details of how I ended up sorting through all that, or expose you to the stream of obscenities I unleashed on my empty room, upsetting the cat and accomplishing little else, but the whole experience got me thinking about the choices one makes about the process of recording vocals.
Now of course this same thinking could be applied to overdubbing anything really but it seems to be primarily vocals where this kind of decision-making rears its head most often. One rarely sees, for example, 8 takes of rhythm guitar to choose from. The only other place people routinely record potentially ‘too much’ like that is probably in basic tracks. But let’s not digress.
If I may, let me take you back in time to the days of analogue recording on 4 or 8 tracks. Odds are that those precious few tracks were filled with instruments and that we had at most one track saved for the lead vocal. I say “at most” because often other things, hand claps, backing vocal, a guitar solo, things like that, had to be put on to share that same track at the same time. So naturally this meant that if the vocal wasn’t up to snuff the choice had to be made on the spot to erase it and do another one. Which meant of course that everyone involved needed to listen and decide and agree right then and there.
As we moved to 16 and 24 track recording, it became more common for there to be a few vocal tracks available. Still not the unlimited number of playlists we can have today, but at least we could record, let’s say, three vocals and then listen and choose, and so began the common practise of ‘comping’. Compositing, a vocal performance from several takes down to one, either by bouncing or just by making notes and making those switches on the desk mutes in the mix. A lot of the time, if one track had the bulk of the ‘good stuff’ we would simply drop in to that track bouncing bits from the other so that it ended up being the composited master track. Again, decisions.
Make Choices, Don’t Defer Them
I’ve talked before about why it’s probably a better idea to make those decisions in the heat of the battle and not put them off to make your, or someone else’s!, life miserable later. Needless to say this used to be a real skill that better engineers developed to be able to do tight drop ins that sounded natural. As tape machines got faster, and with PERC timing circuits to minimize the gap between unerased old signal and the new overdub caused by the physical gap between the erase and record heads, dropping in became easier.
But dropping out was always an adventure. That’s because, in the heat of the moment, and with all of that extra breath and leeway created by only focusing on a short section, a singer or musician could decide to drag the rhythm out or add a turn at the end of the phrase and in doing so ‘accidentally’ overlap the drop out point. The real danger was in erasing the start of incoming side that you were trying to keep by dropping out. For that reason, on anything approaching a tight drop out, we developed the technique of cutting the tape and inserting a safety “island” of leader tape of at least a few seconds. That way you could just let the machine roll and know that if the singer went long, the end of the new pass would be truncated (by hitting the leader) but the old performance of the next line would be protected. Once we had the new line in place, we would take out the leader and splice to two ends back together.
Magic! Except when it wasn’t.
My most terrifying experience with this was on the Cyndi Lauper track All Through The Night. Somehow it had been determined that the very first line of the song, always an important first impression moment, wasn’t “good enough” and so we undertook to have another go at it. This was fairly late in the process and there was only the one vocal track available. That in and of itself wasn’t the issue, because we only ‘needed’ one take of one line, right? But, the line was “All through the night, I’ll be awake and I’ll be with you” with only barely a breath where that comma sits in that written sentence. And all we wanted was the “all through the night” (the bit before the comma.) and to keep, which is to say not erase the “I’ll be awake…” So it went without saying that the slightest timing variation from Cyndi or slightly late drop out from me, could cause us to end up wiping a bit of the next word, “I’ll…” So an island was the obvious choice. And that’s what we did. She went out to sing and as it turned out, getting that line to everyone’s satisfaction under the microscope now took a lot longer than we had thought. Hours in fact. And when finally everyone seemed happy, I asked John Agnello, my then assistant, to take out the leader island and splice the ends together. Which he dutifully did.
Now it should be said that this is a rather soft delicate moment with a vocal over a sparkly synthesizer arpeggio and not much other cover. So now with the join done, we played it back, and with great anticipation we heard:
“All through the night, BANG!, I’ll be awake…”
It was like a firecracker pop went off on the ‘comma’.
It was standard practise for the studio to always demagnetize an entire box of razor blades for editing, but somehow, something had gone wrong and we had hit an unlucky magnetized blade creating a massive spike across 24 tracks at the edit point.
You can imagine that no one was happy, especially Cyndi after singing that bloody line for 10 hours.
The solution was found in the fact that the second verse began with the same arpeggio, just with other elements added on other tracks. So I ended up mixing the whole song, and then mixing the first line from the second verse, with everything but the vocal and arpeggiator muted, and then cutting in just that needed line into the front of the mix. Fortunately it worked. But the lesson was to only overdub what you really need to do. And to never trust a razor blade.
It’s a good thing to ask yourself if you’re convinced there is a good reason to do another take, and having to erase the previous one forced that thought process. But still the point here was the limitations of available tracks. And comping vocals from several takes wasn’t the only way to do things. First off, we didn’t always have several tracks left to play with. But also it’s just not everyone’s preferred method. And on the whole it wasn’t and still isn’t mine.
For me, and many others, the preferred method became dropping in to one existing track and getting the bits that needed to be replaced instead of always shooting for an entire new performance.
The singer sings the song, and then we decide if it’s a good performance or not and accordingly either discard it entirely and do another or decide to keep it and start refining it. Now that might mean that the last chorus doesn’t build enough, or that one line in the second verse has a flubbed word or is phrased incorrectly and so on. But rather than singing the whole song again on another track, we would just go to those points and do a drop in to grab the ‘missing’ bits. Eventually when all the repairs are done you have one, and only one, complete vocal track.
Now for my brain this is a much better method. I’m living with the vocal in the moment and hearing what works and what doesn’t and when it’s done, it’s done.
Would a guitar player replay the entire song just to fix a bad note in bar 64? No, you’d drop it in. But on the other hand, a vocal also tells a story, so continuity becomes much more crucial. And not every singer is the same. Some can put themselves in the moment and the character and just go right after one word in the last verse and match it perfectly and some can’t.
I find that when dropping in most singers do better, in terms of that continuity, if they at least sing entire sections of the song, or at least pairs of lines, couplets, as opposed to just a single line or less. The more experienced singers know what works for them. Mick Jagger told me, for example, that if we needed to fix something he’d rather drop in complete couplets, not less. But he was also, in saying that, telling me that he could make it fit the rest of the vocal like that without needing to sing more either.
By contrast some more ‘technical’, as opposed to story oriented, singers I know could just sing along from just about anywhere and drop in a single word that matched perfectly. You need to know your singer. Often after completing a vocal by dropping in one section at a time to get it ‘right’, a singer will go home with the result to live with and internalize, and come back a few days later ready to sing it again and nail it. But just as often that’s not necessary because that comped together, dropped in, vocal is great.
Some singers though are on a real emotional ride through the song and can find it exceedingly difficult to just tune in to the intensity level at one point of the song without going through the whole journey to get there each time. They want to sing the whole song even if really you’re only aiming for one line.
And of course the good news (is it?) about modern DAW recording is that we can actually record that whole new vocal pass and prepare to be happily surprised if other parts of it beat the original we’re trying to fix one line in. Just as long as someone keeps the focus and it doesn’t become that unmanageable 16 takes to sort through.
If doing multiple whole passes for compositing afterward, a limit on how many takes is generally a good idea. Although there are some superhuman exceptions, almost no one has the brain power to keep track of 20 tracks of vocals even with the best checklist system in the world. Four seems to be a good number where it’s not too many to keep mental track of and not too tiring for the performer so that they start varying too wildly.
Also, just like I talked about when dropping in, ideally one track seems to ‘have it’ enough to predominate and then other tracks can be looked at to contribute small sections the main choice seems to be lacking in. But when it’s a complete hodge-podge with no one track clearly in the lead (as it were) odds are you don’t have it yet. Trying to create a Frankenstein’s monster from a pile of vocal bits, none of which are able to stand on their own, is rarely going to lead to a satisfying result.
When the composite method is employed, I find a written checklist system to be useful for comping vocals. I create a sheet with the lyrics and four (or how ever many tracks, but four is a good number) columns. As we listen to the performance, I rate each line in the track’s column with a check mark or a check+ or check- or just a minus sign and so on. Sometimes it’s a check with two pluses! The likely winner. Then after listening and grading all four performances I can see fairly quickly which was my choice for each line. Then we’ll do a run through based on those choices and see how it all feels together. Often if a second choice line was part of the performance that was chosen for the surrounding lines, it’s a better fit even if it wasn’t the top rated choice.
As always, whichever method you choose, it shouldn’t sound overdubbed or comped. No one ultimately knows how you got there if the result works. And really that’s the trick with any overdub; it should ultimately sound as though the band all played together.
So coming full circle back to my mixing project with the 16 takes, what did I do?
I started by listening to the first take and the last take to see the general trend. Is he getting better? Was he losing some spontaneity or energy? There was little question in my mind that the first take killed the last. So I moved on to take 2. And then 3. And 3 was sounding really good to me. So I simply stopped. I used 3 and only went searching (on 2 or 4) for anything glaringly wrong, like a missing or wrong word. And in the end? They loved it.
That’s 12 takes of singing they didn’t need to do. I’ll let you in on a pro mixer secret: This is the sort of thing we do all the time. Those 6 guitar mics or sets of room mics you send? Odds are the mixer is picking one, sometimes randomly, and throwing away the rest. Overkill in choices rarely makes things better.
So if there is an overriding lesson in all this it’s that it’s a good thing to ask yourself if you’re convinced there is a good reason to do another take, and having to erase the previous one forced that thought process, but even now that you can do as many as you like, perhaps you’re better off still asking yourself if you really need to.
The sanity you save might be your own.