So you want to make an album? (part 22)
KeithHandy posted in Producing, So You Want... on October 22nd, 2007
To read the entire series, go to the “So You Want…†category.
Installment 22: Click or no click (human time vs. machine time)
Before you even begin record a song, there is one global decision you need to make about your approach to it, and this decision will affect every successive stage as you go forward.
It’s the question of “click or no click”, or more precisely, “human time vs. machine time”: How do you establish the tempo of the song when you put that first track down? Do you find a good sounding tempo on a drum machine or computer sequencer and stick to that? Or do you just pick up an instrument (including a drum set) and just play? There are advantages and disadvantages to both.
Let’s start from the initial reasons why, as novices, we might choose one or the other without even really thinking about it.
Why use a click (machine time)? Maybe because we’re using software that, by default, displays a big tempo, shows bars and beats while playing and recording, and clicks on the beats while recording. All we have to do before we start is adjust that tempo until it sounds appropriate for our song. Then, as we continue to pile overdubs on, we can easily see where loud notes and percussive hits fall relative to this pre-established grid. This also makes it easy to mark sections of the song; the intro might start on bar 2, the first verse on bar 10, the first chorus on bar 26, and so on.
Why (again, as novices) would we not use a click (human time)? Again it could be what the software leads us to. Audacity, for example, functions a little more like an ordinary tape recorder than a sequencer (ideal software will be a combination of both). Or maybe you’re not even using software. If you have an ADAT machine or some kind of portastudio (although digital portastudios can have programmable tempo grids, and provide a reference click), you might be inclined to just “think like you’re using tape”.
There’s nothing wrong with either rationale for the person just getting started, because we have to just try it out one way or another to start learning and figuring out what works for us. But the longer we do this, the longer and more complicated the list of pros and cons gets on both sides.
Will it flatter my performance?
You want the music to sound good, but also, as a human being with an ego, you want to sound like you are a good musician. (Subtle distinction there.) Even if the drummer is anonymous, you don’t want the listener’s reaction of “yikes, what a shitty drummer” to toss a wet rag on their overall listening experience. So which method is going to make your drumming, bass playing, guitar playing, etc. shine the brightest? That depends.
Playing to a click is a separate skill that you have to practice and learn. It’s not quite as natural as playing to a drum track, which is why some click tracks will come with the option to sound more like drums — or you could have a temporary track using a sampled drum groove and replace it later. Or decide you like it as is, and just keep it. No matter what sound you use as a reference, there is one predominant challenge in playing your instrument along to it: avoiding “the lurch”. “The lurch” is when you’ve just been playing a hair behind (later than) or ahead of (earlier than) the beat for a few moments, and suddenly correct yourself. The moment when you spontaneously correct yourself will sound worse than what you were doing just before it, especially if you listen to this new track isolated. Having it in a mix may cover it up to some extent, but the less you rely on that, the better. When it comes time for mixing, you’ll want to be focused on creating a good balance, not on covering up awkward rhythmic glitches.
In order to play well with a metronome/click, you have to loosen your mental “grip” on the individual clicks, and try to feel the overall tempo more. In other words, mentally zoom out to the bigger picture. You still need to have an internal sense of tempo, which you strive to keep in agreement with the click, but you’re not checking each click one at a time and asking, “was I early or late for that one?” It’s not target practice. You don’t win a prize each time you “hit the click”.
This same challenge applies when you’re overdubbing on top of an existing performance, even if that was played freely by a human in human time, although in that case it should be much easier, because hopefully the slight timing irregularities in the existing tracks will correspond to how the song naturally “feels” to you.
The act of putting down a first track in human time, however, carries with it its own headaches. “Good tempo” is subjective, and our own idea of what a good tempo is will drift a little depending on what we heard previously, how much energy we have, and so on. Our perception of something as “too fast” or “too slow” also depends on how the rhythm divides the beat up, whether we play staccato or legato, how tight or loose the instruments are, how the song is mixed, and probably a zillion other factors. Compound this with our natural tendency to speed up as we go along.
All that said, there are tons of drummers who will play much better without a click track, for the reasons I outlined above. Similarly, if you’re doing a song that’s more acoustic based, with the “glue” of the song being, say, a guitar or piano, you can sometimes get a more natural performance if you’re not distracted by the “TOC toc toc toc TOC toc toc toc” of the almighty beat-per-minute counter.
What about logistics?
Using machine time from the get-go gives you some advantages later on. As I said earlier, you have a clearly laid-out grid where, at any time, you can see exactly what measure and beat you’re on, and have an easy visual reference if you want to adjust the timing of a note/hit by making careful cuts in the silent areas just before and after it and then sliding it to the left or right. If you decide later in the game that you want to add a keyboard part with a very exact rhythm, you’ve made this very easy for yourself — you simply sequence the keyboard part, by playing into the sequencer, drawing notes on a piano roll grid, or using a variety of other input methods — and then because your project is “aware” of where the beats are, you can easily quantize all those notes to the nearest sixteenth note, eighth note, or whatever your smallest subdivision is.
I wouldn’t say that human time has any logistic advantages, per se, but with all the artistic benefits, it doesn’t have to be a logistic nightmare.
For starters, you could simply go old-skool all the way. Not all that long ago, people used tape, and it was impossible (or extremely difficult) to synchronize a sequenced track to an existing performance. If you were going to combine sequenced performances with human performances, you always did the sequenced stuff first, be it drum machine or whatever. Eventually people came up with ways to apply a time code to one of the unused tracks, so that a computer could remain synchronized to the tape. But in this relatively post-tape era, that kind of workaround seems like a Rube Goldberg solution.
Anyway, I digress; my point was, there is absolutely nothing wrong with recording a drum set, then playing bass to that, then playing guitars to that, then singing over that. “Look, ma, no click!” This is perfectly legitimate. If your software’s timeline lets you choose between “bars/beats” and “minutes/seconds” views, choose the latter, because the bars and beats at the software’s default tempo will be irrelevant. The click would also have no relation to the song, so you’ll be leaving that turned off too.
If your human time performances are mostly good, but have a couple of sloppy spots, you can often look at the peaks on the drum track as a visual guide. If, however, you’ve started off a song in human time, but want all the advantages of using the software’s bars/beats view and quantization capablility for new sequenced tracks, you can still create a tempo grid after the fact that will perfectly align to your human performance. It won’t be as easy, but you can do it. I believe there is some software that will help you with this, but I’m going to assume you have to do it the hard way.
Let’s say you have a drum track recorded that you’re really happy with, and you’ve already edited it and cleaned up any bad timing by ear. First get an approximate idea of of its tempo. Generally, when you set tempos, you can tap something and the software will figure it out based on your tapping. Do this while your drum track is playing, watch the tempo fluctuate up and down a bit, and whatever it appears to average out to, go ahead and set this as the tempo for the whole project… for now.
Find beat one on your drum track, and drag the clip either left or right so that the peak is exactly on a barline, and set a tempo change exactly at that barline — for now, to the same tempo it already is. If your drummer’s tempo is really steady, you may be able to get away with doing four bars at a time. (If you actually started by recording a drum machine, but it wasn’t synched to anything, you should be able to do even more.) In a lot of cases it’s best to do one bar at a time, which will take longer. Keep looking ahead to the next significant downbeat. If the barline is to the left of the hit, you need to decrease the most recent tempo change; if the barline is to the right of the hit, you need to increase it. Make smaller and smaller adjustments until you zero in on it. Then as soon as that barline lines up with that hit, set a tempo change there, and look to the next significant downbeat to adjust that one. If a bar doesn’t start with a clearly defined downbeat, use an earlier or later bar (you don’t have to use the exact same number of bars for each tempo change). Listen back frequently with the click turned on, from the beginning up to where you are, to make sure you don’t hear four clicks where there are supposed to be three or five!
This might seem tedious while you’re doing it, but in the grand scheme of things it takes a negligible amount of time out of your life, and gives you the best of both worlds: a natural drum or guitar (or whatever) performance, and the ability/freedom to quickly add a precise keyboard part. (Of course, if the only keyboard part you’re going to add is something with soft, slow attacks, like a “pad”, as they call it, this might not be worth the trouble — but it’s great if you plan to add anything rhythmic.)
I’ve just described how it’s possible to harness the natural push and pull of a human performance, but remember you could also make a “human-like” tempo grid by starting with sequenced material, and speeding it up or slowing it down as it goes along. Some software lets you draw the tempo curve with an icon resembling a pencil. Other software requires you to enter individual values. Sequenced tracks, as opposed to audio tracks, will actually speed up and slow down according to your newly added tempo changes. Sometimes it might be good to first record a keyboard sequence to an unchanging tempo, and then fiddle with the tempo curve until you’re happy with how it sounds. I would advise you not to record any audio tracks until you’re completely happy with this curve.
There’s some trial and error in doing this, especially if you’re trying to get it to sound “human”; but on the plus side, once you get it to where you’re happy with it, your project will be very easy to work with, whether you’re recording audio or adding additional sequenced tracks. If you’re deciding whether to create a tempo grid first, or base a tempo grid on a human performance, it comes down largely to what instruments you’re featuring. If the song is heavily keyboard based, i.e. you’re using a lot of orchestral samples to simulate the sound of an orchestra, then this curve-drawing method might be ideal. If it’s more drum and guitar (or other “real instrument”) based, then I suggest the method I explained previously, where you record the performance and then map it (if you even need to map it, that is).
A few notes about fixing audio tracks
It’s not as easy to fix audio performances as it is to fix sequenced performances, where you can literally grab a note and just drag it into place. But it can be done. You want to start with the best performance you can, but there will likely be a few parts that still don’t sound 100% right to you. Slicing an audio track into little pieces and nudging them around may seem blasphemous, but it’s not as much of a black art as you might think.
1. Be sure to cut in a good spot. The quieter, the better. Cutting in the middle of a note will generally sound unnatural. And no matter how perfectly quiet of a spot you cut it at, always put a very short fade in/fade out — even if it’s just a few milliseconds — on the beginning and end of every new clip. Some sounds will be just quiet enough to not show up on the screen, but if you cut in the middle of any sound whatsoever without a fade, there’s a chance you will hear an audible click.
2. If there aren’t any good edit points on a track, but you absolutely need to shift something in time, use a crossfade. In most cases a “square root” fade curve, the one that bulges upward, will keep the overall volume most consistent through the overlap. If it still sounds noticeable, experiment with longer and shorter crossfades. The attack of a note should not be in the crossfade.
3. Don’t sacrifice horizontal timing for vertical timing. Any time you’re moving isolated bits of a track around on the timeline, be sure to listen to that track by itself and make sure it sounds consistent.
4. In real life, it’s normal for certain instruments to tend to play a little earlier or later than the beat. A strummed guitar sound will usually start a little before the beat. A piano will generally be a hair later than the beat. Also, a sound that’s slightly off the beat will stand out more, and you can use this to your advantage.
5. If a drum fill is a little too fast or too slow, but the hits are evenly spaced, it could still sound good. If they’re unevenly spaced, it might sound a little more awkward and amateurish. See if you can make a clumsy drum fill a little nicer, but be sure to crossfade carefully, especially if there are any ringing cymbals. (This would be suicide on tape, but we can at least try it, because we’re the “undo button generation”.)
6. We are naturally pickier about some things than other things. For example if we have really tight drums and bass, we can get away with a little looseness in the guitars. Since perfectionism tends to be addictive, we need to know when to stop. Listen for little rhythmic flaws in well-known songs by your favorite artists, and ask yourself if they actually bother you or just give the song character. Having every single sound precicely on the beat (or subdivision) can sometimes be impressive — but what’s necessary, and what’s overkill?
