Recording Techniques (Part 9)
Part 9: The human voice is the most difficult instrument to record, bar none. David Mellor offers some practical guidance.
Despite what it says above, the human voice can actually be reasonably easy and straightforward to deal with in the studio — providing that you have a brilliant singer who is well experienced in recording, and all the necessary tools — such as a good mic, good acoustics, and good studio equipment. But good singers — good in the studio that is — are very hard to come by, harder even than good drummers. There are plenty of average vocalists about, and plenty of pretty good ones, but the sort that you can just place in front of a microphone and record in a couple of takes are worth their weight in platinum — and that's practically what it takes to pay them for session work.
So what exactly do I mean by a 'good vocalist'? The first requirement is to be able to learn the tune quickly, and learn it well enough to put expression into the words and music. The second requirement is to be able to sing in tune all the time, even when wearing headphones. This is probably the factor that will cause more headaches than any other. Put it this way, if your vocalist sings in tune, then you won't have to spend hour upon hour doing drop-ins, retakes, 'fixes' and whatever else it needs to get the performance right, bearing in mind that you are struggling against time as the performer's voice gradually deteriorates. Once again, more on this later.
Another requirement for any vocalist is the valuable personal asset of being able to do exactly what is asked. Obviously this doesn't apply so much to the likes of Michael Jackson and Madonna, who can hire and fire more recording engineers in a week than you or I can spit cherry stones. 'The ability to take direction' is a more appropriate way to say it. There are many musicians who are only too willing to do their best for you and play or sing in exactly the way you want, but willingness alone doesn't get the job done. The ability to take direction means that what you ask for, you get, performance-wise, and is not necessarily related to the amount of backchat you are likely to receive. Really good vocalists and musicians will not only do exactly what you want them to, they will also add suggestions of their own and give you not just 100% of what you asked, but 150% or more.
I could go on rhapsodising about the merits of using a really good vocalist, but if some are that good, why is modern music populated with singers who can't tell the difference between an A sharp and a Z blunt? The history of popular music through the ages, from Bob Dylan to Bob Geldof, is littered with singers who, by any objective criterion, just can't sing. The answer to this conundrum is that singing is more than simply articulating a tune and a rhythm, it's more about putting across an idea or a feeling. And some people who can't do the former very well are very good at the latter, and ideas and feelings are more important than the mere nuts and bolts of the music.
What this all boils down to in practical terms for the recording engineer, is that you are more likely than not going to be dealing with singers who are not at the technical level of being able to turn in a top-notch performance every time. They may be able to put the idea across, but need a lot of help from the engineer to do so in a musically acceptable way. If you look at it like this, it's not so much that recording vocals is a troublesome task which takes a lot of time, sweat, and tears — it's a part of the creative process in which the engineer can play a very full and active role. It's an exciting moment when a vocal comes together on tape, and it has all the elements of an artistically worthwhile and technically perfect musical performance. The next question is 'How is it done?'
The situation I am going to look at is where the engineer/producer has written, or otherwise acquired, a song and has asked a vocalist to come in and record it for him. Recording a singer in a band involves all the same processes, but the engineer would probably not have quite as much say in the proceedings.
As in many things, preparation is essential, and the first stage is for the singer to learn the song. As you would expect, some people are faster learners than others. I have known singers who can pick up a song after just a couple of run-throughs, and I have known others who just don't seem to have the ability to learn new songs! The best way to do it is to make a rough version of the song and give the vocalist a cassette a few days before the session. Obviously, it has to have a vocal on it, but even if it isn't totally in tune, it must put the song over in the way you want to hear it, with the correct feeling.
It's best to ask the singer what he or she thinks of the key; too high or too low. Even if you have to do some rearrangement and re-recording of the backing track, it's well worthwhile to have the song pitched correctly for the singer you are using. The right pitch is obviously when the song falls within the singer's range without too much straining for high and low notes. If you know the singer well, then it may be possible to go for a high key which you know could never be performed live, but can be pieced together in the recording. If it's the first time, though, play safe — there's enough to go wrong without asking the singer to wear uncomfortably tight trousers.
Learning the song is a matter of learning the tune and learning the rhythm of the words. The way the words fit a song is important, especially as they change from verse to verse. There is a world of difference between lyrics that fit and lyrics that don't. Often, lyrics can be made to fit perfectly by singing them in a certain way. If you know how it should be done, make sure that it's done that way on the rough demo cassette, so that the singer doesn't have to relearn it in the studio.
The other aspect to preparation concerns work in the studio. For instance, what time of day do you intend to record the vocal? In the morning, afternoon, or evening?
Funnily enough, most singers don't like working very much in the morning. I know that this applies to most of us, but it is to do with physical reasons, not that they are too lazy to get out of bed. Most people's voices tend to rise in pitch during the day. It's an interesting experiment to try out on yourself — match the lowest note you can sing on your keyboard early in the morning, and do the same just before bedtime. There may be a difference of as many as seven or eight semitones.
Whatever time of day the session takes place, it's good to have some 'hospitality' ready for your singer's arrival. This may range from a can of beer to a bottle of gin (not too large!), or it may just mean a ready supply of glasses of water — whatever keeps the throat well lubricated. Some singers insist on lubricating their throats with cigarette smoke and, whatever your feelings on the subject, you would be unwise to risk compromising their performance by banning smoking in your studio (which is actually a good idea anyway because the smoke, when it settles, contaminates and reduces the life expectancy of your equipment). Conversely, if you smoke and the singer doesn't — well I hope that giving up temporarily won't make you too irritable.
Before starting any serious work, it's time for a chat with the singer. You may want to chat about the song, about the rate of inflation, or about any other subject that takes your fancy. The point of this is to make you both feel more relaxed and able to communicate easily, which will be important later. Even if you know each other well, it's rarely possible to dive straight into work and achieve any useful results.
Hopefully, the studio will be set up already, with a selection of microphones on stands ready to be tried out, to see which best matches the singer's voice. The choice of microphones for vocals is a strange thing. Whereas most engineers will have favourite mics for the various different instruments, it is impossible to have just one favourite mic for vocals and use it all the time. There is always an interaction between voice and mic, which is especially true when the two are used very close together. Some voices will sound fine on Mic A and terrible on Mic B, other voices vice-versa. Sometimes the same person will sound good on one mic one day, and good on another mic the next day. Until someone invents a grand unified theory of vocal miking there is but one solution — try them all out.
When it comes to vocal miking, there are two Grade 1 champs, and a number of lesser contenders. Probably the most used microphone for vocals is the Neumann U87. It's not exactly a new design, but it does the business, as does the ever-popular AKG C414. The number three vocal mic, which for some unaccountable reason hasn't been in production for quite a few years, is the Neumann U47 — the valve U47 that is. Many engineers swear by the valve U47, and the current model of U47 with FETs (Field Effect Transistors) instead of valves is a pretty powerful performer too.
What studio vocal mics tend to have in common is that they are mostly large diaphragm capacitor mics, so it's a pretty good bet to try any you can lay your hands on, such as the Beyer MC740 or AKG's 'The Tube'. Even if it means a hire bill, it's worth it. I find that small diaphragm capacitor mics, such as the Neumann KM84 or AKG C451, don't have it when it comes to the human voice. In theory, a small diaphragm should be more accurate, but it's the ultimate result according to the ear that's important, not the spec sheet. Dynamic mics can also be useful for vocals. The Shure SM57 and SM58 are classics, as is the Electro-Voice RE20 (the world's ugliest mic in my opinion), and can offer the right sound for certain types of music, particularly rock music — but they can never be as 'crisp' as a capacitor mic.
Whatever mics you have available, it is well worth having two or three set up so that you can have a quick listen to each and make a choice. It's not usually wise to appear to be 'messing about' with either the mics or their positioning. Musicians tend to see this as unprofessional, even though it's a vital part of the engineer's art, and this may compromise the performance.
The mic position can have a great effect on the sound of the voice. Distances from zero to about a metre are useful — more if you are recording an opera singer in a large room. The further away, normally, the more natural the sound; the closer to the mic, the more present and intimate the singer appears — it can be like whispering in your ear. Equally, the closer the singer gets, the more chance there is of the mic 'popping'. This happens on consonants such as 'p' and 'b', where a jet of fast moving air bursts from the lips and blasts the mic diaphragm. Popping is nearly always fatal to the recording. Even with a heavy bass roll-off it can usually still be heard when the song is mixed, so it's definitely something to avoid.
There are two ways to deal with popping. The first is for the singer to sing across rather than directly at the microphone, and to ease off on the offending consonants. The other is to use a pop shield. Most mics have accessory wind shields which are either supplied with the mic or are available as an extra, and simply fit onto the business end of the mic. These are great for keeping spit out of the mic and for working outdoors, but they don't do much for popping, unfortunately. The best tool for dealing with popping comes absolutely free and consists of two components: a wire coat hanger (free with your dry cleaning) and a length of nylon stocking (free when you ask nicely). I'll leave the mechanical arrangements up to you, but basically the requirement is for a lightly stretched membrane, eight to ten inches across, positioned in front of the mic. Stocking material is full of holes and will allow sound to pass through unimpeded, except at very low frequencies — ie. the popping frequencies — which are absorbed by the membrane and therefore will not blast the mic, or at least will blast it a lot less. The two variables are the size of the area of stocking, and the tension. Get the combination right and popping should be a thing of the past.
Choosing the mic, setting it up, and adjusting the pop shield can all be done while the singer is running through the track. 'Running through' doesn't mean the same as 'rehearsing', it's just a way of getting used to the song, the setting, and a time to get things (including headphone foldback) basically right. The rehearsal period itself is where serious work starts: the vocalist starts to sing out and the engineer sets the record levels and fine tunes the foldback mix. It is worth taking a closer look at foldback and the best way of supplying it.
Obviously the singer has to hear the backing track, plus what he/she is singing, in the headphones. The headphone volume has to be at the right level, and also the mix of the instruments. The foldback mix may not have all the instruments in it, just enough to give the rhythm and tuning. Although most mixing consoles are geared up to having a mono foldback mix, it really is best to have it in stereo, which simply means using two auxiliary sends instead of one. The level of the voice in the foldback mix is important. Singers react to what they hear, and if the amount of voice is too much, it's possible that he or she will ease off slightly. The singer may ask for more level in the headphones, but while accommodating their requests, think also about what they really need.
When the engineer has such a wide range of effects as are available in the modern studio, it is tempting to use a few of them while recording the vocal. I'm not going to say don't do it, but simply point out the options. If you have rackfuls of effects units, then there is no point in recording any effect (compression, reverb, etc) with the voice — add it to the monitor mix while recording, and for real in the final mix. A flat dry recording of the voice allows much more opportunity later. If you don't own so many effects, then it may be useful to do some basic work on the voice while recording.
Vocals nearly always need a little EQ and compression, so it is fine to EQ and compress onto tape, but not too much! Gating onto tape is more dodgy. You will want to gate the vocal track because it always picks up a lot of background clutter — breaths, headphone spill, and the like, but gating as you record risks losing odd syllables which you may come to regret later. Adding effects like reverb or echo depends on the nature of the recording, but as a rule it's better to leave it for later. Whatever you do about reverb, send some of it to the foldback mix. Reverb on the vocal is an enormous help to the singer. Even the best vocalists can't keep in tune unless they can hear themselves clearly in the headphones, and having a bit of reverb really makes a difference.
When it comes to the rehearsal proper, after thinking about all of these points, it is time to check that all is going well, that the singer is singing in tune and in the correct rhythm. If the tuning is suspect, the first thing to suggest is that the singer moves one side of the headphones slightly away from the ear, so that some of the natural sound of the voice gets through. It is difficult to hear tuning on headphones, so this improves matters. With some singers, it is a problem for them to come in on the right note. There is only one answer to this, if you discount singing lessons (which can actually do more for a vocal performance than any amount of studio trickery), and that is to record tuning notes on the tape at strategic points. This may seem like a chore, but time spent doing this can be a valuable investment in some cases.
Regarding the rhythm of the lyrics, if all does not seem to be going smoothly, then a good trick is to mark on the lyric sheet the points at which the singer should breathe. I do it like this:
/This is the first line/ of the song
/It's all rather slow/ and long
/So the singer needs/ a lot of/ breath
/To manage the/ sus/tained/ notes
Apologies for the rhyming scheme, but I can assure you that it works and has an added advantage in that the gaps in the vocal line will be in exactly the same place in each take, which is good for drop-ins.
There are two basic techniques for recording any overdub, including vocals. One, which I shall call 'assembly', uses only one track. The other, 'compilation', needs several. So it's no good recording 23 tracks and leaving only one free for the vocal, then expecting to use the compilation technique. Let's look at the more economical assembly technique first...
If you have a brilliant vocalist, then you will be able to record all the way through the song and then just drop in a few corrections. But if the vocalist is only average, remembering that even for a good singer it's much harder to record than to perform live, then it can be very depressing when you have several goes at the song, all of which seem virtually useless. The answer is to break the song down into smaller parts. Get a bit of it right, then carry on.
If I use this technique to record — I recently used it on one song containing an extremely difficult vocal line, which the singer hadn't an earthly chance of learning — I would probably aim at getting the first two lines right. This may take five takes, 10 takes, or 40 takes (no exaggeration) until it sounds good. I actually swap between two tracks of the tape, so that when I have a good version, I can always get the singer to try for a better one without recording over the first. With a good first two lines, confidence builds and the rest of the song tends to go more smoothly. I won't deny that it's a slog to work in this way, or that it will produce the most perfect, flowing vocal line, but I do promise that with enough patience it is nearly always possible to obtain a usable vocal, even with a less than perfect singer.
The compilation technique needs a reasonably good singer to start with. The aim is to go for complete takes of the whole song, possibly about six of them. From these six takes it should be possible to compile one perfect version by bouncing down onto another track. Sounds easy?
Well first of all, the singer has to have enough of a good vocal technique to be able to sing all the takes with the same energy and with the same tone of voice, otherwise they won't match up when they are compiled. Secondly, it can be a big job just choosing which parts of which take you are going to use. Well no-one said that it was going to be easy to record vocals, so here goes...
What you need for the selection procedure is either six copies of the lyrics (or as many as you did takes), or six different coloured pens or pencils. Then as the song plays through, starting with the first take, you mark down the quality rating of lines as A, B, C, or X. Guess what 'X' means! It does take a bit of practice to get used to doing this, to assess a line very quickly, mark it, and then move on to the next without stopping the tape. If both you and the vocalist take part in this exercise, you should be able to come up with a happy compromise, eventually.
After doing this, it's time for the playoffs, where you compare line by line, judging one (hopefully) high quality line against another. After an eternity, you'll have a list which you can string together into a single perfect track, maybe with just a couple of drop-ins to fix it completely.
With either recording procedure, you will face the same problems. Here are a few:
You need to comment on the singer's performance.
Develop a selection of platitudes for all occasions, such as:
"That was great but..."; "I think you should have another go at that"; "I know you can do it better"; "Let's try that again"; "One more time"; "That was the best yet, but..."
The sound is inconsistent between takes.
The singer is probably moving around. Explain the problem carefully and you'll achieve a better result. If the singer is coming in and out of the control room, it can be worthwhile to mark the studio floor so that he/she always goes back to the same spot.
The vocalist consistently sings a phrase flat.
If a phrase keeps on coming out flat, and you know it and the singer knows it, suggest that he or she tries to sing sharp — not in tune but sharp. It won't come out right first time, but once the boundaries of the correct pitch have been explored thoroughly from both sides, the next take might be spot on.
The 'feel' of the vocal isn't right.
This is a tough one. One trick that sometimes works is to suggest that the vocalist tries to imagine a mood appropriate to the song — romantic, energetic, aggressive, etc — rather than simply what it feels like to be put on the spot in front of a cold, unfriendly microphone.
There are two qualities that the recording engineer/producer should cultivate which are of prime importance in recording vocals. The first is to be able to draw a good performance from the vocalist, to watch the performance develop during the run-throughs and early takes, and to capture it before the edge goes and the voice begins to fade. The other is to be able to listen to a take and know whether it is perfect, nearly perfect, acceptable, or not good enough, and to be able to tell the singer to sing it again without damaging their enthusiasm or confidence.
More than any other factor, the vocal performance can make or break a recording. The engineer or producer who can encourage the absolute best out of a singer has a rare talent. It's more than just operating the equipment and following the correct procedures, it involves a musically intimate person-to-person contact that, despite its difficulties, can make recording very satisfying indeed.
Feature by David Mellor
Previous article in this issue:
Next article in this issue:
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!