Recording Techniques (Part 16)
The Mix (2)
Part 16: The mix (2). David Mellor goes still further in pursuit of the elusive perfect mix.
There may be no such thing as the perfect mix, but there is such a thing as the best mix you can possibly do within the capabilities of your equipment and surroundings. If you don't emerge from your mixing session feeling totally mentally exhausted then there must lurk at the back of your mind the possibility that you could have done better — this is what distinguishes the real recording engineering men and women from the boys and girls. But even if it's hard work, it's a lot of fun too, and the feeling of achievement you get when things go better than you could have hoped makes it all worthwhile.
In this and the next article in the series I'd like to present a multi-course meal of food for thought, ranging from how an engineer listens to the sound he or she produces right through to how to present your tapes so that the next engineer down the line towards the LP or CD knows that they are dealing with the work of a professional. Here goes...
Monitor speakers, and their environment, have two distinct and not necessarily compatible requirements. In no particular order of importance (I have to say that to prevent 50% of the people who read this sending me angry letters) these requirements are: firstly, to enhance the sounds produced by the musicians and act as an aid to musical creativity; secondly, to tell the engineer what's going down on to tape. As you can see, they are both essential, but almost totally mutually exclusive.
For decades, top recording studios went for the biggest and loudest monitors they could afford. Some speakers were more accurate than others, but sheer weight of sound lends excitement in itself, so all concerned could get all the vibe they wanted from the music, and the sound could still be a reasonably accurate representation of what was on the tape. There's a trend towards smaller speakers these days, though personally I think there's no such thing as a good small loudspeaker. There are some small speakers that are good considering, and I'm not denying that small speakers have their uses in the correct situation, but the laws of physics give all the advantages to big 'uns.
Small speakers — or near field monitors as they have come to be known — were originally used to assess how a mix would sound on a transistor radio or domestic radiogram. As the quality of stereo systems has increased, so has the quality of the near field monitors, and used for their purpose they give the engineer a good deal of useful information. The principal problem is that small speakers lack bass. Sometimes they do sound quite bassy, but if you compare them side by side with a larger speaker you'll realise how much is missing.
The temptation is that if you don't hear much bass coming out of the speakers, then you'll add more to the mix to compensate. If you do this, then when someone does come to play the mix on larger speakers, it will sound much more bass-heavy than you intended, and believe it or not there can be such a thing as too much bass! Adding bass to excess also causes headroom problems where the speaker runs out of excursion because of the deep bass, and mid band loudness suffers as a result. The moral is to think twice before adding bass when you're monitoring on small speakers.
My feeling is that since there's no such thing as the perfect speaker, the best way to get a good mix is to listen to your recordings on as many systems as you can, including the car stereo, to get a feel of how well or badly they travel. Remember that unless you're playing back on a really dodgy system, if the mix sounds bad then it's the engineering that's at fault. A mix which only sounds good in the studio is by no means a good mix.
This isn't strictly a mixing technique, but since it's mix-related, now seems as good a time as any to mention it. Not many people know that it's quite possible to extend the capabilities of a 16-track tape recorder to 17, 18, and perhaps even 20 tracks. And no, it doesn't involve synchronising a drum machine or sequencer. The trick is to record more than one instrument on some tracks.
In the old days when few studios had more than eight tracks, engineers had to use this technique all the time, but since we have got used to having more tracks to play with, people have tended to think in terms of one instrument per track. But in most arrangements, not all the instruments are active at the same time. Let's imagine a song which has a number of saxophone breaks and a guitar solo, but the sax doesn't play during the guitar solo, and the guitarist only has that one spot in the song. If you're running short of tracks then the obvious answer is to 'double bunk' them on one track since they are not getting in each other's way. The only drawback is that the mixer settings you choose — EQ and fader etc. — to suit the sax will probably not suit the guitar when it comes up.
Even when two or more instruments 'timeshare' a single track it is still possible to treat each one differently on the mixer, provided you have enough spare channels. If your patchbay has sockets wired together in parallel — as it should — you can use these parallels to direct the timeshared track to as many channels as you need on the mixer (see Figure 1). The only snag is that you have to mute the sax channel while the guitar is playing and vice versa, but this is still a lot easier than trying to alter EQ, aux and fader settings as the mix is running. If you have a MIDI muting system then you can get as many tracks out of your multitrack recorder as you have channels, with the all muting done automatically.
Mixing your own tracks for the benefit of your own ears is a luxury that the jobbing recording engineer doesn't have. He or she has to use initiative to balance the sounds well and add appropriate effects, but the final guidance will come from elsewhere. As I have said earlier in the series, taking all recording studios from 8-track upwards into account, in most cases the engineer is also the producer of the recording. You are recording a band, one of whose members is the self-styled 'producer', it is unlikely that that person's experience in achieving a good recording and mix is as great as the engineer's.
The problem when other people want their say in the mixing process is that they may, in all truth, not have a clue what they are rabbiting on about. I don't want to appear to be insulting to musicians and budding producers, and I don't deny their ability to learn how to make excellent recordings given the chance, but from my experience there are a lot of 'experts' on recording who think they know more than they actually do, and if the engineer follows what they say to the letter, then the result will be a bad mix. This is where the twin qualities of tact and diplomacy come into play; you can't be a professional recording engineer without them. The ultimate aim is for the band or producer to think that all the creative input into the recording came from them alone, and that the recording engineer was very quick and efficient at following their instructions. In reality, although the mix will hopefully fulfill the producer's requirements absolutely, the recording engineer is usually the driving force behind the creative effort.
The recording engineer's nightmare scenario is where the band that's paying for the session are all gathered around during the mix throwing their comments and suggestions in. Usually the vocalist thinks the vocals could be louder, the bass guitarist thinks the mix needs more bass, and so on. It's a funny thing, but no-one ever asks for anything to be quieter. I wonder why? I was once asked to mix a tape of a 5-piece band and agreed on the condition that only one band member was present. I was slightly perturbed when all five turned up at the studio, along with their manager — but only one of them spoke to me during the whole mixing session, and they kept their band conferences to the breaks. I'd recommend it as a way of working with bands.
"A mix which only sounds good in the studio is by no means a good mix."
Ears are funny things. They don't just look funny on the side of your head, but they can sometimes hear things which aren't really happening. I was once mixing a band and the guitarist asked for a little more brightness on the lead guitar. I thought that, if anything, the guitar was already too bright, but since he was the one with the cash I thought I had better appease him, whilst adding as little extra top as possible because I really thought it would spoil the mix and he would regret it later. I reached for the high frequency EQ control and turned it clockwise an infinitesimal amount.
"Is that OK?", I asked hopefully. "A bit more", he replied. I gradually edged it up more and more, and the guitar was getting brighter and brighter, but still he wanted more top end. Eventually I reached a point where it was so bright it hurt, and the guitarist decided it was just about OK.
I never told him, because he was satisfied with the end result, that although both of us had heard the guitar getting brighter with the extra high frequency EQ I was applying, we had both been fooled. I had accidentally grabbed the HF control of a channel I wasn't using, and I was turning that instead. I wasn't changing the guitar sound at all, but because we both thought it was happening, we heard it. As I said, ears are funny things, and easily fooled.
The basic mixing function is setting all the fader levels to achieve a well-balanced result. Obviously, what you consider to be a well-balanced mix is purely a matter of taste, but there are some points to watch out for, and some tricks which can make life easier.
The first thing to do is to listen carefully to the roughly balanced track and find a section which you think can be mixed without moving the faders as the track plays through — ie. a section in which when you do your final mix you will not actually have to move the faders. The longer this section is the better, although it must always include the most important part of the music, be it melody, chorus or hook. If your multitrack can cycle automatically over this section then it's worth spending a few seconds to set the loop points.
With the machinery running the section over and over you can experiment all you like to get the right balance. When you get to a stage where you think you are almost there, consider each instrument individually and, without moving the fader, try to decide whether it would be an improvement if you increased or cut its level.
With the vocal, it's very important to remember that by now you know the song very well indeed, and have already probably heard it more times than any listener will want to in the future. It's easy to fall into the trap of setting the vocal too low simply because you are hearing it clearly in your mind's ear. It has to be clear even to a first time listener, unless of course it's one of the aims of the production to make the listener struggle a little, which is sometimes no bad thing.
When you have balanced this one most important section, mark the faders with a chinagraph pencil so there is no risk of destroying the levels you have set with an inadvertent brush of the elbow. When you play the rest of the track it is almost inevitable that what works for one section won't work for another. The only way to do it is to go over and over each section working out what the best levels are, bearing in mind that you are going to have to move the faders at the right times while the mix is rolling. With my own work, I often find that I have as many as 10 faders with two or three chinagraph marks reminding me of the levels I have decided upon. By the time you have worked all this out, you'll know when to make the movements — but don't take too long a coffee break, lest you forget!
Sometimes it's not best to vary levels by moving a fader — sometimes a slight adjustment of the EQ can achieve a better result, because you are adding level at a frequency at which it is needed, and not boosting non-essential frequencies unnecessarily.
Once upon a time, delay was the effect to use. Would we remember Elvis now without it? Well, we probably would, but even if we don't use delay in quite the same way now, it's a very good technique for enlivening a mix that would otherwise fall a bit flat. A delay unit can produce either a single repeat or, by feeding some of the output back to the input, a continuing echo. Which you use must be determined by the needs of the track.
"In reality, although the mix will hopefully fulfill the producer's requirements absolutely, the recording engineer is usually the driving force behind the creative effort."
Usually, it's best if the delay time is related to the tempo of the track. This wasn't too easy in the days of tape delay, but with the digital processing available in our multi-effects units now, there's no problem in adjusting the delay in millisecond increments until it sounds right, or taking the short cut and calculating it. The simple way to calculate delay times is to divide the tempo in bpm into 60,000. This gives the delay time for one beat, or quarter note, in milliseconds. For instance, if the tempo is 120bpm, then the delay time is 500ms.
If you don't want to use a quarter note (crotchet) delay, then you can get an eighth note (quaver) delay by halving this time. To make the delay more interesting you can divide the quarter note delay time by three to give an eighth note triplet pattern, or by 1.5 to give quarter note triplets. If you want to be more adventurous, try multiplying the quarter note delay time by 0.2, 0.4, or 0.6, to give quintuplet patterns. With two delays, even more interesting permutations are possible. How about repeats which come two-and-three-eighth notes after the original sound? Try it!
Particularly with synthesizer music, it can be difficult to make the whole thing stick together properly. It's too easy to end up with a lot of instruments playing at the same time, but which don't really sound as though they are playing together. What we need is a way to bring the instruments together as an integral unit. One way to do this is to add a common characteristic to each of the diverse instrumental sounds. It doesn't take too long to think of reverb as the way to achieve this. A little bit of the same type of reverb added to all or nearly all of the tracks, including those which you might have thought didn't need any, can work wonders in bringing everything together.
Of course, it's very easy to overuse reverb. You can make the song sound as though all the instruments are playing in the same cathedral, but that's probably not what a recording engineer with any degree of taste would be after, except perhaps for a special effect.
I explained the use of EQ earlier in the Recording Techniques series, but I haven't mentioned the following application yet. A technique which I am told that painters (artistic painters, not decorators) use to pull all the elements of their work together is to apply a varnish over the whole finished painting. Obviously this has the side effects of making the surface shiny and protecting the painting, but the principal function, apparently, is to integrate the different colours and textures of the paint. The sonic equivalent of this can be achieved by passing the entire stereo mix through a graphic equaliser. You may want to use the EQ to add extra depth or brightness, but a few tweaks in the mid range will also help to bring the sounds together by giving them all the same final EQ characteristic.
Argh! Noise again. Even in the dbxed, Dolbied and digitised 1990s, we still have to battle almost constantly against noise. A multitrack recording made with dbx, Dolby C, Dolby S, or even the fully pro Dolby SR, will still suffer from noise. For one thing, every track you put down adds a little extra noise which the noise reduction system can do nothing about. But more than that, what happens when you mix separately recorded tracks together is that every time you double the number of tracks in use you create, in theory, 3dB of extra noise. In practice the amount of added noise might be more or less than this figure, according to the levels at which you use the tracks, but the fact remains that when you play back a multitrack tape you are bound to hear quite a lot of noise among the music.
The only way to deal with multitrack noise is to fade or cut tracks not in use. This can produce a very clean mix with tracks springing into life out of total silence, if done well. A mix automation system will of course do all this for you, once you input the necessary data. Noise gates on every channel are a great help too, but we don't all necessarily aspire to these heights of grandeur (and expense). MIDI controlled muting, as mentioned elsewhere, is also a wonderful noise killer.
Given fewer facilities, there is still quite a lot you can do to reduce noise when mixing from multitrack. The most basic and least automated way is to fade tracks manually. Noise is usually most audible at the beginning and end of the track, when very often not all the instruments are in use. All you have to do is fade in noisy instruments just before they come in at the beginning, and out just after they finish at the end. Sometimes it's possible to subgroup a number of tracks so you can do it more easily on fewer faders.
A simple automated noise reducing technique uses just one stereo noise gate with the channels linked so they both open and close at the same time. If the mix comes to a definite end rather than fading away, then you will probably hear noise when the instruments finish but before the reverb dies away completely. In this case simply sub group all the instrumental tracks and put them through the gate before adding the reverb in the mix. This way, when the tape tracks finish they will be cut, allowing the reverb to continue and decay into silence. Figure 2 shows the setup.
One problem that can crop up with this is that at the beginning of the track, the gate can 'soften' the initial attack. This is solved by starting the mix with the noise gate active but set to zero attenuation when closed, so that in effect there is no gating at all. But immediately after the music starts, turn the control to full gating so that when the track ends it cuts the noise as described.
It takes a lot to produce a good mix, and I'm not finished yet. Next month's installment will cover more essential topics such as how to juggle tennis balls and how not to get in a spin with your spin-ins. Keep those faders f|ying!
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!