Reverberation - the key to professional recordings - is fully explained. How it's created, and how to use it.
Studio owner Dave Simpson provides an overview of current reverberation techniques and puts the case for the addition of reverb during the recording process.
I believe that I am on fairly safe ground if I begin this article by stating that of all the effects available to the contemporary recording engineer, by far the most important is reverb. It can make a cheap demo sound like one recorded on equipment many times the price; it can make a professional recording sound fresh and exciting; it can even help make Trevor Horn a great deal of money!
The reason for all this lies partly in the fact that professional studios have for some time been designed to eliminate unwanted natural reverb. Before I continue though, it is worth taking a brief look at exactly what the term 'reverb' means.
Basically, reverberation ('reverb' for short) consists of a number of closely-spaced echoes. If you stand in any room and shout, for instance, you will hear a 'ring' directly following the shout. This 'ring' is dependent upon several factors; most notably the size of the room, the reflective quality of the floor, walls and ceiling, and the amount of sound absorbing material present in the room. What is happening is that your voice is sending out sound waves, which are being reflected off various surfaces to varying degrees and at various angles. These sound waves are then picked up by your ears. This is what we call the room's 'ambience'.
Such ambience has the effect of softening the hard edges of whatever sound is being produced, often resulting in a pleasing effect, especially when used in conjunction with a musical instrument or the human voice. It is no coincidence that many of us only sing in the bathroom - the room in the house with the greatest number of hard reflective surfaces (tiles etc) and the fewest sound absorbing items.
So far so good. The problem is that contemporary recording studio design usually aims at a dead acoustic. By reducing reverberant sound to a minimum, maximum sound separation between instruments can be achieved, the philosophy being that you can always add reverberation (or ambience), but it is very difficult to get rid of once it is there (you can sometimes use a noise gate to eliminate the reverb tail but the success of this method depends largely on the type of signal and the type of reverb).
The reasoning behind the dead acoustic principle operates at three levels. Firstly, it helps eliminate bleed-through of sounds from one microphone to another; the last thing you want on your clarinet track is the reverb from the trombone! Secondly, it ensures that the amount of reverb is fully controllable during every stage of the recording process. What might seem like a large amount of reverb when the instrument is heard in isolation might be out of context when the whole recording is listened to. In addition, complex signals such as a drum kit might demand that certain signals are best left dry (non-reverberant), such as the bass drum and hi-hat.
Given a dry signal, reverb can be added to a recording for two main purposes; firstly, to 'fool' the ear into believing that the recording took place in a specific environment (or environments - most notably those which bring out the best characteristics of the instrument) and secondly, to create a special effect.
All reverberation systems employ some sort of delay line. This can be acoustic, as in the case of an echo chamber; electro-mechanical, as in the case of a spring, plate or tape delay; or completely electronic, as in the form of a digital delay line.
What these devices try and do is to duplicate the pattern of echoes set up as a sound is reflected off nearby surfaces. The echo chamber does this in the most natural way; by using a room with walls of hard reflecting surfaces (such as tiles or mirrors), maximum reflections of the audio signal are obtained. Normally, two microphones are arranged to provide a stereo pickup from a single loudspeaker, usually within an L-shaped room to avoid standing wave resonances. The sound to be treated is projected through the loud speaker; it reverberates around the room and the echoes (reflections) are picked up by the microphones. Reverberation decay time (ie. the time taken for the reverberant energy to decay to one millionth of its original level, -60dB) can be determined by covering the walls with sound absorbing materials, or by use of a noise gate applied to the microphone signal.
The reverb plate, patented and manufactured by EMT, works in the same fashion. Instead of a speaker driving a room however, a moving coil transducer drives a large, thin steel plate suspended in a framework, whilst two asymmetrically placed ceramic units pick up the delayed signal after multiple reflections from the boundaries of the plate (the equivalent of the room's walls, floor etc). Decay time is modified by a mechanical damping system, usually motor driven.
Recently the EMT patent expired and several smaller, cheaper plates have appeared on the market, such as the NSF reverb plate (sometimes known as an echo plate).
The most common method of providing reverb involves vibrating a spring or springs, using, for instance, electro-magnetic transducers to provide drive and pickup. These come in many shapes and sizes, from a single spring in a cheap guitar amp, through multiple spring devices (such as the excellent Great British Spring), to top of the range units such as the AKG BX20, which is a remote controlled, dual channel torsionally driven system providing a decay time of between 2 and 4.5 seconds.
More recently, rack-mounting spring units have appeared on the market. The main disadvantage these suffer is from the obvious restriction of spring length, which serves to limit the decay time. However, these disadvantages can to some extent be countered by using compressors in the signal path, both before and after the spring, as well as using several pairs of springs.
The most recent development in reverb systems is the digital reverb unit. The input signal is converted into a continuous stream of digital information and sent to an onboard microprocessor. Here, the microprocesser interacts with whatever controls are available and calculates the desired parameters for any given character of reverb. The reverberant result is converted back into analogue form and appears at the output sockets.
The advantages of this type of system can easily be imagined. The parameters of the reverb (decay, pre-delay, frequency damping) are limited only by the quality of the unit. (This is not the truism that it sounds; no plate or spring could provide a decay time of 25 seconds, and precious few actual structures could.) In addition, such devices have none of the problems associated with some other types of unit, such as excessive noise, a tendency to overload and undesirable side-effects (such as 'boings'!).
Digital reverb units can be bought fairly cheaply. The Yamaha and Dynacord units are both very good value, whilst further up the ladder, the MXR 01a is excellent value. Mind you, the top of the range Yamaha, AMS and Quantec units take some beating (if you can afford them, what the hell are you doing reading this? You should employ someone else to read it!).
I can see some of you asking the question; "if reverb merely consists of delay patterns, how come a digital delay line never sounds very convincing when used as a reverb unit?" (The same goes for tape delay.) The answer is that true reverb consists of many different delay lengths; the ear would never be equidistant from all reflective sources. (In fact, this is not strictly true - given a large reflective surface such as a cliff and a sound source in the open air, such as a mammoth breaking wind, a single repeat would be far louder than any others; this is what we call 'echo'.) Thus a digital delay line can only provide pseudo-reverb because all the delays have a fixed time interval between echoes - this is known as 'flutter echo'.
Right. Let's assume we have our reverb unit. The question is - how best to use it? The problem of adding reverb during a mix, is that the efficiency of the system is determined by the quality and number of reverb units you have available. If you have only one reverb source, be it an elastic band or the top of the range AMS unit, if you add it in the mix, every signal will be subject to the same parameters.
Given a decent, stereo digital reverb, most people can live with this problem. However, with a spring device, not only will certain signals suffer (every reverbed signal tends to be something of a compromise), but the strain of putting a whole multitrack mix through the spring can often overload the unit, resulting in distortion.
Even if you are lucky enough to own a stereo digital unit, you quickly discover that the complex signals of a modern multitrack mix demand many different types of reverb. For instance, synthesisers can sound good with a short decay time (say 0.3 of a second) and a reverb frequency boost of about 9-10kHz. Strings, be they real or synthesised, can 'sing' if long decay reverb (say 2.6-4 seconds) with a mid frequency boost of 1 kHz is added. Drums, especially toms and snare, should be tried using short decay times (I find around 0.7 sec useful). Slow songs though, can handle reverbed snare drums and tom-toms with decay times way beyond this - four seconds or more (listen to 'Mandy' by Barry Manilow - after you've delighted in the way reverb can enhance a snare, take another listen and be amazed at what it can do to the human voice).
How then do you get round the problem of having only a single reverb unit? Modern thinking tends to indicate that the solution lies in recording each signal with reverb, rather than adding it at a later stage. The ambience may be natural (witness the 'live end - dead end' studio design currently in vogue) or it can be artificial, supplied by a reverb unit. Given modern multitrack techniques, it is perfectly possible to record every instrument one at a time and utilise just one reverb (provided it allows enough variation).
This system does have the disadvantage mentioned previously; namely that in the final analysis, the wrong amount of reverb might have been irrevocably added. However, and this is in my view crucial, it does mean that the correct type of reverb is added. Every reverb signal can also be carefully tailored to suit the dry signal.
A further advantage is that the overworked resources of the mixing desk and outboard effects can be harnessed during recording stages to supplement the reverb unit. In practice, this means that even a cheap reverb unit can be made to sound highly acceptable. For instance, the reverb return signal could be brought into the desk via one of the input channels (which might normally be in use during mixdown) and thus EQ could be added.
The cheapest noise gate (and at around fifty quid for an Accessit unit there is no excuse for not having one - they can be amazingly useful - see Dave Lockwood's excellent article in the December issue) can determine the amount of pre-delay (the time gap before the reverb cuts in) and decay; two factors which can be so vital.
The famous 'gated reverb' drum sound (pioneered by Hugh Padgham) is achieved by sending the reverb signal through a noise gate and triggering the gate by the drum in question, say the snare. The noise gate is set so that it opens and lets the reverb signal through a fraction of a second after the drum is hit. This 'pre-delay' both prevents the original signal being swamped with reverb, thus increasing sound clarity, and deceives the ear into believing that the 'environment' in which the drum was played is larger than it actually is. (In really large rooms, there is a slight delay - between 10 and 100 milliseconds - before the reverb becomes audible).
At the same time a fairly fast decay time is selected (usually under 1 sec). The fact that the gate is determining the decay time of the reverb means that the reverb unit can be driven to its maximum. This ensures that a large amount of reverb is fed into the noise gate, which eliminates the long reverb tail which would normally ensue. This effectively prevents the final signal from becoming cluttered as the reverb tail from one drum beat overlaps the next beat.
If, however, you do not own a noise gate, or if you have one with a fixed attack time (such as the Accessit), any digital delay line will provide the necessary pre-delay.
Send the pure reverb signal through the delay, set the delay output to 'delay only' and return the signal to the mixing desk. By adjusting the delay time, you can alter the amount of time the reverbed signal takes to follow the original signal, giving instant predelay!
Studio owners who would rather spend money on their Capri 2.8 souped-up cars than on a noise gate, could try limiting the amount of decay on the reverb by switching the reverb channel in and out as required. (I'd rather fork out for a noise gate myself and drive a Mini.) Alternatively, many reverb units nowadays possess some means of adjusting the decay time; some also contain facilities for adjusting the pre-delay. So you could always trade in your steam driven model!
A further advantage of recording with reverb, is that a stereo reverb image is produced even if the reverb unit is mono. What happens is that you end up with reverb recorded on tape; in the final mix, when the signals are panned, the reverb will, of course, be panned too. This is not actually true stereo reverb, but it is still better than a single mono reverb source in the centre of the stereo spectrum.
Having extolled the virtues of recording signals on tape with reverb, in practice I tend to opt for something of a compromise; I leave two types of signals 'dry' - strings and vocals. This is because they both often require the same type of reverb so that I can set my unit to accommodate them both during the final mix. (For vocals, try 1-1.5 sec decay, 40 milliseconds pre-delay and 5dB off the top end - say 10kHz.) In addition, vocals are especially critical to a final product, and I do like to retain a degree of control.
As this indicates, I do still use a master reverb device over the final mix. It does mean, however, that if any given signal was not recorded with enough reverb, a further amount (albeit of the same type as the vocals) can be judiciously added.
The main thrust of this article then, is my belief that you should at least consider recording some signals with reverb. Given a reverb unit and perhaps a noise gates fascinating and professional sounding effects can be achieved. (One I forgot to mention - set the noise gate for a slow attack and very fast release for reverse reverb.) If you are still unconvinced, re-read the article and then do three things. Firstly, analyse your reasons for objecting; are you sure they aren't based on what someone has told you happens in 'big' studios? Secondly, give it a try - you might be impressed! And lastly, next time a guitarist or keyboard player records with you, ask him if he'd rather record with or without reverb and just see what he says!
Having just mentioned recording vocals dry, here's a final tip which might prove helpful. If a vocalist wants to hear their vocals with some ambience in order to help them sing, but you do not wish to record the reverb, you have two choices. You can route the reverb return to a spare input channel (instead of the vocal channel) and give them foldback from this channel. Alternatively, if you return the reverb via the returns in the desk, route these to a spare output channel and put the off-tape monitor from this channel through the fold-back; either way the vocal track will remain clean, but the foldback will contain reverb.
That's about it. The next article will contain details on vocal technique; how to speak with a Weetabix strapped over the microphone (commonly used in the clubs), and how to make money from your old, blown speakers (sell them to British rail for use in railway stations)!