Nov 16, 2003 - by cbrautHandling the compressor with confidence and dexterity, experiencing and comprehending all of its secrets, from soft knee to hard knee, from threshold to sidechain… this is the goal that we challenge you to through two educative and none-less entertaining episodes.
Compressing a sound source, be it an isolated instrument, a stereophonic mix, etc..., involves reducing its dynamic range. This range, expressed in decibels (dB), represents the ratio between the maximum amplitude and the minimum amplitude of the source. Paradoxically, we often call an abundantly compressed signal dynamic (in the sense of expressing energy), in which case the dynamic range (i.e. the difference between the highest and lowest amplitude) is judiciously reduced. Effectively, upon listening, the linearity obtained imparts a sensation of power. It is understandable that this double meaning of the same word in a similar context can lead to confusion…
|Manual or automatic.|
When an engineer follows the level of a track by ear, by compensating with a gentle movement of the faders according to the variations of level, he is compressing the signal manually. From this come the expressions "following the bass", or "following the voice"... Sure, by the time his brain (even when abundantly illuminated with a good dose of phosphor), interprets the information and transmits it to his hands, it’s a good bet that the sound source will have already started going through the event before being attenuated. If an automated console allows better control, nonetheless technically, man can't pretend to rival the compressor, which for his share, accomplished his task almost without introducing any delay. Like its close cousins that specialize in dynamic processing (the expander and the limiter), our friend the compressor uses a circuit that controls the gain of the signal to process (in this case to attenuate it) with ratios determined by a controlled tension. This tension is derived from the amplitude envelope, either of the signal being processed (most usually), or another signal from an external input (a notion we’ll get to next month).
If the compressor and the engineer that we were evoking, finger on the faders, followed the same objective (controlling the dynamic of the signal), the results would differ. Stupid and brutally, the compressor will blindly accomplish its job without any delay, while the engineer, no matter how smart or experienced, will react somewhat slower, but will follow artistic criteria. The former would be more efficient than the latter in some circumstances (for example, compressing a snare), and reciprocally.
For your information, the dynamic range of the human ear, encompassing the difference between the threshold of audibility and pain, is roughly 130dB. Notice that the dynamic range of a symphonic orchestra, from the weakest to strongest passages, is around 70dB. Coincidentally, for the most, the recording and reproduction mediums (even those of broadcast), are unable to recreate such dynamics. Their range is limited by from one side, the background noise inherent to the medium, and the other, by distortion. For example, a good old vinyl won’t take more than 45dB!
This is to say that before being able to use a disc, cassette, magnetic tape, and so on, it will be necessary to compress the signal for it to hold in the limits of the medium. Sure, the emergence of digital technology has improved lots of things. On paper (but in practice, this number is to be revised lower), the dynamic is equal to the resolution, expressed in bits, multiplied by six. Thus, for the compact disc and its 16 bits, it is 96dB (16 x 6 = 96).
Even with the dynamic range that digital offers, we still tend to compress music a lot. Already, known from the fact that for most, a low volume album is often synonymous with a bad one, the major concern, at least for rock and variety, is to hit the roof: in other words, to get close to the highest possible volume. Another big reason of this leveling from the bottom: radiophonic broadcasting. In fact, as with albums, the stations with the louder volume are more likely retain the attention of the listener frenetically browsing the FM band. Thus, to get the maximum impact, a majority of radio stations are seeking to use their emitter modulation rate in an optimal way: a real race for sound level.
Some of them, notably those diffusing classical, refuse to play this little game. Forget the 70dB of the symphony, which is perfectly adapted to the Opera house, but is less appropriate to listening the hertzian way in your mustang. No, you won't be able to drive your hotrod while fully enjoying the Little Night Music or the Four Seasons. To do so, we would need to savagely compress, deteriorating the music... failing to carry out this massacre, by default, leaves our sad conductor restrained, either to set his transistor to an acceptable volume, agreeing to only hear the fortissimo (as the pianissimos are masked by engine noise), or raise the volume to hear the pianissimo, and in doing so agreeing to have his ears blown away. This is the sad reality of the audiophile driver... Happily, to the great joy of the little ones and grown ups everywhere, the most modern FM stations abuse this process: you can easily scan through your favorite pop stations, without noticing any differences between their levels (not to mention their programs, but that’s another topic altogether…). How not to enjoy the benefits of the technology.
May these considerations not bring us further from our current preoccupation, compression in the studio. From the cassette multi-track to the digital recorder, everybody knows that it is better to prevent breath from the medium (analog) or the quantification noise (digital), and to record as loud as possible without ever exceeding the tolerated level. If not, the signal will be distorted; while analog distortion is occasionally desired, in reasonable limits, by some engineers, digital distortion, better known as clipping, is definitely unacceptable.
If the problem were that simple, you would compress every signal without discernment, which would be the same as limiting it (purely and simply to restrain them from crossing a certain threshold), before recording them... however it wouldn’t take for this approach to lead to disaster. Everything is a question of compromise. Knowing to preserve the audio quality of a take while having respect for its alive quality, without exaggerating its nuances: this is your mission, should you choose to accept it. Everything is also a matter of taste, and certainly varies across musical styles. For example, over-compressing a bass and a drum will probably fit better acid house than cool jazz...
|Before or after?|
Is it better to compress before or after recording? The preceding paragraph gave you a few bits of the answer, since that to keep the quality of the take, you now know that it is indispensable to have enough level, to not flirt with the background or quantification noise from the medium. Then, as for process modifying the recording itself, undoing while mixing seems difficult. Faced with this irreversible character of the applied process before the recording, a good method consists of compressing beforehand, but with parsimony, while keeping the possibility of compressing again after, if needed. Be warned against compressors of middle quality, with which we should take care to avoid using this kind of double compression, as they tend to denaturalize the sound.
|All or parts?|
Excepting the isolated compression of each instrument, before, after, or before and after the recording, it is not uncommon to compress the whole of a mix. While taking care not to go too extreme (in order to keep the blend and balance), putting the piece in a certain range will allow, for one, to optimize the performance of the reproduction medium (the same principle applies to mastering as to recording), and also to follow the needs of the media.
However, it is important to understand the dangers of a stereo compression. Imagine for example that the snare was mixed in front. The compressor, while lowering its amplitude, will simultaneously lower that of the other instruments. Therefore, each time the snare hits, the whole of the mix will be lowered in amplitude. Such settings may be inappropriate (notably the release, which we will examine next month), and will be followed by an unwanted effect known as pumping (figure 2).
If we were to first compress the snare individually, the stereo compressor wouldn't have to go wild like that. All of this to say that the compression of the instruments, taken individually, as well as that of the mix, are complementary: employing the former, to prevent excessively high peaks, will also prevent the latter from overkill, provoking this unwanted effect of pumping.
At the mastering stage, we are confronted with the same problem. Supposing that a brass part's level reaches a little over the other instruments, the engineer will have to either go for audio quality at the expense of the artistic quality (by compressing the whole to optimize the ratio signal/noise, which will, for the time of this brass part, attenuate the rest of the mix), or to go for artistic quality at the expense of the audio quality (by not compressing it, which will, except for the time of this brass part, under-exploit the dynamic range of the medium and lower the whole piece compared to the level of other albums).
Coming soon: the evermore dense part 2 will include a glossary defining the most common terms associated with compression, practical advice, and a truckload of tricks. Stay tuned!
Translation by TritonX