Last week, I talked a bit about the intimate relationship between the threshold and the ratio. It seemed pretty straightforward, and it was. However, this week, we are going to investigate more deeply parameters that still seem misunderstood, even for sound engineers. Of course, attack and release times are less obvious to the human ear than distorsion, for example.
Why attack and release time matter ?
Attack and release times play an important role in the quality of compression [Zolzer, 2008]. It is true that attack and release set to zero will distort more easily. Also, in that case, the compressor doesn’t care about the feeling transported by the waveform; Mechanically, it will simply chop off everything by half (for ratio 2:1) that will trespass the threshold. Like Bootsie said, “the magic is where the transient happens” [Variety of Sound, 2009], if true, which is, therefore it cannot be systematically cutoff whatever it is supposed to express. For those who don’t know what transients are, let’s say peaks.
What are the challenges one might face while adjusting these parameters ?
First, monitoring and acoustics will play important roles on those adjustment. The fidelity of the speaker will play on the attack time, while the acoustic of a room will play on the perception of release time. Why so that ? Let’s go deeper into that:
The attack felt is directly proportional to the ability of the woofer to reproduce the dynamic. If you’re woofer is made of heavy materials like cardboard, the speaker will respond slowly compared to the one made of kevlar. In other words, if the dynamic of the speaker is slower than your attack setting, you won’t hear any difference.
The release time is hard to hear if you room is very echoic. Why is that ? Because the reverb of your room consists of an amalgam of delayed sound. Therefore, more the reverb level is close to direct sound in terms of power, more you brain will take into account of past events rather than instanteneous ones. It is then a fact that dead environment is better than a live environment for compression release setting.
Theoretical background
First, what are the attack and release time ?
Attack time is the time it takes to the compressor before reaching the gain reduction it should apply once the threshold trespassed. Similarly, the release time is the time the compressor will continue to apply the gain reduction after the signal get back under the threshold.
Where is that coming from ?
The attack and release time are originally coming from the analog domain. Since feedback designs were used, the gain reduction applied by the compression was based on the information that came in few milliseconds before. It’s funny to see in the Altec 436 manual, which has fixed attack time of 50ms was considered as a “fast attack” compressor. Nowaday, it would be considered as slow.
How to adjust them ?
Attack time:
I usually start by setting the attack time first since in many compressor design, the release time is function of the attack time. When listening to the effect, you have to focus your attention on the beginning of the peaks. While a zero attack will brickwall the peak, a little longer attack will let it pass a bit. Now it’s a question of taste which also depends on the particular situation. But, a lot of people like to let pass the attack of high dynamic instruments. For drums as an example, letting the attack pass a bit before compression helps to make the compression more transparent, since the hear still feel the punch in the dynamic even if the rest of the curve is compressed. Voice also gains in having long attack time, since the consonants can pass a bit like percussive sounds. On the opposite side, some sounds like slapping bass, with over exagerated slapping noise will gain in being entirely compressed. Same thing when trying to deess (remove harsh “sss” sound in a voice). In other word, if the impact is desirable: go with a longer-than-zero attack time. If not, if the impact is annoying: cut it straight away with near zero attack time.
Release time:
Release time is often used to minimize audible distorsion. The distorsion phenomenon occurs when is squared by the threshold almost like clipping. This occurs since the samples just under the threshold have almost the same values as the one just above and the ear interprets it as if it was a continuous square wave. By adding a release time, we are pushing the data close to the threshold a bit away, so the ear doesn’t hear it at the volume.
So, according to that explanation, when release time is needed ?? Long release time is particularly needed when the overall volume is close to the threshold. Otherwise, if it is an almost instantaneous huge peak and the rest is really quiet, a very quick attack time would do the job without pumping artifacts. Longer release time than required will translate into pumping effect, which is, in most of the cases, undesirable.
To conclude, I hope this article has been exhaustive enough, please do not hesitate to leave your comments or share your ideas. You can like the www.quantum-music.ca facebook page or subscribe to the RSS flux to get news feeds.