Visualizing compression (a weird take on using slow attack times when compressing a snare drum)

I always had a question in mind regarding snare compression. You may think I'm stupid. I'm a slow learner and late bloomer. I don't judge others so please return the favour haha.

So, compression 101: Audio is too loud -> compressor turns down audio. Compresion 102: There's a couple of parameters you should know about when dealing with compression. - Threshold: when the incoming signal becomes louder than the set threshold, the compressor starts applying gain reduction. In other words it makes the audio less loud. Sorry. This is how I write. Deal with it. - Ratio: determines how much the compressor should react to audio reaching over the threshold - Attack (our protagonist in this blog post): determines how fast the compressor reacts to loud audio. - Release: determines how fast the compressor STOPS applying gain reduction to the signal once the audio drops back below the threshold.

These are not exact definitions. You can find more details about compressors on the internet. I just wanted to give newcomers a glimpse of what we're up against. This blog post was born out of my curiosity to discover what goes on VISUALLY when you start breaking the "rules" of standard compression practices.

This is a snare hit recorded in Cubase:

Snare waveform (50ms highlighted for reference)

If you've never given it a thought, this is a great opportunity. The gridlines help with visualizing this. The transient of the snare (the first, biggest peak) comes within 3 milliseconds of the signals beginning. The majority of the snare sound is in the first 20-30 ms of the signal. After that there's still ringing, sustain, room sound, etc. But the main part we're focusing on is definitely within the first 20-30 ms.

Now, here's an image of a regular instance of compression. It's what you'd expect to see. Snare hits. Audio is "too loud", compression begins.

Standard snare compression, red line shows gain reduction over time

I've always though (and I honestly hope I'm not completely alone with this...), "hmm, what happens, if I make the attack time slower?" But like... a lot slower. Like 25 milliseconds. It's not even that drastic, if you consider mixbuss compression, where you might even go for 30 or dare I say 100 ms, but for a snare, it's quite slow.

So what happens? I always thought (again, yes, I know, I'm stupid for thinking this but maybe there are other stupid people out there) does the compression just happen with a lag? Like it happens, just 25 ms later? And if I imagine the waveform of the snare, will I see a small "cut", or "dip" in the body of the waveform, 25 ms after the transient? Like this?

Imaginary snare compression, delayed gain reduction

Well, it turns out it's not the case. And it's much simpler than this, and I have indeed been stupid all this time. The best way to understand this whole thing is to think of a compressors states. A compressor is basically always in one of two states, at least regarding its speed. It's either "attacking" or "releasing". And when we talk about attack, we need to be a bit more specific. What I mean by this is that the attack time actually defines how fast a compressor reaches the 2/3 of what the total gain reduction should be based on ratio, threshold, input signal and so on and so forth. And this is important. Because a compressor can only be in two states, as I've said. If the input signal is above the threshold, it's applying gain reduction. If the input signal is below the threshold, it's releasing, and lowers the gain reduction.

And so the answer to my question (visualized) is this:

Slow attack time myth busted

The compressor, just like in the previous case, starts compressing straight away. Right when the input level exceeds the threshold. It just does it with a more gentle slope. For this, a simple example would be: Ratio is 4:1, threshold is -10dB, input signal reaches -6 dB, so for every 4 dB of excess audio above the threshold, the compressor only lets through 1 dB, so it "stops" 3 dB. So for a 3 ms attack setting, the compressor goes from 0 dB gain reduction to 2dB in 3 ms (because remember, attack time shows the time to reach 2/3 of maximum gain reduction, which is, in this case, 2/3 of 3, so 2, this example worked better than I intended haha). If we set the attack time to 25 ms, the compressor starts to apply the same gain reduction, but it will take a lot more time to reach said 2/3 (so 2dB) of gain reduction. BUT! As soon as the input signal drops back below the threshold, our dear compressor immediately switches back to "releasing". It doesn't really matter what the release time is. It just matters that gain reduction stops and starts decreasing again. So it won't result in the compression starting at 25ms. It will just start slower, more gradually. But in both cases, the compression will stop at the exact same time, when the signal drops below the threshold. In one case we'll have had more GR, in the other not so much.

So there is no magic way to put a dent in the middle of the snare drum's waveform (actually there is, and you can do it with sidechaining a delayed signal to the snare and using that signal to trigger the compression. If you're interested, ask me about it).

Basically that's it! That's what I wanted to know and it only took me a little bit of research and testing for myself. But I always love finding these interesting questions and answers. I believe this post has become a journal entry as I might have lost any readers by now, but if you're still around, I hope you had fun reading hahaha. All the best to you!