Yep lots of the time.You could also automate your compressor!..
Usually threshold, but sometimes attack as well.
Yep lots of the time.You could also automate your compressor!..
Here's the gotchas' in that plan. Yes it can do that -set with very slow attack and release it can pull down on your verse, but being it is set so slow, not respond to the individual words or phrases. Then, slowly ramp back up in level during the quiet section.... For an example, lets say you have a quiet vocal part in a verse and a loud one in the chorus. If you want to make the verse louder, isnt it simpler to use compression instead of automating parts one by one? ..
Which leads to thisHere comes a second question...compression are widely used to reduce the dynamic range of a sound in order to bring balance to a mix. Automation is a manual volume control and its used for the same purpose. How come there be a huge difference between them?
A bit of a nit pick here, but actual compressors don't ever work at a sample-by-sample level. There is always an attack/release mechanism that averages the envelope over time so that the amount of reduction is never exactly tied to the sample that is currently being reduced. If the averaging time gets too short, it starts to sound like distortion, and if there is no averaging, it is not a compressor but a saturator or distorter. You can prove it to yourself with any digital compressor that lets you turn all of its time constants down to 0. Things get grungey quick. In fact, ReaComp is a pretty powerful little clipper/saturation/distortion engine that I have used for everything from mix-bus "warming" to full on guitar distortion.IF you had automation that could operate practically down to a sample by sample range of adjustments, it could do exactly the same as a compressor.
With vocals I'll generally edit/automate the gain of the track, compress it and sometimes automate the volume. Automating the gain evens out the differences in level on a relatively long time scale, sections, lines or single words, which also keeps it driving the compressor evenly from part to part. The compressor also adjusts the gain but on a much shorter time scale that's tied to the signal itself. If the song is about the same intensity all the way through that might be it, but if it has loud and quiet parts I might use volume automation to make the vocal track follow those sections.
On the other hand, I was messing around previously with the idea -'zero attack/zero release might equal no pumping artifacts was the initial thought, which has the flaw in logic you pointed out here.. I did stumble on a very cool successful application. It happened to be on the old 'Cake 'Sonitus comp, but likely many others could be used; But zero x zero' works fantastic as a drum kit/over head or bus smasher! As I recall I could go anywhere from 0' to a few MS release with different effect.A bit of a nit pick here, but actual compressors don't ever work at a sample-by-sample level. There is always an attack/release mechanism that averages the envelope over time so that the amount of reduction is never exactly tied to the sample that is currently being reduced. If the averaging time gets too short, it starts to sound like distortion, and if there is no averaging, it is not a compressor but a saturator or distorter. You can prove it to yourself with any digital compressor that lets you turn all of its time constants down to 0. Things get grungey quick. In fact, ReaComp is a pretty powerful little clipper/saturation/distortion engine that I have used for everything from mix-bus "warming" to full on guitar distortion.
Ok, that.. is a totally new one for me.OTOH, if you have control of RMS time and pre-comp, you almost don't need to use the attack/release controls, or rather they become a bit redundant. When I use ReaComp for "automatic automation", A and R get set to 0. Then, by setting RMS time to 500ms, the reduction happening RTFN is actually based on the average of the last half second of audio. By then setting pre-comp to 250ms, it feeds the envelope detector a quarter second before the audio actually gets to the reduction, so that the reduction is based on the average of the last quarter second AND the next quarter second. This way, the attack/release times are somewhat program dependent, and it's so transparent and natural that it feels like cheating.
Yeah, I got what he meant, and didn't really mean to imply that his point was invalid. Just kind of expanding the point or drilling down to details or making sure nobody else misunderstood it.I'm pretty sure he was saying that you would need that sort of resolution in the automation in order to create the same sort of envelope that a compressor would. In other words the attack slope that would be smooth with a compressor, would be stair-stepped with automation because the volume envelope controls are too course.
0 A/R (and RMS) is exactly full-band distortion. As you make those numbers bigger, you sort of limit the frequencies which actually get distorted almost exactly like splitting the signal into a lowpass and highpass part, distorting the low, and then mixing them back together. IDK off the top of my head how to tell you the cutoff frequency from the time constants. With three different variables, it's got to be kind of complex, and is probably best just dialed in by ear.As I recall I could go anywhere from 0' to a few MS release with different effect.
That's how I understand it should be -or is. What's interesting is it sounds (in my example) very much like the decent sound of compressors you've heard doing the 'smashed drums effect...0 A/R (and RMS) is exactly full-band distortion. As you make those numbers bigger, you sort of limit the frequencies which actually get distorted almost exactly like splitting the signal into a lowpass and highpass part, distorting the low, and then mixing them back together. IDK off the top of my head how to tell you the cutoff frequency from the time constants. With three different variables, it's got to be kind of complex, and is probably best just dialed in by ear.
There are other compressors out there which offer RMS and lookahead parameters, but I only ever use ReaComp nowadays. I suppose I should say that I almost never have the threshold set to actually do anything on its own, I set it right above the top of whatever is happening and then turn the knee parameter up until it digs in where I want it. This way, things are curvier.
Yes it will interpret in between those points, but the compressor will have a complex curve that you don't have enough points to draw correctly. That was more my point.To your point here, though - don't get confused by the time resolution of the actual control points on the envelope and the resolution of the actual control element of the automation engine. Sure, you can only put those points so close together, but the DAW will interpret between those two points to find a point on the line between them on a sample-by-sample basis.
The sampling stair steps don't exist. The idea that digital audio is somehow stair-stepped is the product of an oversimplified graphic that misrepresented the way sampling works.Yes, it kind of has to be "stair stepped" because it's digital, but in a 32-bit (let alone 64-bit) floating point mix engine, those steps can be very small. In fact, a digital compressor's attack/release envelopes must do this also.
Yeah, I got what he meant, and didn't really mean to imply that his point was invalid. Just kind of expanding the point or drilling down to details or making sure nobody else misunderstood it.
The best part of the above is the Times Square Train analogy.
Use the "compression train" to get somewhere near Times Square then use mixing (which tends to be via some form of automation when you're "in the box") to get to exactly where you want to be.
...and that's a hugely common method of mixing a song.