Firstly, thanks to everyone who has replied to my previous threads, you've all been very useful.
I need some help with compression. I'm recording my bands new songs, we're a rock band in the style of Metallica/Velvet Revolver.
I'll be using PODFarm for Vocals and Guitars.
I never know if it's best to use the compressor on PODFarm whilst I'm doing the Guitar/Vocal takes, or record them with no compression and then add it in Sonar 8 after the track has been recorded?
What do you do?
What works best?
How much compression?
if you compress during tracking (not in sonar), you can't undo it later. this is not a problem if you are getting good tracks. there's a bit of background to "outboard" compression - in the old days of tape and 16-bit digital, you needed as much signal as you could get without clipping. since compression reduces the dynamic range of the signal, and because 99 times out of 100 you would compress individual tracks anyway, it made sense to do what you could ahead of time and get that much more signal to tape.
24-bit recording gives you the kind of headroom (or more precisely, lowers the noise floor) so that you don't have to worry about keeping the levels "hot" during tracking. you aren't "losing" anything like you would be on tape or 16-bit recording. you have plenty of room to work with and if it comes in quiet, you can always crank it in sonar's 64-bit engine without worrying about artifacts. you can be peaking at -24db and still be fine - because you have so much room to play with.
so when you use outboard compression, you are "robbing" yourself of detail, permanently - and there's no benefit from a gain staging perspective. however, if outboard compression gets you the sound you want to track, then by all means do it. if you know what you are doing you can save yourself time in the mixing stage.
this leads me to my next point (after which i will get back to compression):
do not attempt to target a specific db level when tracking. it is completely unnecessary at 24bit. instead, focus on gain staging - find the volume settings at which your equipment sounds the best. because "louder" always sounds better, make sure you do your comparisons at the same relative volume level. for example, if you turn down your preamp 6db, then raise the volume of the recorded track 6db in sonar. you can use the track trim controls for this as they are pre-FX. if you are monitoring with equipment, turn the monitor level up or down to compensate for gain changes on instruments and preamps. tracking is about capturing the best sound from your equipment, not hitting arbitrary db levels.
the most important thing for learning to use a compressor is to listen to what the compressor does when you play with it. a compressor controls how a track "breathes" - how much signal gets through before you clamp down, how much you clamp the signal and how you let it go.
the attack controls how much sound gets through before the compressor actually starts compressing. longer attack times let more sound out before it gets clamped down. listen to the transients of your recorded sound (pick attack, kick/snare snap, etc.) and see how the compressor attack settings affect the sound. the release controls how long the compressor takes to return the volume back to normal after the signal drops below the threshold.
that is an important distinction to make. if you set your threshold low enough, the compressor will never stop compressing. that actually might be what you want, depending on your knee and gain reduction settings, but probably not. so keep in mind your threshold settings will affect what happens with your release settings.
i usually try to set the threshold just above the point where the signal sustains - in other words, you have the attack, and it falls down to the "meat" of the note, which then decays until it is released. if the threshold is below the level of the note's sustain, it will continue to compress until the note decays enough to let it go. this is not going to be consistent for any instrument except maybe sampled drums or synths - and of course does not take into consideration louder and softer notes. so you really have to take the entire track into consideration when setting your threshold. but that's not a bad thing - it makes you think about how you want to shape the song. for example, you could set your threshold in a moderately loud section so that in the louder section, the compressor is really working (and if set up right, adding to the intensity of the track) and in the softer sections you are getting some of your dynamics back.
the knee and ratio settings shape how the signal is compressed. a hard knee means instant gain reduction above the threshold. a soft knee will start applying compression gradually before the threshold so that by the time the sound hits the threshold the full gain reduction is in effect, and will generally be smoother and more transparent sounding. you may not want smooth and transparent for drums, but you might for vocals.
another important aspect of compression is observing how much gain reduction you are getting. again there's no magic number, but if you are only getting 1-2db of gain reduction, then you are probably not hearing any difference nor are you really doing much about your peaks. obviously it's up to you how much reduction you want but i usually figure on 5dB or more being worthwhile. you can't really get "too much" gain reduction as long as it sounds good, but anything over 20-24dB you might want to check into whether you are simply squashing the crap out of your track.
since you are making the overall signal quieter, you'll need to apply some output or makeup gain (most plugins let you do this) to maintain the apparent volume of the track. this is important because when you compare before and after (using bypass) to see what your changes are, the louder track will always sound "better" and you may miss undesirable artifacts like colored sound or pumping in the quieter compressed track that will only show up once you turn it up. making them the same apparent volume (use your ears) lets you make a fair comparison.
as to what settings you should apply, it all depends on what you want to do. do you want to transparently limit peaks? do you want to get a "fatter" sound? do you want to emphasize the instrument's attack (beater click, pick attack, etc.)? do you want to squash the daylights out of a cloned drum track? think about the shape of the wave that you are trying to control - is there a big instant peak? is the attack long or short? is the length long or short? is the tempo/rhythm of the track fast or slow? all these things play a part.
generally, i go for transparent compression except for bass and drums, since i have DFHS and the drum samples are raw and need processing. on vocals i typically use a very short attack (2-4ms) for a more natural-sounding reduction. on acoustic guitars and drums i use a longer attack (6-12ms) so that the initial part of the transient comes through louder in relation to the rest of the signal. for bass, it depends on the tone - if you want subdued bass, then 0ms attack is for you. if you want growl or fret noise or it's a slap bass, then you'll need to play with the attack to get that sound out. the thing to remember is that you are altering the relative volume of the initial attack of the instrument against the full peak and sustain of the sound.
my release times depend on the nature of the track and the tempo of the song versus how long the signal is above the threshold. usually i start with a longer release and bring it back until it stops knocking down the next note i want to have shaped by the compressor. sometimes that's really short, sometimes it's really long. but i usually start by determining how i want the track to breathe (half-measure, quarter-measure, full measure, etc.), figure out how long that would be in milliseconds (60000 / BPM), and using that as my release setting. if the compressor does anything that virtually guarantees it will not kick in when i want it to, and i can reduce the release time until the track "pops" the way i want.
i will use a hard knee when i want to color the sound a bit and usually on percussive tracks and acoustic guitar. vocals get a soft knee unless i want to color the sound, and with electric guitars it all depends on the tone. some people don't think heavy distortion needs compression, but you can bring out the pick attack more with well-crafted compression settings and get more impact out of it.
you can listen to my stuff (http://www.soundclick.com/jacktheexcynic)
to see a bit of this process in action, but realize that (a) i've learned most of what i know about compression from others here ("yep" specifically, may his cakewalk account rest in peace), (b) i am far too lazy to apply it fully to my own music, because it's one thing to know the process and another to spend hours learning and doing it and (c) sound engineering is not my strong point.
but hopefully what i've said will be of some help.