There are basically two ways to increase apparent loudness and the first one is real simple; it's called digital gain change or peak normalization. Peak normalization tells your computer to find the highest peak in your file and raise it by a specified amount relative to OdBFS. Simple, easy, and effective but...when you raise the gain, you'raising everything that's in your file, including low level noise that maybe wasn't so bothersome before, so it's a good idea to try to maximize signal to noise ratio beginning with the tracking stage. This is easy enough if you have a loud instrument, even in a not so quiet recording space, but distant-miking a quiet lap harp in that same room could prove to be a challenge if you have to deal with traffic or cicadas or if your mic or preamp exhibit self-noise. You could record everything direct, but you might find your productions to lack a little 'air' if you do too much of that. You could fix that with some tight artificial ambience from a digital or analog reverb, but that can sound uh, artificial. You can move the mic closer so you hear more of the instrument than the room, but if your mic is cardioid, it will have some proximity effect, which means the bass response will increase as the mic distance decreases. which brings us to method two,
compression.
Essentially, a compressor is a tool that reduces
dynamic range which is the distance between the loudest and quietest portions of your file. But hold on a second, aren't musical dynamics a good thing to have in a performance and isn't that something we want to preserve, rather than reduce? Well sure, but there's a catch; the listener is not gonna be in a treated room-she might be at the beach or in a car. Shoot, the car windows might be down; your low-level passages might be completely lost! That quiet interlude you sweated over for hours has been reduced to "whooooosh". So yeah, compression, baby; bring it on! But like with proximity effect, almost every kind of compression (and there are many flavors) will affect the tonal balance of the material in some way, usually emphasizing bass at the expense of trebles, depending on the settings, but often compression will also enhance vocal sibilance and string squeaks.
So what's a poor studio rat to do? Well I'll tell you what I do; I strap a compressor across the two bus and mix into it. Because bus compression will affect different instruments/tracks differently, we know it's going to affect the balance of levels, so as soon as I have a rough balance I start compressing and adjust the levels as I go and compensate as needed with some judicious EQ. Now for a little detour, I wanna talk about meters; we have meters that read peaks and we have meters that read averages, labeled RMS which stands for "root' "mean" and "square" the method of averaging the electrical energy in your signal or more properly what that energy is
gonna be when your parcel of ones and zeroes gets decoded. I trust you've seen a visual representation of a waveform featuring peaks and valleys, right? The peaks represent transients; they're called that because they're just passing through and they don't last long, but they have a lot of energy; some in fact happen so fast, they can occur between samples-more on that later. These transients are collectively responsible for the punch and presence in your mix, so if you reduce them too much things can quickly start sounding dull, but we want to
reduce them so we can
raise the overall level; this means the peak levels and the average levels will be closer together. Essentially, we're trading headroom for greater average level and that
sounds louder because our ears have a slow integration time; they respond better to changes in average level than to peaks.
Back in the dear old analog days there were so many ways for noise to creep into the recording process that a lot of recordists who maybe had only a paltry few channels of hardware compression available would use them at the
tracking stage as well to keep their signals above the
murk. Man, I miss the murk; it hid a lotta flaws. And a lot of edits. A lot of that juicy murk had to do with consoles, transformers and channel leakage, but in a well-maintained facility most of it was down to the process and the formulation of magnetic tape; without getting too deep in the nanoWebers, I'll just note that simply passing signal through a tape deck results in a smooth, natural compression with subtle level-dependent saturation, slightly increased second and third-order harmonic distortion and a little EQ bump around 90-100Hz. All this analog sweetening came free with your thirty thousand dollar Studer A80, but nowadays we need about fourteen plugins to do all that, considering that those tape tracks were then passed through a second tape deck when they did the mix and yet another at the mastering stage. These multiple stages of compression worked out great (save for the noise and the murk) because it turns out that working a single compressor too hard almost always sounds bad.
How hard is too hard? Well, the very best, cleanest, most expensive hardware compressors in the world can do about 4.5dB of gain reduction before you start to hear artifacts. (and if your monitor situation ain't up to snuff, you might
not hear them and that's
no bueno) A common old school solution is to chain together a compressor and a limiter; the compressor raises the average level and the limiter clips off the peaks. Today we also have software compressors, look-ahead limiters and mastering suites with a collection of tools for your enloudening pleasure. Usually you'll have a linear phase parametric EQ, a multiband compressor, (two or more bands of compression separated by one or more crossovers) another linear phase parametric EQ, maybe followed by a limiter or two. There is also a new third method of increasing apparent loudness through psychoacoustic means that I don't really comprehend, but it works and it's called Oxford Inflator.
For further reading, I recommend Mastering Audio by Bob Katz or this
free resource.
