THE "TRIM" OR "INPUT" KNOB ON YOUR MIXER/INTERFACE IS YOUR FRIEND. GET TO KNOW IT INTIMATELY. IT'S THE ONLY KNOB THAT REALLY MATTERS WHEN RECORDING.*
You need to record healthy levels in the first place. If you don't, no amount of tweaking plug-ins will enable you to produce a good clean audiophile recording. No matter what you're using to record (a computer, a portable hard disk recorder, a 24-track analog deck), you're bound by two extremes: you can't record louder waveforms than the media can handle (that creates distortion), and you can't record waveforms that are quieter than the noise floor of the media (that means that the inherent hiss or quantization noise is louder than the music).
Here's an example of a healthy level for an audio signal:
This signal may or may not actually peak at 0db, which is the digital ceiling, but that's OK. The signal is big enough that any quantization noise at the zero crossings will be minimal as compared to the signal itself. Also, if I use a compressor on this waveform, the levels are high enough that the compressor will actually be doing something. (More about that in another lesson.)
Here's an example of the same signal that's been recorded much too quietly:
In order to hear this properly, I'm going to have to boost it, either in the mixing board, or with a "gain" plug-in, or by normalizing it. The problem is, I'm taking what would have been a glorious 24-bit audio signal and turning it into the equivalent of an 8 or 12-bit signal. It's like blowing up a tiny picture you got off the web and printing it out - it's way too pixellated and crappy looking. You're throwing away all of the money you spent on your interface, computer, and software if your audio waveforms look like this. Figure out how to record a healthy signal. If you're using a microphone or instrument plugged into the interface, adjust the input level (sometimes called the trim or gain) until you get a good healthy signal on the meters. A good rule of thumb is that the peaks should be at around -6 db. That gives you room for error in case things get loud. And here's what happens when things get too loud:
This signal is clipped. There's distortion in it that you can hear, and it usually sounds pretty damn ugly. The nice rounded peaks of your waveform have turned into dead flat mesas. The edges of those mesas sound like fingernails on a blackboard. Two hours of such sounds can drive a person insane. Don't do it.
Side note: if you're using virtual synths in your computer (like EZ Drummer), you should "print" all of these synths so that they're audio files, just like your guitars and vocals. That way, if you open the project two years later and the virtual synths are no longer working, you can still get at those tracks and do a remix or whatever else floats your boat. Otherwise, that part is gone forever. It's like losing the lead guitar track because you sold the strat you used to record it.
Anyway, when making the audio files from those virtual synths, the same rules apply for the level of the resulting waveform. If it looks too quiet or too loud, go back and print it again after adjusting the instrument's volume knob.
Another Side Note: Garageband seems to take whatever it is you record and automatically "normalize" it, which means it boosts it until it looks like a healthy signal, regardless of the recording level. This is not a good thing - it's the same as blowing up a tiny web graphic and printing it. The opposite is true as well - garageband will compress the signal to prevent it from distorting. But that compression is uncontrolled, and can sound pretty awful. So learn to set the input levels in garageband while you're recording - it's critical.
THE TAKEAWAY: THE "TRIM" OR "INPUT" KNOB ON YOUR MIXER/INTERFACE (OR THE MASTER VOLUME KNOB ON A VIRTUAL SYNTH) IS YOUR FRIEND. GET TO KNOW IT INTIMATELY. IT'S THE ONLY KNOB THAT REALLY MATTERS WHEN RECORDING.*
*I'm taking it for granted that you're recording your signal "flat"- in other words, no equalization, no compression, and not "after" the mixer board's controls, which would color the signal. This is the way I record. I figure I can always tweak the sound after I've captured it.
back to index