I was told today that slightly compressing an instrument early in the chain, prior to moving the sound through the soundcard and onto a hard disk was a good idea--that it would allow more of the sound to reach the final recording.
In other words, if I want to record a guitar track or sample a note, for that matter, it\'s better to use a little compression at the mixer level, and then send the compressed sound through the soundcard to the hard disk, instead of recording the raw sound and then applying compression.
If you\'re going to the hard drive from a sampler, you generally don\'t go through the soundcard. For instance, when you tell Giga to \"capture to wave\" the bytes go directly to the disk, just the way Giga created them. Assuming that you\'re not overloading things and going past full scale, it doesn\'t matter if you compress in the Giga mixer, or if you do it later in, say, Vegas or some other DAW.
That said, compression is often applied early, before EQ and reverb. For instance, it\'s best to apply \"crunch\" to an electric guitar signal, before feeding the other effects. Doing this in general can help tame an overly dynamic track before you do any other work to it.
I use an analog compressor when recording vocals or a direct input bass guitar, so I can keep the signal hot, and not go over full scale with P-pops (vocals) and thumb strikes (bass). Maybe your friend was talking about this - live recording. In that case the sound does go through your soundcard then to the hard drive. In this situation an analog compressor (I use the Behringer Composer Pro) is the way to go. Once you\'ve gone past full-scale and digitized the peaks of the signal have already been clipped and lost forever. Digital compression in this case is too late. The idea with the analog compressor/limiter is to tamp down the big peaks before they get digitized by your soundcard.