Funkybot wroteThese days, I like to keep my levels low-ish. Fun challenge: try to get your levels where your mix already sounds pretty good with all faders at unity. With virtual instruments, it's as easy as doing a pre-mix using the output level of each instrument, but with acoustic instruments and anything else you're mic'ing up or bringing in via a DI, it's not a bad idea to track around the same level you expect to mix at. The added benefit: it makes mixing easy as hell when everything already sounds pretty good before you move a fader.

Now, why even bother? Why not record everything as close to 0dbfs as possible? Well, 1) your converter was likely calibrated to operate with more reasonable reference levels in mind (probably close to -14dbu), and 2) if you're using "analog modeling" plugins, you probably don't want to slam their input unless you're intentionally trying to drive a track into distortion. Remember: most analog modeling plugins are generally calibrated with a reference levels between -18 and -12dbu. If you go hit them a signal peaking close to 0dbfs, then you're going to drive these plugins way harder than anyone in the analog domain ever would (unless they were intentionally trying to distort the signal).

Ideally, you'd do your gain staging on the way into the DAW, but doing it after the fact will still have benefits if you're using analog modeling plugins. So doing the trimming in the DAW or via gain plugins isn't entirely without benefit.

kaosikosmos wroteIf the point is to get all tracks at the same volume, why not just select all audio events and normalise, then drag them down to -18 or whatever you like? Will doing this change the waveform or destroy the dynamics?