I did a little searching around, but couldn't find specific answers to my questions. I know there really isn't a definitive answer on the matter. But I'm searching for a system that works for me. So here goes;
Currently I am recording 16 bit and usually run my levels fairly hot so there hasn't too much need to add gain in post (NOTE: I know there is a debate whether to normalize or add gain. I'm comfortable with adding gain to the entire wav file at once and plan to continue.) In the past when I have had levels on the low side I would usually add 1-3dB of gain to boost the playback levels. Sometimes they would exceed 0dB a few times thoughout the recording, but I've never experienced any really hard clipping.
What i have been doing recently (and plan to continue when I begin recording 24 bit) is setting my levels a little more conservatively. Then, performing a global analysis in wavelab 5.0 to check the average dB level and make sure the channels are pretty close (within 1dB). I will check the peak level (for example the peak will be -3dB). From this point I will add 2.8dB of gain to ensure I am still under the 0dB mark. Sometimes i do have a couple spikes in the dB level which exceeds 0dB, but I still add a little gain and never have experienced any clipping that is noticable in playback.
So my questions are:
How do you determine how much gain to add, if any?
How much gain is too much? More importantly, when does clipping become noticable? I've never encountered any audible clipping during playback.
This is really the only type of editing (besided a fade in/out at the beginning/end of sets) I perform. Since most of the shows I record are rock shows, I haven't noticed the introduction of noise. I really like to have the most boost in volume as possible. Should I be concerned about headroom when adding gain in post?
Thanks in advance...