Peak Levels In Pro Tools for optimal sound quality

[d_matt]

Someone (O.G. Killa) mentioned in another post that one should keep track peak levels around -18 to -14 dbfs in order to avoid clipping on the mix buss.

I haven't heard this before and I'm just wondering if this is a known fact among engineers or one persons opinion, or a golden nugget of a tip.

This also makes me wonder, what is the point in normalizing samples inside a sampler if you're just going to turn them down at least 15 to 20 db inside PT? Does it make more sense to normalize them to peak at say -15 dbfs?

[where02190]

Normalizing increases your s/n drastically, I don't recommend it.

Shoot for average levels of -18dbfs. This leaves ample headroom for transient peaks, wile preserving max bit rate capture.

[tele_player]

Normalizing raises noise and signal equally, the ratio doesn't change, and lowering the levels afterwards pushes the noise back down.

[where02190]

Absolutely incorrect.

[mikeW]

Absolutely incorrect.

When you normalize aren't you boosting both the noise and the signal? And isn't the boost % the same for both the noise and signal?

If so, Tele's comment seems logically correct. I would disagree with Tele if he had been referring hard limmiting but his comment stands to reason with regard to normalization.

Please correct me if I'm wrong...

[thomsbrain]

(i'm asking here, not telling...)

i thought normalization was hard limiting, like with no attack or anything, just instant volume cut. i didn't think it made it louder at all, it just cut off the peaks. am i wrong?

[Rail Jon Rogut]

Normalize finds the peak of the selected audio.. and raises the gain uniformly for the selection so that the peak of the resultant audio is at the peak you've selected for the normalization.

Rail

[Phil O'Keefe]

What Rail (correctly) described is peak normalization. There's also RMS normalization, which raises the average level to a given point instead of the peaks. When using RMS normalization, it is possible to clip peaks.

With RMS normalization, everything gets raised - including the noise and peaks, although peaks can get clipped off if things are pushed too far. For example, if you have program material with a -18 dBFS RMS level and maximum peaks of -6 dBFS and you do a RMS normalize with +9 dB of gain, you won't clip the average but you will be clipping off peaks.

With peak normalization, everything gets raised - including the noise and peaks, but peaks are normally NOT clipped off because most apps won't allow you to normalize peaks beyond 0 dBFS.

[pedrohead]

this might be a stupid question, but how can you tell the dBfs output on individual tracks? i know how high i set my faders, but as far as where the green-yellow-red (peaked) on the individual track meters goes, how do you know where that is dB wise...or am i talking about something else?

[mikeW]

Phil and Rail,
What you guys have stated matches my understanding of peak normalization, only you guys explain things a bit more eloquently.

Oops, I have brown on my nose.