when to normalize?

[nsterken]

Sounds like a means of consolidating and then rendering to a new working track. Not sure with your app but in Sonar for example there is 'bounce to clip- consolidates and processes down to one in place of the existing clip, or 'bounce to track- consolidates to a new track. I believe though if the source clips or track is just muted' it doesn't unload processing it (could be wrong.. we have to 'Archive the old track to free it from processing.


Again to my way of thinking if you have a gain change target it mind there's a half dozen places or ways to do it, and peak normalize would seem to be the most haphazard way to do it. But then I also don't see the utility of the alignment to peaks as a track prep method.
What lets say we have a string of guitar overdubs ready for assembly for mixing. A few happen to have some nice sharp peaks, some others don't. How is aligning to their peaks making the track fall in to your new better leveled' nominal read to mix state?

hmm... i think, not sure, but i think u sort of misunderstood me there...

i'm not concerned at all about lining up seperate tracks to have SIMILAR or shared peak levels. The -10 to -7 dB range I mentioned simply seems to be a good, strong range for my recording signal to provide an adequate noise-to-signal ratio, that's all. So, my process is, to usually record at like -12 to -10 dB strength for the GENERAL level (is this called RMS?? meaning, the average level of the take?) Anyway, with the average level around there, I usually end up with a take that has some peaks and they usually spike around -10 to -7 dB.

so my compressor then reduces THOSE peaks back by maybe only a few dB and then, one way or another, I raise the gain. I was simply speculating whether or not normalising was another way to bring the gain up, although, as far as I understand it, all it's doing is the same thing as raising the fader?? Which I'm already doing, so what would be the difference, one way or another?

All this I just described I do simply after I have recorded and edited the track. I then do these minute processes to simply create a track with a strong, clean, somewhat controlled signal (reffering to the comp here) with a good amount of head room before moving on to recording the next instrument, track, whatever. I, at this point, am nowhere even close to thinking about mixing so my answer to ur question is that I do not know how it may help me mix as I'm not using it as prep for mixing, simply printing a strong, pre-processed take to track, if that makes sense. LOL

So I'm not doing this to get ready for mixing besides getting the tracks to read at a strong, coherent level. Does that help?? Not quite sure how to explain better, sorry...

 

[chuckduffy]

hmm... i think, not sure, but i think u sort of misunderstood me there...

i'm not concerned at all about lining up seperate tracks to have SIMILAR or shared peak levels. The -10 to -7 dB range I mentioned simply seems to be a good, strong range for my recording signal to provide an adequate noise-to-signal ratio, that's all. So, my process is, to usually record at like -12 to -10 dB strength for the GENERAL level (is this called RMS?? meaning, the average level of the take?) Anyway, with the average level around there, I usually end up with a take that has some peaks and they usually spike around -10 to -7 dB.

so my compressor then reduces THOSE peaks back by maybe only a few dB and then, one way or another, I raise the gain. I was simply speculating whether or not normalising was another way to bring the gain up, although, as far as I understand it, all it's doing is the same thing as raising the fader?? Which I'm already doing, so what would be the difference, one way or another?

All this I just described I do simply after I have recorded and edited the track. I then do these minute processes to simply create a track with a strong, clean, somewhat controlled signal (reffering to the comp here) with a good amount of head room before moving on to recording the next instrument, track, whatever. I, at this point, am nowhere even close to thinking about mixing so my answer to ur question is that I do not know how it may help me mix as I'm not using it as prep for mixing, simply printing a strong, pre-processed take to track, if that makes sense. LOL

So I'm not doing this to get ready for mixing besides getting the tracks to read at a strong, coherent level. Does that help?? Not quite sure how to explain better, sorry...

I think you may be overthinking it. Many tracks don't require any kind of compression, processing, normalization, eq or anything else. You don't really need to 'prepare' all your tracks before mixing, and IMO it's not going to give you a better end product.

 

[miroslav]

I was simply speculating whether or not normalising was another way to bring the gain up, although, as far as I understand it, all it's doing is the same thing as raising the fader?? Which I'm already doing, so what would be the difference, one way or another?

Yes...normalizing is similar to raising the track level, except with normalizing, there is always a numerical target which is always first reached by the highest peak in your entire track...so you have to establish those targets...which is messy.
You are better off just *listening*....and raising levels per track to get the best balance without overloading your master bus.
If at that point some tracks are struggling or have peaks that are causing problems...THEN go fix just them...but when you just apply compression, normalization and other broad strokes to initially adjust track levels...you may be doing it for no reason and no benefit...and I think that is what mixsit is getting at when he says it's a haphazard way of adjusting overall track level.

For individual tracks...I get a lot more mileage out of cutting them up in "chunks"...there are usually naturally occurring cut points, or if one section of a guitar part is too low VS the rest, I just cut that part and adjust only that part instead of applying some process to the entire track to fix a few peaks or low points.
This is the beauty of digital editing...you DON'T need to work/apply processing on the entire track like you often have to with analog tape tracks (though there are ways to "cut" them up too, but not as simple as in the digital world).

 

[miroslav]

I think you may be overthinking it. Many tracks don't require any kind of compression, processing, normalization, eq or anything else. You don't really need to 'prepare' all your tracks before mixing, and IMO it's not going to give you a better end product.

DITTO!

Exactly what I was getting at in my previous post.

 

[chuckduffy]

For individual tracks...I get a lot more mileage out of cutting them up in "chunks"...there are usually naturally occurring cut points, or if one section of a guitar part is too low VS the rest, I just cut that part and adjust only that part instead of applying some process to the entire track to fix a few peaks or low points.
This is the beauty of digital editing...you DON'T need to work/apply processing on the entire track like you often have to with analog tape tracks (though there are ways to "cut" them up too, but not as simple as in the digital world).

This is important - pay attention to what miroslav wrote - It works for words, phrases, choruses, bridges, solos - no compression required.

 

[mixsit]

hmm... i think, not sure, but i think u sort of misunderstood me there...

i'm not concerned at all about lining up seperate tracks to have SIMILAR or shared peak levels.

I was really gearing my answers to the general sense of the thread- when to use it, on tracks, clips what have you vs other means.

... I was simply speculating whether or not normalising was another way to bring the gain up, although, as far as I understand it, all it's doing is the same thing as raising the fader?? Which I'm already doing, so what would be the difference, one way or another?

Audio Process-Gain is what I would use (vs normalizing) Its straight forward- you say give me x amount of gain. But all these functions including pre gain' or 'trim, are all up stream in the chain, pre track insert etc. Fader is after (too late fto protect plugs on the track for example.

All this I just described I do simply after I have recorded and edited the track. I then do these minute processes to simply create a track with a strong, clean, somewhat controlled signal (reffering to the comp here) with a good amount of head room before moving on to recording the next instrument, track, whatever. I, at this point, am nowhere even close to thinking about mixing so my answer to ur question is that I do not know how it may help me mix as I'm not using it as prep for mixing, simply printing a strong, pre-processed take to track, if that makes sense. LOL

So I'm not doing this to get ready for mixing besides getting the tracks to read at a strong, coherent level. Does that help?? Not quite sure how to explain better, sorry...

Yeah that's cool. It just like others have mentioned seems like a lot of moves, for dubious advantages and I guess it's splitting hairs, or semantics, it just seems like an odd reason to be slapping on a limiter right off because of some peaks (before you get to mixing as you say?) The one reason (the main one!?) to set up with good head room was to allow for them naturally within the mix and gain structure? (Or maybe I'm the on that's got lost here

 

[DrewPeterson7]

What is the philosophical objection to normalisation that so many on this forum have? It's just a gain tool like any other. If it can get the levels of a file into a more convenient range for working/mixing, then why not use it (subject to my comments above about lowering a file by too much on a DAW using integer processing--but I've had use of floating point operations for more than a decade).

All the normal caveats about working with suitable levels apply of course--but not "good practice"? That's incorrect.

I'm actually curious here too- not because I disagree, but because I want to make sure I understand it right.

I THINK the argument would be that normalization increases the volume of both the signal and the noise floor, and some carefully-applied upwards compression should theoretically be better suited to boost the former while leaving the later unchanged. But I'm guessing here, and I'd love to hear the argument as laid out by someone who actually knows what the heck they're talking about.

 

[nsterken]

I think you may be overthinking it. Many tracks don't require any kind of compression, processing, normalization, eq or anything else. You don't really need to 'prepare' all your tracks before mixing, and IMO it's not going to give you a better end product.

I'm not arguing with you there, chuck LOL

I guess I do it just to kinda shave a few of the highest peaks off as personal preference. I do not really believe its getting me any better product at the end but its kinda the same reason I give the wife a good rubdown before the main course: it just helps us both feel better about the whole experience altogether LOL.

Mostly cuz it's something I prefer to waste my time doing and also cuz I can. I never said logic really had to apply to any of this.

 

[Bobbsy]

Well, to me normalisation and compression are two different (and sometimes complimentary processes. For example, I'll sometimes compress a track then use normalisation as a form of make up gain to get the compressed track into the more linear range on the mixer faders. Are there other ways to do this? Sure--for example the "make up gain" knob on the compressor. However, my way works for me.

Or, as others have said, it's an excellent tool at the end of the mix. Within reason, I mix to make things sound right then normalise to get the levels right. (That's an over-simplification--obviously I have to be aware of not getting into clipping or whatever, but you see the principle.) I produce a lot of tracks (music and effects) for playback use in theatres--and normalising to a standard playback level is almost always my last stage in the process.

I'm not an advocate of over using any process and I counsel against any idea of having a standard way of "preparing" your tracks. My only point is that normalising itself doesn't do anything automatically bad or wrong to your tracks. As with ANY process or effect, it's over using it or using it incorrectly that causes problems. The impression I'm getting is that the biggest objection isn't normalising itself, it's that some people normalise individual tracks to peak at or near 0dBFS. You might just as well say "never push the fader all the way up" as "never normalise".

 

[sixer2007]

This might be running astray with the question a bit, but i'm curious about normalizing for a slightly different reason after reading this... So, I do a bunch of mixoffs where I mix others raw tracks, and I feel like when I import them they are always too loud and I lose all my headroom after just a bit of mixing. To counteract that I tend to take every track and simply take the faders down to -15 or -20. This works about 90% of the time, but now and then I get a signal that I want to compress for example, but the track volume is so low that the compressor can barely react to it, if even at all. If I try to raise that one track's fader, it's totally out of balance. My question is, can normalizing be used to reduce all of the tracks by some specified level? And if so, will it be any different than doing what I have been?

 

[Bobbsy]

Well, normalising tends to reduce (or increase) tracks TO a specific level, not BY a specific amount. It's a subtle but important difference. Instead of saying "raise everything by 10dB", you're setting the highest peak in your track TO 10dB (or whatever). This means that a single high peak in a 10 minute track can keep everything else much lower.

A couple of things:

First, you should be able to make your compressor work at whatever level you need by adjusting the threshold control to the desired point.

Second, as I said earlier, once you've compressed, normalising is a good way to bring the track (now without the random peaks) to the level you want to work at.

Hope this helps.

 

[RAMI]

but the track volume is so low that the compressor can barely react to it, if even at all. If I try to raise that one track's fader, it's totally out of balance.

I'm not understanding this. If you use a compressor as an insert, it would be pre-fader, which means your fader level shouldn't affect it at all. It should be compressing the level of the recorded signal. Also, I'm not sure how a compressor wouldn't react to a signal. What happens when you bring the threshold down on the compressor? There has to be a point where the compressor will start compressing.....unless you have your ratio set at 1:1.

 

[Farview]

I'm not understanding this. If you use a compressor as an insert, it would be pre-fader, which means your fader level shouldn't affect it at all. It should be compressing the level of the recorded signal. Also, I'm not sure how a compressor wouldn't react to a signal. What happens when you bring the threshold down on the compressor? There has to be a point where the compressor will start compressing.....unless you have your ratio set at 1:1.

Just a guess, but not all software compressors have a threshold that goes down to infinity.

But usually there is a way to just turn up the volume of the sound clip in the daw, or at the very least, a trim control on the channel. All of these do the same thing as normalization, just without having to choose a specific peak level.

Of course, this sort of thing shouldn't really happen because things should be recorded at proper levels to begin with.

 

[sixer2007]

Well, for example, I like to use the stock cubase 5 comp now and then, and it's the one that gives me most trouble. Even when the threshold is incredibly low i can sometimes only get it to cut 1 or 2 dbs if my track fader is at -20. and no, i'm not at 1:1. Sometimes this will happen as high as 4:1, so i often must make a high ratio to act on the signal, but i don't typically want to

I do have other comps that has a lower threshold ability though, so maybe this is something I just have to deal with..

 

[mixsit]

Well, for example, I like to use the stock cubase 5 comp now and then, and it's the one that gives me most trouble. Even when the threshold is incredibly low i can sometimes only get it to cut 1 or 2 dbs if my track fader is at -20. and no, i'm not at 1:1. Sometimes this will happen as high as 4:1, so i often must make a high ratio to act on the signal, but i don't typically want to

I do have other comps that has a lower threshold ability though, so maybe this is something I just have to deal with..

Yeah, something has to be outa' whack here- the levels, not shure what.
How about attack and release times? If attack was long anf the peak short it could pass right by (through) with out triggering reduction for example.
As well, is how do you go from 'track is a bit too high ('clips), so you lower it some and it becomes to too low for the compressor is also odd?

 

[chuckduffy]

Yeah, something has to be outa' whack here- the levels, not shure what.
How about attack and release times? If attack was long anf the peak short it could pass right by (through) with out triggering reduction for example.
As well, is how do you go from 'track is a bit too high ('clips), so you lower it some and it becomes to too low for the compressor is also odd?

i dunno - the compressor should be operating on the pre fader signal, or else the whole world would fall apart....

 

[sixer2007]

As well, is how do you go from 'track is a bit too high ('clips), so you lower it some and it becomes to too low for the compressor is also odd?

Well, it's not that the sounds are clipping, I am always sure to avoid that, and who ever records the tracks do a decent job with gain.. It's just that when all the tracks go to the 2 bus, it's up around -5 db or so, but i want it much lower. So, i just drop every individual track by a good amount to keep my headroom. Then that cubase comp wont see some signals. I understand that inserts are pre fader, and this only happens now and then, and other comps can find the signal sometimes.. I guess it might just be that the original gain of a signal was in a good place, but then dropping it by -15 or so throws it out of wack or something. That doesn't explain the pre fader situation though... I'm sure i'm missing something here. Thanks for all the help so far, i'll do some more messing around

 

[sixer2007]

double post..

 

[RAMI]

, I'm sure i'm missing something here.

Don't feel bad. I think we all are. Too much of that contradicts itself or just doesn't make sense. I'm not saying you're not making sense, I'm sure you're calling it as you see it. But something is very weird in all this.

 

[sixer2007]

Right, well, who knows... It's never stopped me from mixing a track. Screw it I suppose. I bet there's some button directly in front of my face that i see right past. It wouldn't be the first time!