don't normalize in this situation?

[mixmkr]

I recorded several live performances onto a DAT and want to make up some CD's. (Pno, ac gtr, upright bass, and three fiddlers). However, I kept the levels fairly conservative so as not to clip the DAT....not to mention I didn't want to try and "ride" a master fadar "live", but rather just set a level and go. I feel a got a really good balance, and the 'live mix' is fine...no problems there...just the volume issue. I had mics on the separate instruments and the violins shared a mic...plus two room mics out about 25 feet for some natural verb. (in a small 200 seat church)

I am not really interested in [obvious] compression at all, but would like to bring up the overall volume for the CD and basically keep the mix as is, dynamics wise, etc.

would you:

1. normalize?
2. edit the MOST offending peaks and then normalize?
3. use something like the Waves L2?
4. do nothing and accept the low relative volume for the CD mix?
5. mild compression and/or limiting?
6. something else?

since this is not really that critical of a recording, (but I'd like to make it the best I can, obviously) my first choice(s) would be to edit the peaks down a tad, and then normalize the 'shootin' match up and/or just run the live mix thru the Waves and call it done.

I can also edit a couple of peaks, and raise the overall volume up in CD Architect with the master fadars. Not sure how that might differ from normalizing, though.

for all the NON NORMALIZING recordists here...what is the answer? What would you do?

 

[Jeyan]

If you have a clip or a 0 db peak amplitude in the mix, Normalizing is not going to to have any effect. So there is no way you can raise the volume without comp/Limit. You can manually bring the peaks down a little and then normalize, which is exactly what a compression is going to do. Heavy compression may affect the sound, but mild compressions sounds very nice and tweak it till you get the sound you want.

 

[Blue Bear Sound]

Limiting is the only correct option for this....

As Jeyan pointed out, normalizing will not effectively raise your level, since any transient spike will limit the number of dBs you might have gained had the spikes not been there. This is the major reason normalizing is so ineffective.

The Limiter is your best friend in this situation....

 

[TexRoadkill]

I would probably put about 1-3db of limiting on it.

 

[mixmkr]

thanks for the replies. In afterthought, I should have left out my option #1 of just normalizing. I realize that a couple of peaks that might shoot up to -1dB would negate just normalizing only. Therefore I asked option #2.... edit those peaks down, THEN normalize.

I guess my question is really this.. what are the opinions of doing this (option 2) VS running thru the Waves ultramaximizer.....which IS basically a brick wall limiter, as you cannot make your file clip thru that plugin?

Is there really a difference between the L2 and normalizing(after peaks are edited)?...providing I am not slamming the L2 and using it like a compressor too

I pretty much like the Waves stuff, and feel the that quality is fine for this, but I am interested in what others might suggest.

I guess in hindsight, a GOOD brick wall stereo limiter "might" have been a better option while I was recording, but I kinda steered away from wanting to put too many variables in when recording. At the time I felt that I am only going to only process ONCE after I jump the DAT to the computer, so I felt the noise floor, and the DAT format of 16/44.1 would be just fine as a starting point. If I felt I was going to get into all kinds of other editing, my choice might have been different.

 

[kylen]

I guess in hindsight, a GOOD brick wall stereo limiter "might" have been a better option while I was recording, but I kinda steered away from wanting to put too many variables in when recording.

I think you were right not to do that unless you could see you were only limiting 1/4 dB every once in a while. In other words barely touching the ceiling every so often ...

Normalizing after performing 'edits' consisting of limiting the peaks would consist of 2 DSP operations (if you're counting). Using an L2 and pushing into it 1 or 3 dB would be a single DSP operation and be quicker and sound real good.

I do it both ways at 24bits or 32bits - then dither if necessary. Shouldn't be a problem. If it is pick the method that gives you the least amount of artifacts. Each wave file is a new adventure !
You gotta use whatever tool and process that gives the best quality and at the same time allows the amount of production you need. Sometimes a balancing act - quality vs speed.

kylen

 

[Timmy2000]

Why are you so worried about normalising? Have a play around with it and see if you notice any difference in quality between your normalised and non-normalised recordings.

My tip would be stick it in to your DAW at a decent level (avoiding any clips) and take it from there. You'll probably surprise yourself with just how much you can get away with limiting-wise before the material starts to become noticeably affected (for better or worse). I think all the talk of how modern CD's are butchered by excessive limiting scares people in to thinking limiting is always a bad thing - in situations like this the limiter's your best mate! It's all about listening to different versions and deciding what sounds best to you. Any sacrifices you feel you're making by doing this will be more than outweighed by the overall improvement to the track.

In all honesty though, if you can hear deteriation on a live recording just because it's gone through 2 DSP stages rather than 1, that must be some recording and you must have some serious ears!

 

[crosstudio]

mixmkr:
i use to limit the peaks by hand so that i can boost the volume of individual tracks before mixing, but the waves and ultrafunk compressor plugins work so well that i stopped.

having said that, as long as you are creating envelopes and riding the faders instead of doing destructive editing, you should be fine.

for each sample (i.e. 48000/sec) the waves plugin is effectively doing two things at once.

//sum the clip, track, trim, and limiter gain values. if the total volume
//output is > zero, then make it just below zero.
sampleVolume = clipVolume + trackTrim + trackVolume + limiterGain;
if (sampleVolume > 0) {
sampleVolume = -0.1;
}

 

[punkin]

perhaps not the place but, could someone tell me what "normalizing" is?

I find that after mixing a number of track to a final mix then when exporting for burning to a CD, if I "normalize" all the music on a given CD come out about the same listening level...is this what it's intended for?

sjl

 

[crosstudio]

yup. you've pretty much got it.

normalising means finding the peak and raising the volume of the whole song such that the peak is at 0db.

 

[Blue Bear Sound]

pif I "normalize" all the music on a given CD come out about the same listening level...

Not very likely.... normalizing works on peak values.... listening level has nothing at all to do with peak values. Apparent loudness (what we consider "listening level") is determined by the frequency balance of a song relative to the ear's weighted response to those frequencies.

This is not something that DSP can determine.

 

[Jeyan]

If you are confident in using your compressor, none of this question would arise. All the peaks can be brought down with a little compression (lets say for example ... threshold at -10 and 5:1 ratio) Then you can do a normalize and finally apply a 1 or 2 db limiter if you want. Your signal would be loud enough without any distinct peaks.

 

[crosstudio]

why normalize at all? use a multiband compressor followed by a peak limiter on the mains.

then again, maybe your normalize function is non-destructive. mine is not. so i don't use it.

 

[Timmy2000]

why normalize at all? use a multiband compressor followed by a peak limiter on the mains.

then again, maybe your normalize function is non-destructive. mine is not. so i don't use it.

I don't get it. You're talking about 2 different processes as though they were the same thing. Why normalize at all? To make the recording peak at 0dBFS and still keep the same dynamics of your original recording. Use a multi-band compressor and although it might still peak at 0 dBFS, in terms of dynamics what comes out will sound very different to what goes in.

 

[Blue Bear Sound]

Why normalize at all? To make the recording peak at 0dBFS and still keep the same dynamics of your original recording.

And why is that important? Presumably, you want to do this to gain some level.... but since normalizing works on PEAKS, you could very well have an inaudible transient that is already close to 0dBFS, so normalizing will give you nothing in terms of gain.

 

[Timmy2000]

And why is that important? Presumably, you want to do this to gain some level.... but since normalizing works on PEAKS, you could very well have an inaudible transient that is already close to 0dBFS, so normalizing will give you nothing in terms of gain.

Correct. And if it doesn't have the inaudible transient at 0 dBFS already, normalising (like my post said) makes the recording PEAK at 0 dBFS. My point was actually more about people looking at normalising, limiting and compressing as just different ways of doing the same thing.......

 

[Blue Bear Sound]

And if it doesn't have the inaudible transient at 0 dBFS already, normalising (like my post said) makes the recording PEAK at 0 dBFS.

Typically, that won't happen unless you're recording single frequnecy sounds.... I've yet to see an example of a raw mix that DOESN'T have a transient that would negate any gain-effect the Normalize process might have.

My point was actually more about people looking at normalising, limiting and compressing as just different ways of doing the same thing.......

But it DOESNT get you the same effect at all - like I pointed out, you'll typically get little gain at all from normalizing unless one records their mixes extremely low or is recording single frequencies....... limiting on the other hand, has the potential to provide a significant gain difference (good or bad, at the expense of some dynamics) And the trade-off in that (level vs. dynamics) is a whole other matter entirely.

 

[Timmy2000]

Typically, that won't happen unless you're recording single frequnecy sounds.... I've yet to see an example of a raw mix that DOESN'T have a transient that would negate any gain-effect the Normalize process might have.

So when mixmkr inputs his pre-recorded live mix in to his DAW at a 'safe' level, he manages to get his inaudible transients to hit 0dbSF (or somewhere near there) every time? Sorry man, don't hold with that. I do most of my work around -12 when inside my DAW. When I print to CD I normalise the audio to zero. Not once have I had a mix where when I came to normalise, the amplitude stayed the same. I don't spend my days recording sine waves either........

Originally posted by Timmy2000
My point was actually more about people looking at normalising, limiting and compressing as just different ways of doing the same thing.......
--------------------------------------------------------------------------------

But it DOESNT get you the same effect at all - like I pointed out, you'll typically get little gain at all from normalizing unless one records their mixes extremely low or is recording single frequencies....... limiting on the other hand, has the potential to provide a significant gain difference (good or bad, at the expense of some dynamics) And the trade-off in that (level vs. dynamics) is a whole other matter entirely.
.

Re-read my posts Bear - putting aside the issue of normalising for a second, you seem to be repeating what I say as regards limiting and compression.

 

[kylen]

Once I get in an audio editor I can normalize and move the relative gain of a wave all around the scale if I wish. I may not want the peak at 0dBfs I may want it somewhere else if I'm trying to adjust the relative loudness of 2 different waves. Sometimes 'normalize' will get it in the ballpark sometimes not - if you understand differences in program material, rms, loudness, and fletcher-munson a little then you know why it doesn't always work. But we're not talking about that here - the original post just said how do I bring the entire mix up in overall volume.

I've only had an inaudible transient block the normalize process once - it was inaudible but not un-detectable. I simply hand limited or adjusted the gain of the peak then normalize works fine.

You can substitute the word normalize (meaning the automatic computer process) for volume handle, slider, fader, whatever you do manually to get the job done. The end result is the same thing. I have shifted the entire wave form up or down in gain.

If I choose to use a dynamics processor to limit to a certain peak dB, or compress then we all know that that has an associated sound change with it in most cases (unless you're using very transparent limiter settings) because the dynamics have been changed. The end result here may be different because I am only changing the gain of the portion of the wave that has crossed a dynamic threshold.

The original post said how do I raise the overall volume and keep the mix as is (eg. no dynamics changes). You can either hand limit the peaks a bit, or stick a limiter on there and set it so you don't 'hear' any dynamics changes and raise the gain, or (if the wave has no spurious transient peaks that would incorrectly apply the normalizing feature) normalize the wave to raise the gain a specified amount.

Take your pick - whatever sounds the most transparent is cool for that particular job.

kylen

 

[Blue Bear Sound]

So when mixmkr inputs his pre-recorded live mix in to his DAW at a 'safe' level, he manages to get his inaudible transients to hit 0dbSF (or somewhere near there) every time? Sorry man, don't hold with that.

Well - I can only speak from MY experiences with mixes... I get them around the -6dBFS mark, and so some transients do get close to 0.... so if I *were*to normalize, it would gain me very little.

But sure - if the mixes are way down low, you'll pull a couple dB out of the Normalize process... that still won't help much with respect to apparent loudness, but hey... whatever works for ya!