Topic: why do you normalize your recordings?

[deadheadcorey]

SSIA,

Trying to figure out if I should normalize by recordings or not...

[ashevillain]

It might be a good idea for people to also specify which type of normalization they are using and why. (i.e. RMS vs. Peak)

I use peak normalization. Most of the time I don't want to pay close attention to my levels so I set it conservatively and forget it. I don't like to ride the levels during the show. Since I'm recording in 24 bit and setting levels low I normalize to get essentially what I would have got if I'd run the levels to peak exactly at 0dB (running this way eliminates the risk of clipping and gets you the loudest possible levels while preserving the dynamics of the recording).

The only situation this doesn't work as well is if there is that one or two times the drummer (or anyone really...but usually the drummer) hits really hard and you get a peak around -1dB and the rest of the show is peaking between -12dB to -6db. Peak normalizing in this situation only will give a 1db increase across the board. In these situations I will use some compression (but that is a whole other issue).

[twatts (pants are so over-rated...)]

It might be a good idea for people to also specify which type of normalization they are using and why. (i.e. RMS vs. Peak)

I use peak normalization. Most of the time I don't want to pay close attention to my levels so I set it conservatively and forget it. I don't like to ride the levels during the show. Since I'm recording in 24 bit and setting levels low I normalize to get essentially what I would have got if I'd run the levels to peak exactly at 0dB (running this way eliminates the risk of clipping and gets you the loudest possible levels while preserving the dynamics of the recording).

The only situation this doesn't work as well is if there is that one or two times the drummer (or anyone really...but usually the drummer) hits really hard and you get a peak around -1dB and the rest of the show is peaking between -12dB to -6db. Peak normalizing in this situation only will give a 1db increase across the board. In these situations I will use some compression (but that is a whole other issue).

Good answer on how, but maybe not why???  Why not just turn up the gain on your playback???  Essentially, all you are doing is turning up the volume, right???

Not picking, but I'm curious if there are other reasons...

Terry

[rjp]

Why not just turn up the gain on your playback???

Sometimes, the playback system has a limited amount of gain available...

[Will_S]

More convenient playback, but also I record 24 bit with conservative levels but releaee 16 bit, so I peak normalize before dithering to take maximum advantage of 16 bit dynamic range.

Although admittedly there probably is not 90 dB+ of SNR on most of my recordings...but got to get that venue noise floor just right.   

[twatts (pants are so over-rated...)]

Why not just turn up the gain on your playback???

Sometimes, the playback system has a limited amount of gain available...

Sure, but that could be addressed with better playback...

Other than the practical convenience of not having to turn your playback volume up, is there any reason to Normalize??? 

For me,  I don't bother.  I record at a volume that is loud enough for my listening/playback, and I don't really want to take the extra time in processing in the PC if I don't have to.   

Terry

[Shadow_7]

I record in 24 bit and run the levels conservatively.  There's always applause or other localized noise that makes the peaks a bit hot.  So I hard limit those down a bit and add gain to the whole.  I'm not sure which type that is, I try to avoid audacity's normalize function as it goes to the peak of each L+R channels separately.  Resulting in mismatched levels on most of my captures.  So I use the amplify function instead, after taming the peaks with the limiter beforehand.  But my current preamps have stepped gain.  And with 8+-ish dB difference between steps, I can't always get it where I want it at the point of capture.  And I generally have to play it safe because of all of the unknowns.  I don't engage the limiter on my capture devices either.  Probably more of a battery life thing than any sort of real concerns as to why.  I just don't, I guess I don't trust them.  Those pesky content changing limiters.  But I also don't record monolithic volume wise rock concerts either.  My situations tend to have a much wider and much less predictable dynamic range.  Where my mics hear better than I do from roughly the same location, even if I have to add 10+dB after the fact.

[ashevillain]

Sometimes, the playback system has a limited amount of gain available...

Sure, but that could be addressed with better playback...

I understand what you're getting at but there are 2 other factors that come into play.

First is the cost factor...yeah I'd like to upgrade the playback systems in my apartment, in my car, better portable playback for walking around and in my boat and vacation house (actually I'd love to even have a boat and vacation house at all!  ) My point is that upgrading this stuff isn't cheap. While it would be nice to never worry about normalizing it's something I do because otherwise I have to crank the volume higher than I want to (and induce more noise).

Second issue is uniformity....theoretically I could leave the volume knob set on each piece of gear and listen to various different shows and never adjust the volume because the levels will be fairly close. In reality this doesn't work exactly mostly because I listen to more studio recordings than live recordings. But at least I know about where to start with the volume knob for live shows vs. studio recordings.

You are right though, it could be addressed with more gain on the playback side. I look at normalizing like adding the cleanest and most uncolored possible gain to a recording. Hopefully my logic here makes some sense.

[deadheadcorey]

Why not just turn up the gain on your playback???

Sometimes, the playback system has a limited amount of gain available...

Sure, but that could be addressed with better playback...

Other than the practical convenience of not having to turn your playback volume up, is there any reason to Normalize??? 

For me,  I don't bother.  I record at a volume that is loud enough for my listening/playback, and I don't really want to take the extra time in processing in the PC if I don't have to.    

Terry

This is what I currently do...

I can't really tell the difference between normalize a recording vs adding gain or riding your levels to get as close as possible to the 0db

[ashevillain]

I can't really tell the difference between normalize a recording vs adding gain or riding your levels to get as close as possible to the 0db

Theoretically there should be no difference at all. However some preamps (such as the SD MP-2) color the signal differently as the gain increases. This is another reason why normalizing could be beneficial. A lot of people don't like the sound of the MP-2 at high gain because it tends to get a little flabby in the bass dept. Run the MP-2 low and add gain by normalizing and I end up with the same levels but not the coloration that I'm trying to avoid.

Plus it takes 5 min to normalize vs. constantly checking my rig during the show.

Forgive me if it sounds like I'm trying to convince you to run conservative levels and normalize. I'm not. It sounds like you are comfortable running the way you have been and there's nothing wrong with that. I'm just trying to present as many angles to answer your question as I can.

[deadheadcorey]

No worries man. I might try what you do and see what I get! Greatly appricate all advice and like reading what peoples input is!! 

I can't really tell the difference between normalize a recording vs adding gain or riding your levels to get as close as possible to the 0db

Theoretically there should be no difference at all. However some preamps (such as the SD MP-2) color the signal differently as the gain increases. This is another reason why normalizing could be beneficial. A lot of people don't like the sound of the MP-2 at high gain because it tends to get a little flabby in the bass dept. Run the MP-2 low and add gain by normalizing and I end up with the same levels but not the coloration that I'm trying to avoid.

Plus it takes 5 min to normalize vs. constantly checking my rig during the show.

Forgive me if it sounds like I'm trying to convince you to run conservative levels and normalize. I'm not. It sounds like you are comfortable running the way you have been and there's nothing wrong with that. I'm just trying to present as many angles to answer your question as I can.

[twatts (pants are so over-rated...)]

I can't really tell the difference between normalize a recording vs adding gain or riding your levels to get as close as possible to the 0db

Theoretically there should be no difference at all. However some preamps (such as the SD MP-2) color the signal differently as the gain increases. This is another reason why normalizing could be beneficial. A lot of people don't like the sound of the MP-2 at high gain because it tends to get a little flabby in the bass dept. Run the MP-2 low and add gain by normalizing and I end up with the same levels but not the coloration that I'm trying to avoid.

Plus it takes 5 min to normalize vs. constantly checking my rig during the show.

Forgive me if it sounds like I'm trying to convince you to run conservative levels and normalize. I'm not. It sounds like you are comfortable running the way you have been and there's nothing wrong with that. I'm just trying to present as many angles to answer your question as I can.

Ahhh!!!  Thanks, this is the kind of answer I was looking for...  I can see the difference, though I suppose you would (theoretically) have to take into account any subtle change presented by the Normalization process... 

I might try your method and see what kind of results I get...

Thanks!

Terry

[Shadow_7]

Well...  In terms of conversation content, running levels low might truncate (dynamic range) those softer sounds of little consequence.  As long as the speaker always stays above the cutoff threshold.  So in a way a sneaky way to get some noise reduction.  For those few times when you have some control of the content being recorded.  Otherwise a bit risky to do live IMO.  And always an option to do that digitally in post.  Depending on how involved you want to get in the editing and file format wars.  Just lots of ways for subtle improvements.  Some of which involve mixing back in the original audio track as they are mostly destructive edits.

I'm tending to do more and more edits in post these days.  Sampling at 24/192 at the point of capture, applying a low pass filter at 18kHz and THEN resampling to delivery formats.  And other little tricks where adding a little bit of gain is a pretty benign edit in comparison.  And I could see where choosing your gain on the preamp would be more relevant than the correct gain at the point of capture.  I had an Art Tube MP when I first started off, and running that near or past 75% gain was pure hiss.  Staying near or below 50% and adding gain in post was much preferred over running that preamp that hot.

[newplanet7]

Hai Corey. Do you run 16-bit? That would be a reason to NOT run conservatively.
I run semi hot because I still run 16-bit. Sometimes I still normalize using soundforge 8 peak normalize.
16 bit is a little more tricky but I still love it, and love my results.

When I run 24 bit I still run hot just out of habit.
Also how long does it take you guys to normalize? Terry?
It's not like it's a huge processing step IMO. Also I understand Ashevillian in that adding gain on playback, depending on what playback you're running,
would add more noise than normalizing in 24-bit??

[georgeh]

I try to set gain levels where i don't need to use normalize, but when recording quieter acoustic/bluegrass i end up setting levels lower for crowd noise. then bring down  the crowd and normalize the rest. I use WaveLab 6 stock normalizer

[andromedanwarmachine]

perhaps pulling back to the core question;

the purpose of normalisation is to present an even and sensible level of program material, either to playback on any system safely, or act upon with further post production signal processes which will not be complicated by dynamic peaks.

As I understand it...

JimP

[ashevillain]

perhaps pulling back to the core question;

the purpose of normalisation is to present an even and sensible level of program material, either to playback on any system safely, or act upon with further post production signal processes which will not be complicated by dynamic peaks.

As I understand it...

JimP

That sounds more like RMS normalization. Peak normalization does not affect the dynamics.

[morst]

I used to raise peak levels in Audacity to just under -0.2dB peaks using the Amplify effect, but now I use hard limiting and sometimes even multiband compression in Digital Performer.

Live shows with peaks near 0dBfs don't play back as loud as studio or other mastered recordings with average levels raised up. I try not to squash the music, but if you're willing to squeeze the occasional snare drum hit or bass guitar fart, you can get a LOT more playback level out of your recordings. Another byproduct of this is that crowd sounds don't stand out as much after limiting, as you're raising the level of the music but not the crud! 

[audBall]

I used to raise peak levels in Audacity to just under -0.2dB peaks using the Amplify effect, but now I use hard limiting and sometimes even multiband compression in Digital Performer.

Thanks for bringing this up.

Can you elaborate on what exactly you're doing when using these?  My limited (no pun intended) experience with them generally ends up in quick confusion.  I have some recordings that could definitely benefit something like this and probably end up taking much less time versus editing each section independently.  I use Audition and it appears to have a HL and MC.

[Shadow_7]

I use hard limiting to quiet the audience clapping if it's not that different from the content.  I also use it to tame the over jealous drummer who has to hit the very last note (and others) 3x's louder than any other sounds in the entire set.  I don't do much of that other stuff.  But I do do a highpass and lowpass filter to ensure that the remaining frequencies fall within the specs of the speakers that I most use.  Depending on the content and location.  And I do that before resampling from 24/192.  Which does wonders for the perceived noise floor.  And the visible one in the waveform.  While most mics spec from 20Hz to 20kHz, they actually pick up and transmit frequencies well outside that range in most cases.  Which, if only in perception allows for those beyond human hearing frequencies to be accurately filtered off when sampling at such a high sampling rate.

[phanophish]

Why not just turn up the gain on your playback???

Sometimes, the playback system has a limited amount of gain available...

This

[morst]

I used to raise peak levels in Audacity to just under -0.2dB peaks using the Amplify effect, but now I use hard limiting and sometimes even multiband compression in Digital Performer.

Can you elaborate on what exactly you're doing when using these?

Barely. I'm just learning but my buddy digtalist0ne has some tricks in his sleeves, and I'm starting to get the hang of it. All plugins sound different, so make sure you actually like the ones you're using. The Masterworks limiter & the multiband are adequate, but it would not be unreasonable to spend thousands on software to emulate classic gear costing hundreds of thousands!

While most mics spec from 20Hz to 20kHz, they actually pick up and transmit frequencies well outside that range in most cases.  Which, if only in perception allows for those beyond human hearing frequencies to be accurately filtered off when sampling at such a high sampling rate.

some people can hear beyond that too. Wonder if your bandpass limiting is audible to them? If we get bionic ears, or direct brain implants, you'll want that stuff back, so save the masters!