Skip to main content

Hey everyone. I hope I can get some insightful ideas and emotions surrounding Normalization of individual tracks during mixdown. Heres a situation I had to use this feature. I recorded a band live. The board had post-everything direct outs. EQ, Inserts, Fader Movements. So I had my preamps set very low for my drum toms as to not introduce additional noise from re-preamping an already preamped but reduced in output level signal from the Front of House console and also to prevent unfixable input distortion when the FOH engineer boosted the toms for fills. But unfortunately, that's not really the reason. It's what sounded like the best excuse to use to use my normalization. I just forgot to set MY input trim during the set that I was mixing Front of House for. Doh! It sounded good in the house and I completely forgot to monitor my tom input signals and therefore I forgot to boost the toms.

So I am listening to the mix in the studio. The drums sound great, however the tom fills are very weak. The drummer also played very light. So I normalized the drum tracks. Made the toms sit better in the mix.

The questions I pose is:

When you record a signal as close to 0db on the input, will I get an audibly different result if I record a signal at a lower (or much lower) db level and work some Normalize magic on it.

I am recording at 44.1 or sometimes 48k at 24 bits using Focusrite Liquid Saffire, Focusrite Saffire Pro and PreSonus DigiMax D8 preamps. The Liquid Saffire 56 is my AD and D/A interface via Firewire to my Mac Pro Running Cubase 5.5.1

Just wonder if anyone has Normalizing stories or technical data or feeling regarding the use of Normalization during the mix. Normalization of MIXED material is not what I am curious of. I want to hear about normalization as a tool.. not as a volume war destructo-bomb.

Thanks in Advance!

Tavon Markov
Horizontal Experiment Productions

Comments

audiokid Tue, 11/02/2010 - 16:43

Well good question Tavon.

I never stop learning so I cannot give a blanket statement of good or bad for this in the final mix. Something one engineer will do, another would not. We all have our tricks and burdens. I will go as far to say... I definitely don't recommend any normalizing of individual tracks. However, thats me and like I say, maybe others do normalize in certain situations or sounds.

Others here are more technical savy than I but I'll dive in and share my two cents.

Normalizing individual tracks IMHO products terrible undynamic mixes, added noise and boring one dimensional songs. Depending on your DAW and gear, no matter what you do, your mixes will allways sound smashed or saturated if you normalize the tracks in the mix. Long topic from this point on, and a good one that has been discussed many times. If you really want to start learning, start searching about levels from this point on. (I don't even like to touch my fader after it has been printed. But we do have to, but the less you do in your daw, the better in my opinion. I don't ride faders much either. Oh, I'm starting it now lol.... :wink:

Don't try and record every track as hot as possible. Leave room for space and room for Mastering. The space you don't see is the magic in sound. But, it is also noise in less pro recording systems. The finer the gear, the more glorious space is. Space is what I am in search of with DAW.

Welcome to the next level.

Big K Tue, 11/02/2010 - 22:42

There is a lot of wisdom in audiokid's words...
But I define Normalizing as lifting a signal to a certain level ( max- -0dB) without changing the the ratio of the levels inside the clip/track/event/parts...
If you follow some rules it can be a helpful tool for you.
If you cut up a track for editing ( guitar for example) you can't lasso the edited pieces and normalize them individually, because all low volume parts will be too loud, then, compared to the massive ones (unless intended).
You can't normalize overheads, either, when recorded as 2 mono tracks...It kills the balance L/R completely...etc.

Why normalizing? First of all, there is no need to normalize to -0 dB. I never do that... But, I normalize low level tracks to get within reasonable and practicle (best sounding) ranges for compressors and other FXs. Since normalizing does not do anything like compressing or limiting, you can fine-tune the track or event volume during mixing, anytime, without unwanted influences. Normalizing safes you weird settings and you have to raise the volume at some point, anyhow.

Ever since I had to cope with an ( digitally recorded ) o-tone of a movie that arrived at my place at -45 dB, ( those guys messed up big time...) I learned to appreciate normalization on my Nuendo DAW.
After a week of leveling and sorting out a big mess of 30 badly documented DAT tapes and tracks, I was able to obtain a reasonable o-tone sound, even after lifting this all by approx 40 dB. Usually, volume changes on a good DAW on 24 or 32-float material are not doing much harm within a reasonable range (40 dB is not). Up and down and up again, forth and back, is, of course, bad practice. Producing imbalance on live recorded multiple tracks through normalizing would be stupid, too. If you are in doubt, you can always check the max level of all live recorded tracks, and then determine to raise all levels by fixed x dB, to reach maybe -3 dB for the loudest. And, after all, there is something like Undo and non-destructive editing...
At the end you need a good mix and that needs balancing the tracks. If you do it with normalizing, or volume handles, faders or on compressors and limiters is almost the same..
Leaving some headroom and enough dynamic for the mastering house is very welcome, though.

tmarkov1 Tue, 11/02/2010 - 23:31

All very good thoughts. I used it for bringing up the level of the quiet drum tracks to better drive the compressors and gates. Straight ahead. Nothing else. Not for crunching the levels or anything like that. Its just that I have heard about different things being done along the mixing path that can be the detriment to a mix. Stereo Interleaved files as a mix master being one. Never have I heard any professional criticism of normalizing the kick drum track because it introduces noise or drops the bit depth or creates another generation through dither. All good insight though. Hope more people comment on this.

soapfloats Tue, 11/02/2010 - 23:49

K -

Would using the Audio>Process>Gain function in Cubase/Nuendo achieve the same effect of normalization for your purposes?
Often a player/vocalist will change levels drastically between takes. If it's a good take, I do not want to stop it and would rather amplify it in post.
I am guessing that my process is similar to yours? (Rather than Normalizing, I am adding gain at a set % or dB)

In either case, it seems the noise floor would also increase. At which point I either find another way to get the track to a better level (rerecording, duping or fader boost w/ surgical EQ) and/or use the pencil tool to bring down the "noise-only" sections.

The pencil method is great if you have a lot of bleed.
In one case, I recorded a 5 piece, all live, and w/ little isolation.
The vocal tracks were (of course) awash w/ drum noise.
A little careful use of the pencil tool (not too low between phrases, or the ambient drum sound changes noticeably!) to follow the vocal lines made a world of difference.
Yes, it was a major PITA, and correction for something that might have been prevented in the first place (as others have alluded to regarding normalization), but I think it offers an alternative to normalization.

In the end it's all about gain-staging and knowing where your levels need to be.
Still, we all will have to deal w/ low-volume audio from time to time (be it our fault or another's), and any method is worth a look, IMHO.

Boswell Wed, 11/03/2010 - 04:03

There's little point in normalizing quiet tracks in a multi-track recording prior to mixing. It's an unnecessary step that does not win you anything. Take each track as you have it recorded and adjust the level of it in the mix. This may mean (as in your case) giving it a fair amount of level boost at mix time.

For your drum fills, use automation to increase the level of these above their general level at mix time, or do it manually if your mix process has no automation.

Big K Wed, 11/03/2010 - 06:28

Hello Soap..
No, gain is not the same function as normalizing. Gain adds or deducts the set value to all marked events the same, no matter if they have a peak level of -2 dB or -15 dB.
Gain is handy to use with a number of tracks or events that need to keep their natural recorded volume balance between them, for example L+R overheads.
Normalize increases or decreases the levels of the chosen events or tracks* individually by the entered value (plus or minus) from x dB up or down, but to a maximum of -0 dB.

It is a drag to record in an environment where there is a noisefloor that becomes disturbingly audible when adding some 10 or 20 dB. In that situation and with no chance to get rid of the noise, I would prefer to try intelligent gate-ing and noise reduction per plugin, first. With an adequate automation feature in your DAQ, gain riding is worth trying, too. Try some restoring software, with Algorythmix ReNovator at it top ( my personal opinion...) With ReNovator you can inaudibly erase any unwanted noise in a recording. I tried myself at mastering those stereo tracks to believe: a dropped bunch of keys or clapping chairs out of a classic philharmonic recording, a motorbike pasing by during a church choire recording and completely erasing bass drum or snare from a Rock CD for replacement without artefacts or influencing the wanted signal. Of course you can also achieve better separation between ..say.. brass instruments in a life recording, quite easily.

* with normalizing, the tracks must be in one piece! In cut-up tracks the normalize function works on the individual events, producing incoherent levels of the instrument / voice.

Hey, tmark..
Youp, this danger might exists on some DAWs (was it those with fixed point math?). Therefore, better check the max level dB and don't add gain above what your DAW can cope with.
For your tom track, either way would be fine. Normalizing, as you did, is the faster and maybe safer way. If there is too much bleed from the other drums, cut the tom track up and edit out the time between tom hits, but give it decent fades to blend smoothly in and out.

A little hint: If you have the option to use the broadcast wave file format....Use it! If you move any events or tracks by accident, not knowing when or which, and you have no fast way of re-adjusting, use the "place at original time" feature many DAWs have. This will cause the chosen events / tracks to snap to their original recording position. Even if you have moved the whole project in your project window, at least, you have them aligned accurately to each other and placing them back is easier to do as grouped events then one by one...

BobRogers Wed, 11/03/2010 - 10:22

tmarkov1, post: 356021 wrote: ...When you record a signal as close to 0db on the input, will I get an audibly different result if I record a signal at a lower (or much lower) db level and work some Normalize magic on it....

Yes. The closer you record to 0dB the lower the signal to noise ratio. This used to be a very big problem with tape, less of a problem with 16 bit digital, much smaller problem with 24bits.

BobRogers Wed, 11/03/2010 - 12:46

Before I comment on the poll question let me explain the way I think of the difference between gain and normalization. If I have misunderstood this or mistate it please correct me. First both gain and normalization involve multiplying a signal or collection of signals by a constant factor: if you want a waveform with twice the amplitude you multiply each sample by two. In practice, the factor can be any floating point number between zero and saturation. The difference is in how the factor is determined and applied. In gain the factor is determined by the position of a fader (or the equivalent) and is applied to each sample as it is played back. Normalization looks at a collection of samples (a track or region), determines the largest amplitude sample, and computes the factor that will convert the largest sample to a desired level. It then applies the factor to the entire collection of samples at once, changing the level of the entire collection "permanently." This is complicated by the fact that if you use a normalization tool on multiple tracks or regions you can tell the tool to apply the algorithm to each region separately or to the aggregate.

So I voted for "nondestructive" because normalization really is really a simple multiplication. But I agree with the points above: applying gain is easy to understand and easy to undo. I've never encountered a situation where you could not do the same thing as normalization several other ways. You could have achieved identical results by pushing the faders on the toms higher. Of course, this points to the (small) advantage of normalizing the tom track to have about the level as the other drums. If all tracks have the same base level then your faders give a nice graphic picture of the mix. I've had a few times when I have normalized for one reason of another, and I've done it without any worry. It's just another level tool - occasionally handy - never, in my experience, necessary.

Note: Multiplying amplitudes and adding dB (as referred to by Big K) are, of course, equivalent. This is what you get when you let a mathematician on an audio board.

Big K Wed, 11/03/2010 - 14:34

Oh Bob, where have you been when I needed you the most... during my school days.
Math was appreciated by me like a big fat black spider in the corner...
I still wonder who got me that B, in the end...lol...

Now, to clarify that: what do you mean by "between zero and saturation" ?
Saturation is another animal..., although I would appreciate it very much if digital would saturate instead of distort nastily. Btw, it is not (tape) saturation what everybody is after. Tape saturation is what generations of audio engineers have always tried to avoid and complained about. It is the tape compression they wanted, but somehow the terms got all mixed up and with this MP3-Generation saturation sounds just as good as compression... awful..., but they would not know that...

What you described under normalization is not corresponding to the normalization function of Cubase/Nuendo, but can be true for other DAWs. That, I do not know.
Here, normalization checks the peak levels of all the regions you marked and increases or decreases the level of each marked region or event individually to the set level. A feature to apply the same gain to all regions alike is not what I understand under normalization. It is this complication, as you name it, that I want: applying different dBs ( + or -) to individual regions and obtain a peak level unity of all regions.

Normalization comes in handy if you, e.g., have tom tracks, unevenly played and you want them all equal. Cut the track up, select all events and apply normalize to -2 dB.. safely done... All the hits have a peak level of -2 dB, now (fades or crossfades might be necessary, but can be applied automatically and fast).

The gain function of cubase and nuendo is also a tool like normalize function that is applied permanently..done once..no danger of accidental altering, anymore. No danger, at all, anyhow, because Nuendo 5 has unlimited undos were you can undo applied treatments even if it is 250 lines of changes down the row and it is a 32-bit float engine with non-destructive editing. If you mess up, just get the original recording from the pool...

Of course, all that depends on the personal workflow and tasks. Some are never using Normalize and some use it regularly....

audiokid Wed, 11/03/2010 - 14:57

great topic and expected.

It has been my experience that normalization in older pro tool and I'm pretty sure its this way with all the DAW, (better or worse) is not a good thing or necessary most of the time.. It boosts the levels yes, ( and I agree with Bos ) but to me, severely downgrades the openness (which I call space and hear). I blame the crap sound of today's music for this over used and over rated function. Even lifting (normalizing) audio to - 10 on separate tracks, I can hear it close in and loose space. ( it does something to how the music all gels together) Do that to all your tracks and you are IMHO, making your music sound more one dimensional. Yup, loud and maybe more even, but even isn't necessarily a good thing. I know, math will tell you different. I'm just saying get it right and don't record so hot. Leave room for space. Especially for the 2-bus.

I use Normalizing in Sequoia for the master CD and that sounds killer. It seems to put the entire mix right where I want it, but I save it for last. I could totally change my mind though, because I'm never too sure I fully understand it until the final hour of each and every song.

Big K Wed, 11/03/2010 - 15:59

With all due respect...I do not agree, since I can not hear any significant or better non at all difference between original and normalized signal other then the volume change.
It is nothing else then adding or multiplying numbers and that is the computers domain, no doubt. If it can't do that correctly, then ...good night... No processing and no plugin can be trusted, anymore.

I will try the following: I take a track of well recorded audio or a CD track with decent dynamic at -15 db and I will normalize this track to -2 dB.
Then I will normalize it back to were it was and do a phase cancellation test on it. I do not know what the outcome will be.. I will report..might take some time, though.

We are NOT talking about getting anything loud as it gets, but bringing audio levels to a usable and technically better level for further processing without the use of compression or limiting.
You can just as well use the track volume fader and automation for this, but it is much more complicated and often subject to errors and accidental altering of the settings.

We do totally agree upon leaving healthy headroom and dynamic to tracks and music and getting it right is our trade ...but it is not always the 1st priority to the musicians you record.
;-)

BobRogers Wed, 11/03/2010 - 16:04

By "saturation" I meant the largest number producible by the bit depth 2^16-1 or 2^24-1 depending (0 dBFS in either case). In fact no faders go up that high for good reason - you'd clip anything that isn't the lowest possible resolvable signal. It's confusing and I should go back and add a correction.

Pro Tools offers you the option of normalizing region by region or normalizing as a batch so the normalizing factor for all regions is the same.

audiokid - Of course, I don't know what the actual algorithm is, but the way I recall it being described in the PT manual normalization is just a simple volume boost - the same that would be achieved by setting a fader to the correct level to get the peak to the correct level. I have not used it that much, but I have not heard the artifacts that you have heard. If I am misunderstanding the algorithm I'd appreciate someone correcting me.

BobRogers Wed, 11/03/2010 - 16:11

Big K, post: 356100 wrote: ...I will try the following: I take a track of well recorded audio or a CD track with decent dynamic at -15 db and I will normalize this track to -2 dB. Then I will normalize it back to were it was and do a phase cancellation test on it. I do not know what the outcome will be.. I will report..might take some time, though.

I was thinking the same thing, but instead of the double normalization I was thinking of normalizing one copy and then sliding a fader on an inverted copy to see how much cancellation I can get.

audiokid Wed, 11/03/2010 - 17:08

When you record something acoustic at say -20. Do we think there isn't anything usable above that peak worth keeping? Notably the space we remove in normalization and what I think contributes to the organic sound and what I refer to as space.

I'm guessing we agree there is noise but the majority are saying it is useless and not worth keeping, correct?

RemyRAD Wed, 11/03/2010 - 21:14

I guess I will chime in here. Normalization is a great function. Especially if you know how to use it properly and take advantage of it. I called it a destructive process because it is either increasing or decreasing overall levels. When you do that, you generally need to include some kind of dither. Or you can elect not to. Adding dither can change the overall perceived sound of a source. So, Destructive. And unlike gain or compression/limiting, etc., it is only normalizing the overall summation of peaks of the entire audio file.

Now this is where it can get dicey. Normalizing can be abused. For instance, those toms that were recorded at too low a level. Normalizing them up still didn't quite give you the definitive crack you wanted to hear. So on that particular track you could try to over normalized to 110%. Now this will take that peak and flat top it off. And what kind of distortion do we get when that happens? 3 RD harmonic distortion. Third harmonic distortion is a dissonant distortion that can in crease the overall cracky sound of that Tom. But then you should also, after after creatively clipping only its highest peaks, you must then normalize down to say, -.6 DB. This will prevent the signal from then exceeding the converters parameters while still retaining the creatively clipped peak.

It's also a helpful tool when dealing with stereo tracks that were not recorded at even levels. You normalize both tracks without the stereo tracking interconnect checked.

Creative clipping through careful normalization manipulation can enhance drums or any other digitally under recorded material. I know, I know, people think I'm crazy for clipping anything. Well I'm not.

I'll clip my clients recordings on purpose but I won't clip my clients on purpose.
Mx. Remy Ann David

Big K Thu, 11/04/2010 - 02:35

Hello Remy...
Idk, but I have a haaard tim to clip anything inside my 32-bit float ( audioengine ) box. Also, dithering is not even an option for normalizing with Nuendo and seems not necessary in the majority of cases. It adds low level noise to to what? Eliminating quantisation noise...Hmmm..
Although my live room is down to < -70 dB noisefloor the human factor produces more noise on the recordings then any quantizing noise should ever produce.
In my DAW you can't over normalize. It ranges from -50 to -0 dB (not plus xdB, but max level). If you need more then that I use the gain function to add or subtract dB values defined to a cent of a dB, or I use the handles on the events to boost and cut ( a fast and easy method, too ) and, of course the usual bunch of gain altering possibilities like fader, etc.
Remy, I am not sure I do understand the procedure you describe with it is only normalizing the overall summation of peaks of the entire audio file ( language barrier...sorry...).

You are right with the sound of just normalized drum hits. They still sound different and might need extra pimping and fade-in/out or crossfades. But with the usual inserts like compressor, etc., it is not often necessary. I see it as part of human imperfection and keeps it less polished, ..I like that.. . Since speed is often essential I usually chose erase silence and get rid of the bleed inbetween and give the remaining hits enough pre and post play with automated fades. Bad sounding hits get copy/paste replaced.
I hardly ever use normalize on single sides of stereo tracks, though. I rather try one of the 3 available panning options, usually with good results. Normalizing only 1 side will boost the center sound source too much, e.g., snare at overheads. Anyhow, if this is a stereo recording from a KB or a CD track this would not be a problem, of course.

Hello audiokid..
I am quite sure that there is nothing and I mean not a thing, at all, above that point. Otherwise it would be detected as peak level. Normalizing is not looking for any threshold under wich it ignores any signal. Even if it is smallish -100 dB at 19 kHz it still would be detected as signal. and if it is the highest level within this event, it is THE peak level. Next event - next peak level, a.s.f. Normalize then adjusts the individual peak levels to the set dB level (higher or lower) in the normalize function It has nothing to do with frequencies. The whole spectrum including lo rumble to the highest harmonics within the recordable range is looked at. Nothing gets lost, only a few empty bits might get added at the extremely silent end...lol..

Hello Bob...
This function is not available to me. I was not aware of its existence in PT.
What ever it is, I rather use the MetaNormalizer in WaveLab, which checks the power ( rather then the peak level ) of all tracks and adjusts the level for each track to come to an even loudness perception over the whole CD sampler, for example. This is also a non-destructive and fast way to reasonable results, but I prefer my own ears and set levels by hand, following the mood and song parts that need to be kept prominent. That is something a machine can't do, ...yet......

audiokid Thu, 11/04/2010 - 12:36

I'm with Big D on the dithering Remy . Sequoia and Nuendo are similar. You set the level and go... Its very simple and non destructive. But, I still hear a difference (which could be due to to snake oil voodo of having to move the faders because the levels have changed so dramatically after its printed and me suffering from post traumatic fade disorder!) = so I don't do it on individual tracks any more (since 1999).
I'm with Remy on the attack too and certain, we are all past NOT being aware of clipping the attack off of anything. But, this is what newbies are doing and I think where this topic gets all piled into the newbies and experienced quagmire. Newbies normailze individual tracks, cram it all into the 2-bus and then compress and smash more. They then start thinking, hey, this sounds great (nice and loud, I should get into the mastering business!) but its too loud now so... Next thing they start dropping their mix faders and it really starts the gong show. This is what people are doing at different stages.

Normalizing all your tracks ( not just one) is a very bad thing and this should be clarified. I'm hoping you normalguyzers agree?

Cheers!

mx abnormal audiokid

RemyRAD Thu, 11/04/2010 - 14:04

Big K, okay, let me be a little more specific. Your description of normalization is quite correct. It is an overall mush of mash that is handled equally. Everything gets moved up or down in an equal manner. No dynamics enhancements should be added. But the fact remains, that different types & manufacturers of software have different ideas of how " NORMALIZING", should be accomplished. For example, Adobe Audition a.k.a. Cool Edit has had an open-ended level peak normalization function. Whereas Sony/Sound Forge offers a restricted 100% Peak Normalization or a RMS plus dynamics control with Normalization feature, which produces a hole nuther aminal. 32-bit float can still be purposefully clipped. As you indicated, overall preset static/fixed level adjustments can be applied with or without dither. Normalization is just a function of the computer figuring out what & where the highest peak resides & setting it to zero DB FS (or some other predetermined peak level) within the entire recorded performance. If you decide to increase gain to a recorded sound file, you can clip anything & everything. Plus even if you think your software won't do something, chances are, there is always a workaround should one desire to go one step beyond. So Big K, if your software only allows for 100% peak Normalization then, you would peak normalize. Want a little light clipping on some of those drums? Kick up your gain by 1/2, 1, 2, 3 DB. Then tell your software to Normalize again and you'll have that harmonically enriched drum thwack with your "creative clipping" at 0 DBFS or, lower so that it's not clipping the converters. Anybody who hasn't thought about using some kind of distortion in their recordings hasn't really thought about their recordings.

Clip lightly and the sun shall shine
Mx. Remy Ann David

Big K Thu, 11/04/2010 - 15:29

I have done the following test, now, done on my single core office pc with Nuendo 5 on OS XPpro 32, file format: 44.1 kHz, 32-bit float, using gain and normalizing:

2 minutes CD track imported from CD

lowered gain to -15 dB and bounced the event without anything else in the path and fader at -0 dB

this file of -15 dB I normalized up to -2 dB and bounced it like above

I took that file and normalized it back down to -15 dB and bounced it, again

Now, I took the original CD track at original level and the bounced file, which I normalized one more time to the same level as the original CD track

I reversed the phase of one track and the resulting signal was zero...
It cancelled out to nix, nada, zip... total silence...
I did expect something like this, but not to such perfection.

In short: the test compared the original imported track to a copy that was:
Gain reduced by 15 dB, bounced, normalized to -2 dB, bounced again, normalized to -15 dB, bounced and finally normalized to the original CD track level.

After all that, I altered the fader gain of one track from this cancellation test, by about 0.3 db, bounced the resulting signal and, after normalizing it to -2 dB, I obtained a still well sounding piece of music with no obvious shrinking of room or air.
Computer Math works....
:-)

I am sure, one or the other will do a similar test, too. Please, post your findings, as well.
After we have collected a few more results from other testers we talk again about destructive or not...

--------------------------------------------

Hello Remy

Thank you for your explanations...
It seems, I am not all that familiar with the sometimes strange variations of other DAW's "Normalizing function". My fault....I must be spoilt by what I have...

Now, I have really no problem with clipping when I raise a bag full of channel above the mark zero. Inside this audio engine you enjoy an enourmes freedom in this regard. I mean, I work on it almost every day for the last 9 years... After the last Punk CD I would have noticed, the latest...lol...
Is there another Nuendo user who has such a shortcoming? I guess, fixed point engines do have a limit, though.

Big K Thu, 11/04/2010 - 15:54

Quote from audiokid:
Normalizing all your tracks ( not just one) is a very bad thing and this should be clarified. I'm hoping you normalguyzers agree?

-------------------------------------------------------------------------------

Hello audiokid...
I fear, you really must have had a traumatic experience with that...
Normalizing* all tracks is nothing else then recording the tracks with a higher level, instead, except the small neglectable difference in sound caused by driving the pre amps somewhat harder, maybe.
The mixing is were the audio engineer decides for the levels of single instruments and the whole picture....

Peace..?..
:-)

* Here, i must emphasise I mean the normalize function of Nuendo and not that of a potential clipping machine.

dvdhawk Fri, 11/05/2010 - 00:57

As in most things, moderation is the key. I certainly agree with audiokid, in that I also put a high value on the 'space' between the notes and hate the trend toward 'music' that has had all the dynamics wrung out of it.

Normalizing the finalize stereo mix in moderation can be useful in avoiding ugly sounding 'overs'. For me, normalizing individual tracks would be a last resort unless something went wrong somewhere along the way and re-doing it wasn't an option.

In reality even the simplest function in a DAW. Basic things like volume change and EQ will ultimately and eventually re-draw the original waveform to reflect those changes when bounced down to stereo. The fact is, some algorithms will do this better than others. After several re-draws, will it be accurate to the nth degree? or result in something more like the telephone game we played as kids? Where after you'd 'passed it on' and had it relayed, and re-translated a few times it barely resembled the original. Again, some algorithms will do this better than others. Let your ears decide how much is too much.

Like a lot of tools, when misused it goes from tool to weapon.

Big K Fri, 11/05/2010 - 08:44

Hear, hear...

Dear friends
Be adviced that I am a forerunner of the brutal, fully militaristic, war-banner weaving, dreaded and infamous league of mercenaries for preserving natural dynamics and sound in music!!!
:-)
Don't blame the tool, blame the stupid bugger who abuses it....

Hello Hawk...
You are right that it is possible that multiple digital processing could warp and twist the processed version to the worse.
Therefore, it is important to reach out for a DAW that does all that non-destructive and in real-time, keeping the original as base file.
So, changing a track EQ multiple times only changes the original to a new sound. It does not bend the EQed audio you hear. Nothing is adding up...
If you bounce that processed audio to a new file, you have exactly that what you wanted. If you are not satified with it, go back to were you were and remix the original file, to fit.
It occurs not very often that you need to process copy after copy after copy.
If somebody has chosen this as his default way of working, he should rather try his luck at selling vegetables on a market...
;-)

RemyRAD Fri, 11/05/2010 - 12:50

I still feel it important to inform those, that do not realize it, that most audio software will automatically re-dither anything that has had level changes made to the original recorded file. Dither which amounts to a digitally created noise comes in different flavors. All of which will color the character of the sound. Most dither settings are preset and unchangeable. Like all software manufacturing blah blah, their processes are claimed to be inaudible. Perhaps so? But it is one of the primary reasons why I have left using Adobe Premier Pro. Unbeknownst to most, Premier Pro takes your 48 kHz 16 bit audio and automatically converts it to 32-bit. This is not a feature that one can disable/disengage. I informed Adobe that I knew how to watch my levels within a 96 DB window and didn't want my audio converted. Especially since it takes quite a long time when working on longform projects and doubles the amount of disk space needed for the audio tracks. This is stupid. So, it's Sony Vegas & Avid for me and no more Premier except in the older 6.5 version before Pro.

And yes, BigK, a good example of how Normalization works. Everything should cancel out completely if no clipping was introduced into the Normalization process. As you have described, NUENDO's Normalization feature does not offer any other kinds of manipulation like Sony or Adobe offer. This provides less chance for stupid mistakes & errors especially if you don't know what kind of color you want your dither to be. Brown? Pink? White? Sinewave? Square wave? Triangle wave? But again, the audio should still cancel when phase is inverted in spite of the noise print background. This is not a dynamics processing function. But an overall raising or lowering audio process, based upon the highest peak. And dither is still automatically calculated and included. Which would be correct und very Teutonicly demonstrated by Steinberg (pronounced Schteinberg) ya? Est gude? I don't habla German. Except that I know I'm a Tonmeister. And a Tonmeister by any other definition is still an audio schlep. Just because you guys created the world's best microphones, nickel cadmium batteries & the first good functioning magnetophone doesn't mean you aren't the best. Actually it does mean you are the best. Achtung! Although I still like PZM microphones in spite of the argument I got into with a German recording engineer. He thought they had too wispy of an over exaggerated high frequency response. Fine. Good. That's why I like them for audience microphones. Achtung, Laffing outen louden!

And hey, there really isn't any harm in Normalizing each individual channel of a multitrack recording. Although I find it generally unnecessary since even in software we have trim control during mixing. The biggest issue with that is that numerous re-dithers is performed increasing the noise print. And doing so now, also makes it much easier to overload the digital or analog summing bus due to poor technique in mixing. But in the proper hands shouldn't be a problem or an issue. Which is why understanding gain staging, be it analog or digital is so important to get a handle on it. And not to press the issue but some Peak Normalization processes also include a peak limiter so as to prevent any " Over Normalization". Those can be really dangerous to the clueless at hand.

I've never been normal much less Normalized in any stretch of the imagination
Mx. Remy Ann David

Big K Fri, 11/05/2010 - 16:18

Is this all true with floating point math? A lot of audio engines in DAWs convert files in 32 and 64 bit , inside for higher esolution processing, which is much appreciated.
But I have not seen, that they safe those files at any other resolution then the original one.

With this dithering after processing, I need to do further tests to find the resulting, stacked-up dithering noise in a file.
I will squeeze and boost the living daylight out of an acoustic recording and if I find any disturbing noise from unsolicited dither processes in it after the 8th generation, I kick ol' Charly (Steinberg) in the nuts.

To come to an agreement, can we say: in knowledgeable hands and carefully designed DAWs normalizing is a usable tool and can become a monster when abused by the ignorant and misled and /or poorly designed DAWs...

Remy, I, for once, have nothing against PZM mics. The same as above applies to the use of those...lol...
What was the argument you had, all about?

There is one thing, I am very sad about German science and engineering: we have also invented MP3....
:-((

Big K ( who wants his dinner, now...)

audiokid Fri, 11/05/2010 - 16:25

Big K, post: 356259 wrote:

To come to an agreement, can we say: in knowledgeable hands and carefully designed DAWs normalizing is a usable tool and can become a monster when abused by the ignorant and misled and /or poorly designed DAWs...
:-((

Big K ( who wants his dinner, now...)

I agree and I'm in!

BobRogers Sat, 11/06/2010 - 10:28

I think it's important to keep in mind here that whether it is dither or rounding error in a floating point calculation, we are talking about noise in the least significant of 24 bits. Yes, errors can compound with many layers of digital processors, but as someone whose favorite songs went through several layers of bouncing on tape (with a much higher signal to noise ratio) I can't get really excited about this.