Normalizing! Yes, that old nugget.

[badgererer]

I've been thinking...and i know a million questions have been asked about normalizing, but I was just recording bass DI and normalized it up to about -0.2db because the level was very low. And I thought to myself, that wont have any negative effect because it's DI anyway. That is true right?

 

But then I thought, my mixing life would be a lot easier if everything was normalized to a certain level. I'd have a better immediate mental picture of how loud everything was, comparitively. And I realise that some instruments may be quite quiet but with a loud peak or two, therefore making them quieter even after normalization - but the difference would still be less.

 

So what is stopping me normalizing all my tracks after tracking? I don't see why I shouldnt....after all, what difference would it be between simply having the volume up louder on the fader whilst tracking? It would be the same right..? Just without the nuisance of the possiblity of clipping past that 0db mark.

 

Ok I'm done typing

[Phil O'Keefe]

To me, the better solution is setting the proper levels when tracking.

 

In this day and age of 24 bit converters, I don't really see the need to "slam" levels going in (IOW, when tracking). I normally shoot for levels in the -18 / -15 dBFS range. That's dB Full Scale"... about 15 to 18 dB "below" the 0 dB / clip point. I find that leaves me with plenty of headroom and a good strong signal when mixing, and it greatly reduces the liklihood of overloading the summing / mix buss at mixdown time.

 

I very rarely ever normalize anything.

[Johnny Storm]

Especially with something like bass, I would use compression to bring my track level up, not normalization. Compression can eliminate problems with transient peaks, too. Normalization just makes the track louder.

[badgererer]

Originally posted by Phil O'Keefe To me, the better solution is setting the proper levels when tracking. In this day and age of 24 bit converters, I don't really see the need to "slam" levels going in (IOW, when tracking). I normally shoot for levels in the -18 / -15 dBFS range. That's dB Full Scale"... about 15 to 18 dB "below" the 0 dB / clip point. I find that leaves me with plenty of headroom and a good strong signal when mixing, and it greatly reduces the liklihood of overloading the summing / mix buss at mixdown time. I very rarely ever normalize anything.

 

I see, so you're leaving plenty of headroom to mess around when it gets down to mixing then? What overall level do you end up with on the finished product? I usually aim to keep it under 0.5db - obviously you need to keep it loud enough so when mixed in with other songs it doesn't sound unnaturally quiet.

 

The only reason I normalized the bass was because the level was very quiet (about -23db) with the line in, I'd imagine cause there was no preamp. So I had to normalize it, even if only a little bit if I were to be tracking around the -18/-15 mark? Thats not really a problem is it?

[Kiwiburger]

A bass DI'd into your soundcard probably won't have the ideal impedance match to get the best sound from your pickups.

 

If you had an instrument level preamp of some sort that should solve both problems.

 

I believe the problem with normalising, in general, is that it is a destructive process that does nothing more than what your fader can do. Cubase SX for example can give a 6dB boost with the fader, if that's all you need. For example - it would be silly to normalise, and then subsequently drop the level at mixtime. Two processes instead of one, with rounding errors.

 

If you used compression or limiting later on, that would be further processing that 'undo' the effect of normalising - so basically it's a wasted step that possibly harms your signal unnecessary.

 

Normalising, like compression with makeup gain, brings up the noise floor. You are better to track as hot as you can, then when you bring the fader down you are bring the noise floor down too.

[Zooey]

 

Originally posted by Kiwiburger I believe the problem with normalising, in general, is that it is a destructive process that does nothing more than what your fader can do.

 

 

That's a good point. I see too many recordists going to all sorts of extreme measures to avoid fader moves. The first thing any newbie should do IMO is to crack the manual and completely grok the automation of his or her DAW of choice. Isn't that what mixing is all about?

 

The other problem I see with normalization to 0.2 dBfs is that there is zero headroom left for processes like additive EQ.

[Zooey]

 

Originally posted by badgererer I usually aim to keep it under 0.5db - obviously you need to keep it loud enough so when mixed in with other songs it doesn't sound unnaturally quiet.

 

 

It's a fallacy to assume that making your overall mix peak as close to full scale is possible is going to make a louder stereo master. A mix that peaks at -5 dB has the same potential (or lack thereof) to make a loud master as a mix that peaks at full scale.

[Phil O'Keefe]

Originally posted by Zooey That's a good point. I see too many recordists going to all sorts of extreme measures to avoid faders moves. The first thing any newbie should do IMO is to crack the manual and completely grok the automation of his or her DAW of choice. Isn't that what mixing is all about? The other problem I see with normalization to 0.2 dBfs is that there is zero headroom left for processes like additive EQ.

 

Exactly!

 

Another thing to consider - when you normalize, you bring up the level of EVERYTHING - and that includes noise.

 

If you have a track that was accidentally recorded a bit low, and you feel you want to normalize it a bit to bring the level up a bit, then by all means, go ahead. However, by tracking at reasonable levels to begin with, it's one less step you have to mess with.

 

If you track with the levels of individual tracks in the -15 / -18 dBFS range, you should have plenty of signal level.

 

One thing you might be getting a bit confused about is the overall level of the mixed stereo file when you're done with the mixing. When mixing, I also shoot for -15 dBFS levels. Occasional peaks might go hotter than that, but that's the average level I shoot for. And if you were to play that back, then play back a commercially released record, yes, you WILL notice a significant volume difference between the two songs. Why? not because you did anything wrong, but because that's an UNmastered mix.

 

If you print your mixes at those levels, the mastering engineer (which is ideally an outside person, but sometimes people DIY on their own projects) then has a good signal level to work with, with plenty of headroom so they have some flexibility in how they work on things. As Zooey said, if you have peaks up in the -.02 dBFS area, you can forget about using any EQ boosting - that would most likely put it over the edge and into clipping.

 

Getting the mix "loud" is really the mastering engineer's job... but be careful! There is a trend on many records these days to overcompress in mastering. A nice "loud" mix is one thing, but IMO, when it gets slammed to the point where it kills off all the musical dynamics and is audibly distorting, then it's gone too far. I feel it's more important to make it sound GOOD than it is to make it sound LOUD just for the sake of "being loud", and at the expense of musical dynamics and sonics. After all, if people want it louder, there's always the volume control on their stereos, right?

[WFTurner]

Most normalizing functions in audio software are destructive processes.

There's too many ways to get more volume out of a track via the tools available in most DAWs and their mixer

without making a permanent change to a track and then realizing you wish you had the original track at hand. The feeling you get when you come to that realization is the feeling of destruction and not encouraging.

 

If you feel you must do it this was then make a backup of the track.

 

If this is something you normally have to do with tracks, then it's time for a look at you're recording techniques.

[Phil O'Keefe]

One other thought - if you normalize individual tracks REALLY hot like that, what happens when you go to do the mix? You either have to pull those faders WAY down, or you overload the mix bus... both of which are not good things.

 

When you pull those faders way down, the fader resolution becomes coarser, and ultra-fine mix moves; subtle volume moves, become harder to do.

[boosh]

Way back when I was on the Musicplayers forums two guys,...One named Phil and one called Where0235522436474?,they told me not to normalize and don't record to hot.

Keep everything between -10 and -20db,....

 

My mixes are so much brighter now and I have loads of headroom....

 

Listen to this Phil guy people!!! He tellss the truth

[badgererer]

okay...lots of advice, thanks guys. You've completely discouraged me from ever normalizing in any normal circumstance, and have also informed me that i will want to keep the level at somewhere around -15db, which is funny because I have always been trying to ride it close to the line!

 

I should explain better on the bass situation though. I have no bass amp, and no budget to get a decent preamp (Unless you can get a decent one very cheap). The signal going into my comp directly is very low, with everything turned up to 11. So...would it be that destructive to normalize it up to -15db to meet with all the other unnormalized tracks? Kind of doing some evil to acheive good, or some crap like that.

 

 

From some of the things you said Phil, I wonder about my full understanding of mixing, and then mastering. It sounds like you leave all EQing to the mastering? And compression as well? If this is so, what do you actually do when mixing? Forgive my ignorance, I read a lot of conflicting opinions on the internet so am rarely able to form a conclusive...conclusion. Thanks!

[boosh]

The way I do it:(Dunno Exactly how Phil does it)

 

1) I record everything with peaks around -10db(no EQ)

 

2) Then I mix,...use my plugins when needed and a bit of EQ here and there.

 

3) Then I mastermix and render the complete audiofile and use some EQ on that if needed.

 

Then the mastering process,...EQ Compression,...Limiter,..whatever you choose.

[doug osborne]

 

Originally posted by badgererer okay...lots of advice, thanks guys. You've completely discouraged me from ever normalizing in any normal circumstance, and have also informed me that i will want to keep the level at somewhere around -15db, which is funny because I have always been trying to ride it close to the line! I should explain better on the bass situation though. I have no bass amp, and no budget to get a decent preamp (Unless you can get a decent one very cheap). The signal going into my comp directly is very low, with everything turned up to 11. So...would it be that destructive to normalize it up to -15db to meet with all the other unnormalized tracks? Kind of doing some evil to acheive good, or some crap like that. From some of the things you said Phil, I wonder about my full understanding of mixing, and then mastering. It sounds like you leave all EQing to the mastering? And compression as well? If this is so, what do you actually do when mixing? Forgive my ignorance, I read a lot of conflicting opinions on the internet so am rarely able to form a conclusive...conclusion. Thanks!

 

 

The advice you have received here is generally great.

 

"Destructive" simply means that it changes the audio file permanently, and you generally can't return the file to its un-normalized state. Destructive is not necessarily bad if you know that the process is exactly what you want, kind of like marriage...

 

There are many ways to bring a file up to a desired level that are not destructive - Clip Envelopes come to mind.

 

I often use CD Architect's Normalize function (which you can undo without harm) as a quick 'n dirty way to bring levels on all tracks in a compilation up before burning a quick CD. I wouldn't do this as part of a serious mastering effort, and this in no way will result iin all tracks sounding consistently and appropriately loud.

 

And BTW, kiwibuger's DI advice was a little off - the function of a direct box is precisely to act as an impedance interface between your instrument pickups and mic-level inputs; the mic pre that the DI is connected to provides the gain, and an instrument preamp can do the same thing.

[rim]

Note that the guide of -15 dB is for recording at 24-bits. If you record at 16-bits, would you aim higher, like -8 dB? Or will -15 dB still get you full range at 16-bits?

[Phil O'Keefe]

Each bit is roughly 6dB... if you record at -12 dBFS on a 24 bit system, you're essentially getting 22 bit resolution. On a 16 bit setup, you're down in the 14 bit range. I think tracking hotter was a bit more important to people back in the 16 bit days, and some people used to push old ADAT decks pretty darned hot... but I always preferred a bit of headroom for the safety net - I hate clipping, and I'd rather not ruin a great take with some aggressive playing overloading the converters, so I tended to record a bit lower than some people did, even back then.

[Kiwiburger]

It's as I feared - badgererer is connecting his bass directly into his PC soundcard.

 

Doug - why exactly do you say my advice was "off"? This is what I said:

 

A bass DI'd into your soundcard probably won't have the ideal impedance match to get the best sound from your pickups. If you had an instrument level preamp of some sort that should solve both problems.

 

Two statements - which was wrong?

 

I never mentioned DI boxes, because I don't believe they would be the best solution in this case (budget considering).

 

This guy really needs a preamp, and ideally one with an Instrument level input, not just Mic level. (Instrument being Hi-z, Mic being Low-z). That would "solve both problems" he currently has, namely compromised impedance match and lack of gain.

 

Or am I wrong? Maybe he should plug a Countryman DI into an MSS-10?

[badgererer]

Originally posted by Kiwiburger It's as I feared - badgererer is connecting his bass directly into his PC soundcard. Doug - why exactly do you say my advice was "off"? This is what I said: Two statements - which was wrong? I never mentioned DI boxes, because I don't believe they would be the best solution in this case (budget considering). This guy really needs a preamp, and ideally one with an Instrument level input, not just Mic level. (Instrument being Hi-z, Mic being Low-z). That would "solve both problems" he currently has, namely compromised impedance match and lack of gain. Or am I wrong? Maybe he should plug a Countryman DI into an MSS-10?

 

I'm not running it directly into the soundcard...its going through my Behringer Eurorack which I believe has some sort of built in preamp in each input. I could be mistaken though. It isn't going directly into the soundcard though.

[Kiwiburger]

Thanks for clearing that up. You originally said

The only reason I normalized the bass was because the level was very quiet (about -23db) with the line in, I'd imagine cause there was no preamp.

 

 

Sorry - I assumed you were plugging your bass direct into the line line input on your soundcard.

 

Your Eurorack mixer has Mic preamps, and Line level inputs, but I don't think it has Instrument level inputs.

 

So you really would benefit from a DI box. I use a Sansamp Bass DI box, which is quite popular. I also have an MXR M-80 for different sound.

 

There are many other options, at all price levels, for DI'ing bass. All of them require a preamp of some sort in the chain. Just be aware that you need a hi-z input for a guitar or bass pickup. Unless they are active, but that's another story - I don't like active pickups myself, but that's not to detract from anyone who does.

 

[geek_usa]

Originally posted by Phil O'Keefe To me, the better solution is setting the proper levels when tracking. In this day and age of 24 bit converters, I don't really see the need to "slam" levels going in (IOW, when tracking). I normally shoot for levels in the -18 / -15 dBFS range. That's dB Full Scale"... about 15 to 18 dB "below" the 0 dB / clip point. I find that leaves me with plenty of headroom and a good strong signal when mixing, and it greatly reduces the liklihood of overloading the summing / mix buss at mixdown time. I very rarely ever normalize anything.

 

 

Yes, this man knows what he is talking about. I just read that today in fact, that -15 to -18dBFS is actually the ideal tracking and mix level.

[TonyCrazyMan]

In my understanding, normalizing is harmful to sound quality when you edit in 16-bit resolution, because dithering is necessary. Also the more loudness that is added during the normalize, the worse. But when editing in 24-bit resolution dithering becomes unnecessary, hence you practically get no added distortion when you normalize a 24-bit track, only a louder signal, with of course the amount of distortion you would have got anyways in the digital conversion by playing louder in. That is not harmful since the louder sound compensates for the distortion and the amount of loudness adjustment is minimal. You get the same amount of distortion by increasing the fader. Now, where is the advantage of normalizing each track before the editing takes place?

 

There are two main advantages:

 

- You get the input signal as loud as possible for each track, hence improving the dynamic range of the instrument recorded and results in a little sharper picture of the sound for digital processing. This aids for example digital compressors.

- You get signal balance on the tracks, which is good because all the faders are correctly showing the real volume level when they are at equal levels, before you start the digital editing. You simply get more signal control this way before the editing takes place.

 

But don't make normalizing a habit, once the input signal on each track is at maximum level you get no advantage by further normalizing, as long as you don't add completely wrong amount of output on some effect you add. Also pay attention to the clipping risk when you start processing effects on the tracks that now peak at -0dB. Digital effects all have output signal level control, they are meant to be used to avoid clipping and used to keep as much of the original signal left after the effect was applied.

 

This is just my understanding of when normalizing should be used. Normalizing is a very useful functionality.

[boosh]

 

Originally posted by badgererer I'm not running it directly into the soundcard...its going through my Behringer Eurorack which I believe has some sort of built in preamp in each input. I could be mistaken though. It isn't going directly into the soundcard though.

 

 

I'm using a Behringer Eurorack 1604A myself,...Never use a DI box,...My faders are on 0 and the Main Mix fader is also on 0

I use to gain knobs(your pre-amp on the Behringer) to adjust recording level. My peaks come around -15.

 

I do this with bass ,guitar and vocals....

 

Have absolutely no probs when I do it this way and my mixes are as clear as Cristal.

[greendoor]

I have to disagree with TonyCrazyMan. Normalisation cannot give you more dynamic range, or sharper resolution. That boat has sailed, back when you under-recorded it in the first place.

 

Normalisation is just making it louder with empty bits that provide no more detail.

 

Digital gain is multiplication arithmetic - not addition. That means rounding errors. Agreed - it shouldn't matter whether that is done in a normalisation process, or when you raise the fader.

 

BUT - if you normalise, what are the chances you will then have to lower the fader when you mix? Very high. So - that means to achieve your final mix level, you will need two gain change calculations instead of one, each with rounding errors. For what advantage? None.

 

The claim that the higher level might benefit plugins, I believe, is wrong. In some cases this will cause plugins to clip, because some plugins need headroom.

[TonyCrazyMan]

 

Originally posted by greendoor I have to disagree with TonyCrazyMan. Normalisation cannot give you more dynamic range, or sharper resolution. That boat has sailed, back when you under-recorded it in the first place. Normalisation is just making it louder with empty bits that provide no more detail. Digital gain is multiplication arithmetic - not addition. That means rounding errors. Agreed - it shouldn't matter whether that is done in a normalisation process, or when you raise the fader. BUT - if you normalise, what are the chances you will then have to lower the fader when you mix? Very high. So - that means to achieve your final mix level, you will need two gain change calculations instead of one, each with rounding errors. For what advantage? None. The claim that the higher level might benefit plugins, I believe, is wrong. In some cases this will cause plugins to clip, because some plugins need headroom.

 

 

resolution. That boat has sailed, back when you under->recorded it in the first place.

 

Yes, I agree, I started to question that after I had written it. The dynamic range is what it is when the signal has been sent to the digital audio converter. I think what I meant actually was the signal-to-noise ratio, but that is also the same unless the normalization process doesn't include some kind of distortion cleaning, then you would actually win by normalizing since you would get a better signal-to-noise ratio. Also, in terms of rounding errors on the multiplication calculation, my understanding is that when you process bits on the 24-bit depth these calculation errors are on the atomic level which makes even dithering not necessary. That means the quality loss is only teoretical. In such a situation no noticable distortion would be added during normalization, you would get the same distortion simply by raising the volume on a digital input level fader. I agree that a signal that peaks at -0dB will probably need to be decreased in amplitude when digital editing takes place. But even the loss in quality of this modification is on the atomic level, hence not leading to any hearable noise in practise. I think the advantage of being able to get more control on the amplitude of input tracks through normalization is greater than the disadvantage caused by teoretical sound quality loss due to calculation rounding errors, in 24-bit depth mode. If you want headroom from there on all the tracks you can simply lower the volume level on each track as much as you want to get a suitable headroom. If the normalization program also includes distortion cleaning you would actually get a little better sound quality with a better signal-to-noise ratio.

 

wrong. In some cases this will cause plugins to clip, because >some plugins need headroom.

 

I agree that clipping will happen easily when the peak is at -0dB. As soon as the amplitude is affected positively by the effect, the clipping will happen. That's often the case. However, if the peak is already at -0dB a compressor effect set to limit at -0dB wouldn't have to make any limitation calculations on the waveform, hence it would aid the plugin. This is often the case because you want to add the compressor the first thing you do after the track has been recorded to get some instruments stand out in the mix and to smooth out instruments with a high dynamic range to get a good base balance to start editing with.

[badgererer]

Opened up a real can of worms here didn't I! Well at least I got my questions answered