normalizing

Blue Bear Sound said:
Ah, I see... no - I meant that it uses peaks...

No... it's not limited to integer values at all... you can add 0.75dB of gain -- DEFINITE round-off error there!

Plus, adding a decibel is a multiplicative process, decibels being a logarithmic form of measure. So even if you're adding exactly 1 decibel, you're really multiplying the SPL by approx 1.2599 (I think).
 
Blue Bear Sound said:
And I already explained that normalization DOES NOT give you the intended results of equalizing levels between tracks....... your comment is simply wrong - period. LIMITING is what allows for level balancing across tracks - and limiting is NOT the same as normalizing.

Wrong again.

Simple case of 2 tracks, one at normal levels and one at low levels. Set mixer at defaults, and produce the result WAV. Listen to result. Normal track is much higher in volume that other track.

Then go back, select second lower track. Use normalization to bring up to normal levels. Remix. Listen. Now the're the same. Ears don't lie. VU meters don't lie. WAV editors where you can see the WAV form don't lie. They all show the same result.

Ed
 
A much simplier example is telephone voice mail systems. When a caller leaves a message, many modem based recording systems result in a WAV file that is too low in volume to be easliy heard either over the telephone or PC speakers. Thjis occurs for various reasons, but the result is usually the same.

Enter normalization, which is done after the message is completed, but before the call is completed. All of a sudden, the volume is quite respecetabe and quite audible. The difference is night and day.

Ed
 
Ed, c'mon.... do you even know what apparent loudness is? Because it ALONE determines the track level, NOT the meters, not the waveform in a wave editor...

Normalizing works on the differnece between the peak and the max threshold (typically, but not always 0dBFS)....

As I explained earlier, a single momentary transient peak could cause the normalization process to net you only a minor increase in level.

You're arguing for the sake of arguing now - because you're completely neglecting the very real fact that I'm presenting.
 
The mechanics of basic audio normalization are well known. The idea is to raise the level without distorting the result. When one has mismatched tracks, that is generally a good thing to do.

Ed
 
geez, man...

Even ignoring the very important point of frequency distribution relative to our ear's weighting being critical (apparent loudness), how do you get around this fact:
"...a single momentary transient peak could cause the normalization process to net you only a minor increase in level."
 
Blue Bear Sound said:
Ed, c'mon.... do you even know what apparent loudness is? Because it ALONE determines the track level, NOT the meters, not the waveform in a wave editor.

Again wrong.

What determines the track level is the actual track audio values. These aren't driven by anything on the PC but by the original source and recording process. A 44KHz 16 track has 44K such samples each second, each with 16-bit of volume resolution.

Levels are what they are, and no amount of words will change a low level track to something else, only something in either signal processing or mixdown can do that.

Ed
 
Sorry, Ed.... you're way, way off... I'm stating facts and you're simply ignoring it.

Kinda like you insisting the world is flat while I've given you scientific proof that it isn't...

You really need to brush up on your theory... one day, I suspect you'll be rather embarrassed to discover I was correct, but until then, believe what you like...

I'm done with this discussion...!
 
Blue Bear Sound said:
Sorry, Ed.... you're way, way off... I'm stating facts and you're simply ignoring it.

Kinda like you insisting the world is flat while I've given you scientific proof that it isn't...

You really need to brush up on your theory... one day, I suspect you'll be rather embarrassed to discover I was correct, but until then, believe what you like...

I'm done with this discussion...!

Actually I think it's you who is ignoring the facts. Perhaps when you been around this area longer some of this will sink in.

Ed
 
C'mon, Ed.:rolleyes:

Try this:
1 - Take an uncompressed track with some dynamics.
2 - Normalize it to -3db.
3 - Duplicate it, then limit one to -10db.
4 - Now norm them both to -1 and listen.

Result - The limited one will sound a hell of a lot louder.

If you want it in visual terms, it's area under the curve that matters. All norming does is set the loudest peak in your wav to the given level, and moves the other peaks a corresponding amount. It's the level of all the peaks that matters. Norming will only give you approximately equal volumes on approximately equal tracks, with no wild random peaks.
 
Slaps Forehead said:
C'mon, Ed.:rolleyes:

Try this:
1 - Take an uncompressed track with some dynamics.
2 - Normalize it to -3db.
3 - Duplicate it, then limit one to -10db.
4 - Now norm them both to -1 and listen.

Result - The limited one will sound a hell of a lot louder.

If you want it in visual terms, it's area under the curve that matters. All norming does is set the loudest peak in your wav to the given level, and moves the other peaks a corresponding amount. It's the level of all the peaks that matters. Norming will only give you approximately equal volumes on approximately equal tracks, with no wild random peaks.

I understand this process. As you said, "All norming does is set the loudest peak in your wav to the given level, and moves the other peaks a corresponding amount". However for many audio tracks, this alone works fairly well. For those cases where a single spike exists, other means are needed to get things more in line.

That's when the limiting process enters the equation. Good PC recording software will generally have options for both areas so the result can be tailored as needed. In some cases, it can do both as part of the same audio pass.

Both have their value and both are useful tools for the audio engineer.

Ed
 
this isn't normal

I've listened to more stereo songs, normalized and not normalized..i didn't hear a difference in my car.

- standard Kenwood CD player 45wattea ch/ Pioneer speakers.
no difference heard form my home stereo either, ONKYO w/ INfinity,Paradigm,RAM and Small speakers.
anything crappier, boom box or computer speakers and it would be hard to say it was the "normalizing software" distorting the music!

Someone said they can definitley hear the difference.
maybe this is caused by the quality of speakers and related acoustics, but normalizing is a "novice" friendly tool. I would have to agree it is a "lazier way" and does "fhk with the waveform again"..no way around that.

The same person who hears a difference however, can probably hear the difference between a "compressed" minidisc vs. wav file vs. CD vs. DAT...
Blindfold me and i couldn't...someday maybe I'll have some $$$UREI speakers!...

it's getting kinda clearer reading this thread, and at least i learned about normalizing and peak sensitive/dependentcy.

reducing the processing chain? 6 of one 1/2 dozen the other?
1)one chooses to use compressors and limiters and get a nice smooth stereo mix before burning a Stereo CDR to listen to while burning one in the car.

2)the other records crazy sht, samples, voices on his portable DAT, wants to mix stereo songs from different sources(his entire fhkn home recording mixes of 40 years! from his 4 track cassette to his PRO TOOLS Triton).
....this one chooses software to limit and noramlize very quickly. (probably listens to most music on plastic computer speakers and doesn't have a problem with listening to MP3's).

probably the same argument with MP3 degradation..there has to be degradation anytime you "fhk with/process a waveform".

but can a human really, on an average day, tell a huge difference? I don't think it flys.

TEST:
I did do a "Blindfold test while driving the other day.
I would close my eyes, change songtracks and listen to the various songtracks, one song, Normailized, Limit and Normalize, Normailze and compressed(software), no processing....no major difference heard.

Note:
1)DO NOT KEEP YOUR EYES CLOSED VERY LONG WHILE DRIVING!
2) don't do this in your garage with the doors closed- may cause death.
3) sitting in driveway is ok, but your nieghbors may think your having a nervous breakdown sitting with your eyes closed with the car running.
 
Just a tidbit...

I used to use PC normalizing (along with editing to tame those stray peaks, so I could get more normalization). But then, I found the limiter, world of difference! Like others have said, it raises all of your apparent levels a certain Db level and then clips everything above that. Of course you lose some dynamics, but you achieve maximum volume without clipping.
That is, if you can negotiate the loss of dynamics with the gain in volume.
BTW, I've recorded live stuff on minidisc and have had fine resullts using normalising, but if I could do it all over again or if I do it in the future, I will use a limiter (if necessary).
 
what is normal?

so what i'm getting burned into my brain is...

Normalizing is good for beginners, lower end equipment with like ASS mixes & sounds and low end monitors & non-sound designed rooms to hear the details.

Normalizing comes up short when tested against high end equipment, compressor/limiters, nice monitors in a sound-designed room etc...
1) And optimum tracking techniques that don't have "noise floors" that get amplified when normalizing.

Logically and expected that with pro equipment..
you can hear the difference (or degradation of Normalizing).
I wouldn't argue that logic.

I ran a series of listening tests this week comparing A/B/C/D Normalized vs. Non Normailzed (minus the hi-end monitoring), posted in Mixing thread...basically in the car I didn't think Normalizing destroyed anything. but still thinks it the "lazy way" and a "novice tool"...and if your listening thru plastic little computer speakers with a 10watt card-amp wht the fhk difference does it make.

yeh, some can hear a difference....some can tell the difference between Sgt. Pepper Mono version and Stereo...
some can't/dont' try. :)

the Rutles said....the Mono Pepper version was the only one where the Beatles where in the studio with the Engineers mixing...the Stereo version was for the Marketing guys moving into this NEW and IMPROVED STEREO market" and was done later....still depressing & amazing how G.MArtin and crew did all this with the equipment at hand...and Paul was Dead, and it was actually Eric Idle filling in on bass.
 
Re: Just a tidbit...

freeztar said:
Limiting:
Like others have said, it raises all of your apparent levels a certain Db level and then clips everything above that.

so it sounds like Limiting raises ALL levels, including Noise?

....you can't always get what you waaa uuunt...but if you try sometimes....you get what you need......owwhhh babbbyyy

man, talk about ASS mixes, I was listening to my Stones "DIGITALLY REMASTERED" CD's...first tracks up to Spend the Night...were fhkng ASS....horrific ear bleeding...
 
From The Mastering Engineers Handbook by Bobby Owsinski

From interview with Bob Katz, co owner of Orlando-Based Digital Domain, Bob Katz specializes in mastering audiophile recordings of acoustic music, fromfolk music to classical . The former tachnical director of the widely acclaimed Chesky Records, Bob's recordings have recieved Disc if the Month in Stereophile and other magazines numerous times, and his recording of Portraits of Cuba by Paquito D'Rivera, won the 1997 Grammy for Best Latin-Jazz Recording. Bob's mastering clients include major labels EMI, WEA-Latina, BMG, and Sony Classical, as well as numerous independent labels.

Q: Do You ever normalize?
B.K.: "Normalize" is a very dangerous term, I think it should be destroyed as a word because it's so ambiguous. If you mena do I ever use the Sonic Solutions normalize functions so that all the tracks get set to the highest peak level, the answer is no. Or do I ever use TC electronics finalizer normalize function to find the highest peak and bring it up to 0db? No, I never do that. Do i use my ears and adjust the levels from track to track so that they fit from one to the other, then use compressors and limiters and expanders and equalizers and other devices to make sure that the highest peak on the album hits 0db FS? Yes, I do. I don't call that normalizing, though.


From interview with Glenn Meadows, the two-time Grammy winner and a multi-TEC award nominee who has owned the Nashville-based Masterfonicssince the 70's (most recently purchased by Emerald Entertainment). He has worked on scores of gold and platinum records for a diverse array of artists including Shania Twain, LeAnn Rimes, Randy Travis, Delbert McClinton, Widespread Panic, and Bananarama, as well as for producers and engineers such as Tony Brown, Jimmy Bowen, and Mutt Lange.

Q: Do yoy ever use normalize fuction?
G.M.: No.I don't use a computer to decide how much to bring something up. Typically, I will process on the way into the workstation. I am not a load-it-in-and-then-master kind of guy. I prefer to take the original source material and go thru whatever processing gear I decide I need or would like to use on the project, and come into the work station and deal with it that way.

I have a reffernce point where I park the monitors when I start working on the cut, and I kin of get a feel for what it is doing and then look at the head room comming in to see where I am at. Invariably I end up within 1/4 or 1/2 db at the top, maybe because there is a little bit of a peak limiter sitting there as a protection. But once in the work station, I will use the processing only as subtle final tweaks. I don't use the internals of the workstation as my mastering tools per se. The work station is an editing area. It is a scratch pad to do all the work in and compile it and put it together. The outboard gear is what I use for mastering and that is just the way I have grown into it.

Just a couple points of view on Normalization.
 
Back
Top