What Does “Normalize” do to Sound in a File?

[Old Quad Guy]

I’m curious to dig deeper to know how sound quality can be affected when a certain method is applied to a sound file.

I first used “Normal” or “Normalize” with Opcode’s “Studio Vision” in 1991 (discontinued in the 1990’s) on a ripped CD file to experiment with the program and thought it mucked up the sound by making the sound not quite there (compressed sounding) from what it was. Some of the highs seemed to be gone. I know there are improvements and more options with the sound editing programs we have now. However..

What does “Normalize” actually do to sound when one uses Adobe Audition’s “Normalize” option under “Effects” or “Group Waveform Normalize” various settings?

Doesn’t “Normalize” also raise the inherent noise in a sound file, depending on settings? Is there a way or tool to raise the volume in a group of sound files to a certain peak level so that they’re the same sounding without ruining or adding negatively to the sound? Thanks.

 

[neil wilkes]

Normalize essentially fucks up your audio.
There are 2 types...Peak & RMS.
Peak will raisethe level of the highest peak to the amount specified - this does very little indeed except adding a layer of useless DSP & Quantization distortion to your innocent audio.
RMS normalize is the same as a gain change except with added DSP - see above - and it will of course also raise the noise floor.
Best avoided.

Is there a way or tool to raise the volume in a group of sound files to a certain peak level so that they’re the same sounding without ruining or adding negatively to the sound? Thanks.

Details?

 

[Old Quad Guy]

[FONT=&quot]Thanks Neil for the info. Now we get to the heart of the matter of what I'm getting at. I've been using WavePurity Professional's "Find Volume Scale 95%" in the past to raise the aptitude (or volume, at least I think that what it does) so that the separate mono .wavs have the same volume. Is the "Find Volume Scale 95%" the same thing as "Normalize?" I was under the impression that "Find Volume Scale 95%" just raised the volume without compressing the sound and was a different thing. The Charlie Rich “Behind Closed Doors” DTS-CD conversion I did in 2007 used this method.

Is the "Find Volume Scale 95%" ok or not to use or is there a better way to achieve matching the volumes of different files? I'm looking for the best way to match volumes (or maybe not) after SQ computer decoding. As we know, we get different volumes after decoding and depending on decoding script, one is supposed to raise volumes or not according to instructions. Thanks.

[/FONT]

 

[DKA]

The "normalize" function in Sound Forge performs the same function as well. I won't go as far as saying it messes up your audio, but it does something you may not be intending it to do.

The same function exists in other programs (Nero, for one) to normalize groups of files to the same volume level. When burning DTS discs such as upmixes or quad transfers, you should always be turning any normalization options off. This is where people wind up with a disc full of nice, ambient static.

 

[Old Quad Guy]

From [FONT=&quot]WavePurity Professional 5.60 help section:
[/FONT]

 

[Old Quad Guy]

This is the window/settings I've used in [FONT=&quot]WavePurity Professional 5.60 for multiple uncleaned Mono tracks after SQ decoding:
I've had good results using this method, especially if the files are very clean. But should I be doing this? Perhaps it might be better just to clean before decode and just use the volumes one ends up with? What to do or not do with volume is the question and where I'm blocked, because volumes after decode can be uneven.
[/FONT]

 

[FredSoles]

is normalizing the same as doing a volume increase? Is there a problem with increasing the volume of a wav file in a wave editor (other than the noise level being increase)
i would think that if you adjusted all of your volumes before you do anything to the file (NR, etc) you're OK. right?

 

[Old Quad Guy]

To keep it simple my 2 questions are this:

1. [FONT=&quot]Is the "Find Volume Scale 95%" or "Auto Scale Volume" in [/FONT][FONT=&quot]WavePurity Professional [/FONT][FONT=&quot]the same thing as "Normalize" in Adobe Audition?

2. What if anything is anyone doing after SQ / QS decode in regard to changing the volume on the 4 mono files?

Thanks.
[/FONT]

 

[0tto]

1. try to avoid use of tools in sound editor which are based on automation with minimum next to nil control, on how it set by you.
changing volume level in AA by the tools under "Amplitude" - "Amplify/Fade" and/or "Envelop" on the fragment of .wave when need smooth fade in - fade out.

2. i would advise do cleaning of the tracks prior the script processing.
if still needed, you can do second run on the tracks after decoding.
try ClickRepair, standalone, simple but yet powerful appz for this purpose
(hello kap'n krunch if you read this )

 

[Old Quad Guy]

Thank you all for opinions and help. I have a clearer idea now about what is going on.

While the "Find Volume Scale 95%" can be most useful, in that it does raise the "Amplitude" to a certain percentage and all the channels will have similar volumes, if that's what one is going for, it must also be somewhat similar to "Normalizing." Unless one's original source is absolutely clean, it will also raise the "bad" noise in the recording one wouldn't want. Otherwise, I have liked the results on some records. Even though it's relatively easy to decode stuff, there is no “one size fits all” for volume / decoding because each result is different, with its own set of problems and rewards. It really is a “high art” doing these conversions and getting the most out of a recording without ruining the sound.

Thanks Otto, I’ll try “ClickRepair” and do a cleaning before and after to see what we end up with.

 

[JonUrban]

I've used click repair and think it's great. Also, I ususally do all of the "repair" to the audio before I adjust the volume level. Is that the proper way to do it? Once I adjust the levels at that point, if more is needed then I do it at a different setting.

 

[neil wilkes]

Personally I would always say to leave all these automatic level adjustments well alone.
Trust your own ears - they are far more sensitive than any algorithm could ever be.
Normalize is very, very bad - it is almost impossible to overstate how bad this is. It cannot ever improve your audio, it can only degrade it.
Let me try to bust the myths by borrowing from Bob Katz - quotes from "Mastering Audio, the art & the science (2nd Edition)"

1 - The Aesthetic myth.
The ear responds to average levels, and normalization measures peak levels. As a result, a ballad with little crest factor can be disproportionally increased in level & end up louder thana rock piece with lots of percussion.
2 - The technical myth.
Normalizing adds an additional level of quantization distortion. Since the material has already been mixed (to fixed point) it has already been quantized, which pre determines the signal to noise ratio - which cannot be further improved by raising it. Repeat. raising the level of the material will not alter it's inherent signal-to-noise ratio but will add more quantization distortion.
Material to be mastered does not need normalizing anyway, since the ME will be performing further processing. Songs should be levelled by ear, and by ear alone.
3 - Average normalization.
This is another variant that attempts to create an intelligent loudness algorithm based on the average level of the music as opposed to the peak.
neither peak nor average nor any "intelligent" algorithm can do the job properly, because a computer does not know that a ballad is supposed to sound soft.
There is no substitute for your ears.
Having said that, there is one area where this type just might be suitable - where everything is supposed to have the same loudness even if this is not natural (which it won't be).
This is cases like Muzak in a supermarket, or broadcast sound.

Sorry to keep banging on, but the importance cannot be overstated.
The problem seems to me to be exacerbated in fixed point applications, as these will invariably apply rounding & dither after each process.

 

[Old Quad Guy]

Thank you Neil very much for your information! We will avoid "Normalize" at all costs.

Because WavePurity Professional's native language is German they must not call it "Normalize," but the "Find Volume Scale 95%" and "Auto Scale Volume" in WavePurity Professional has to be the same thing as "Normalize" and “Group Waveform Normalize” in Adobe Audition. We will avoid these tools. While the Charlie Rich “Behind Closed Doors” DTS-CD conversion in 2007 gave us "equal sound" and a nice soundfield, it also raised the noise of the turntable slightly as well as other aspects in the overall sound, not good.

When recording into computer from Quad reel or Q8 it is a simple matter of raising or lowering the input of the recording level. But SQ / QS decoding seems to be a different animal in that, as I understand it, we're left with 4 different volumes sometimes, which may or may not be the correct volume overall.

Question: After SQ / QS decode do we then just put the disc together? If we need to raise or lower volume what tool or method should we use?

To try and answer my own question, would it be a simple matter of putting the 4 decoded Mono files in Adobe Audition's "Surround Encoder" and adjust the volume for each channel there by ear using the "Track Level" slider, then export the files?

Again, thank you for your help.

Jim