Subject:Compress & Normalize

Posted by: spongebob

Date:5/20/2003 1:58:48 PM

After creating a track in Acid, I usually render as a wav, normalize in Sound Forge, and then apply the 16 bit master compression preset. I was wondering if it is better to compress BEFORE or AFTER normalizing?

Subject:RE: Compress & Normalize

Reply by: groovewerx

Date:5/20/2003 2:42:55 PM

compress the individual tracks (as needed). limit then normalize the final mixdown.

calibrate your mixing environment to insure proper levels in the first place. then you won't have to normalize individual tracks.

Subject:RE: Compress & Normalize

Reply by: spongebob

Date:5/20/2003 3:57:24 PM

Thanks. BTW, does it make a big difference if you compress/normalize individual tracks within a piece vs. compress/normalize the entire piece after it is complete?

Subject:RE: Compress & Normalize

Reply by: Vocalpoint

Date:5/20/2003 4:23:35 PM

I would recommend to stay away from any Normalize period. Grooveworx hit it on the head...a well crafted mix with proper levels to start should always give way to a better sounding track in the long run. I do use compression on my tracks (as sparely as possible) but Normalize would actually alter the source material permanently and that is a no no in my workshop.

Cuzin B

Subject:RE: Compress & Normalize

Reply by: SonicSounds

Date:5/20/2003 4:40:19 PM

The problem with not normalizing or maximizing is that you don't get a very "loud" signal from your end product. When you compare them to any commercially-produced track, you'll notice right away that they aren't anywhere near loud enough. Normalizing will limit your peaks and boost your valleys.

Subject:RE: Compress & Normalize

Reply by: Vocalpoint

Date:5/20/2003 10:17:37 PM

Maximizing and Normalizing are two very different things. I will maximize and limit as needed but not at the expense of meeting the RMS level of any "commercially" produced track. Most stuff out there today sounds like complete shit because of too much maximizing.

Normalizing kills any dynamic the individual track may have. If that's your thing - then go for it. But while normalizing might seem like a good thing - Loud does not mean good.

Pink Floyd and Steely Dan seem to have done well for themselves without being overly "loud"...if ya get my drift.

Cuzin B

Subject:RE: Compress & Normalize

Reply by: Chienworks

Date:5/20/2003 10:29:50 PM

Normalizing doesn't do any limiting or compressing.

Subject:RE: Compress & Normalize

Reply by: stuffedspacedog

Date:5/20/2003 11:16:18 PM

I use a compressor first (Waves Renaissance Compresser) on a fairly light setting, then stick it through a maximiser (Waves L1 Ultramaximizer). You end up with a significantly louder end product without squishing all the dynamic range out of your work. Normalizing is better if you're working in "top 40" or broadcast territory.

Subject:RE: Compress & Normalize

Reply by: Vocalpoint

Date:5/21/2003 6:58:41 AM

Stuffed,

I work in broadcast everyday. Make no mistake - the L1 is in fact squishing all the dynamic range out of your work. Can you see that straight line across the top of your wave form after a bout of L1....there's your dynamic range - gone forever.

We avoid it at all costs. And normalizing is definitely not better. Normalizing simply alters (most of the time ruins) the recorded material.

You guys need to really examine your "loud is better" fixation.

Subject:RE: Compress & Normalize

Reply by: dkistner

Date:5/21/2003 11:28:40 AM

Chienworks, that's what I thought! These guys here are talking about normalizing as if it were compression. I thought all normalizing did was adjust the amplitude of the entire audio file up or down so the maximum peak is at the level you say (e.g., 96%, -3dB, whatever). (I'm assuming here you're normalizing in an editor post-mixdown.)

As I understand it, normalizing the final mix as a strategy to get your audio louder can be bad because you're also increasing the noise floor. Since I don't have any noise floor, I don't worry about it. Someone recording into the box, however, or who has a setup that results in noise should definitely worry about it. That's where getting the levels set just right going into recording is super-important. (Believe me, I know. I spent three months dragging some voice recordings up kicking and screaming out the noise floor for a project for a friend of mine. I literally had to rearticulate everything almost sample by sample.) I'd normalize before dithering in any event.

Somebody correct me if I'm wrong!

Subject:RE: Compress & Normalize

Reply by: Vocalpoint

Date:5/21/2003 12:52:24 PM

Diane,

You are correct. I would normalize only as the very last resort - like I could not re-record a part or it was a clip flown in from somewhere else where the levels were horribly low.

Recording at the correct level the first time is always the best way to go. Properly recorded tracks should need too much help come mix time.

Subject:RE: Compress & Normalize

Reply by: SonicSounds

Date:5/21/2003 1:53:15 PM

CuzinB, when you say stuff nowadays sounds like shit, what albums/tracks are you referring to?

When I say "loud" I don't simply mean volume. You can't say that by simply limiting your tracks out of clipping range on your original mix that you're getting all of the dynamics from your song are you?

Do this as a test and you'll see what I mean by "loudness". Take an old Steely Dan tune ("Peg" for example) and play that through your stereo at a set volume. Then take something new like N*E*R*D's "In Search Of..." (although the highs are a little too much IMO) and play it at the same volume. You can't tell me that they both have the same dynamic "loudness". Even though both were mixed to the clipping point (-3db to 0db), they both aren't even close to being the same dynamic "loudness".

From what I know, some normalizers allow you to limit your peaks as well as boost your valleys (which is like compression, isn't it?)

CuzinB, are you saying you don't compress, normalize or maximize anything?

Subject:RE: Compress & Normalize

Reply by: Vocalpoint

Date:5/21/2003 4:21:53 PM

Sonic,

"CuzinB, when you say stuff nowadays sounds like shit, what albums/tracks are you referring to?"

This list would encompass almost every mainstream rock release in the last 5 years. But for one recent shining example of this completely unnecessary habit in today's recording landscape - I would point you to the fairly large outcry (from a good chunk of their fanbase anyway) over the latest Rush release Vapor Trails. I have followed Rush since 1975 and the sheer distortion and shitty sound of their latest is a prime example of how even a world class act like Rush can ***k things up by getting carried away with a mastering limiter. The latest Foo Fighters and Creed are also very poorly done.

As far as your test goes - I master a variety of materials for my clients everyday so I know exactly what you are getting at - BUT piling on the RMS (which what you are really talking about) just doesn't make your material any better. In fact - it can make it much worse.

Today's practice of maximizing the crap out of everything got it's start as a record company ploy to be "louder" than the next CD in a radio station music meeting (or on the air). For some unexplained reason - some moron figured that if Loud is good then Louder must be better and this logic would somehow help when programmers are picking tracks to add. I worked radio for 13 years and I know all about it.

The dirty secret that these guys really don't know is that deep inside all FM radio stations there are much bigger, larger, nastier optocompressors that pound the shit of of these songs even more because they are packing so much volume.

A radio station compressor has only one goal in mind - to even out the signal so radically that nothing sticks out. Ever heard one of your real "loud" songs on the radio? Won't happen - given the method that these compressors work - a song with less volume will actually be brought up in volume where your NERDS example will be caught by the optocomp, pounded down even more and in the end made to sound even more like shit than it already does.

Also for the last time - normalizing has nothing to do with compression.

Finally - compression and limiting are excellent tools that have been around since recording began and when used sparingly and correctly can have a wonderful and necessary effect on a mix. I use em all the time.

However - If my track isn't as "LOUD" as the NERDS...I could care less. Steely, Pink Floyd or Dire Straits didn't get their legendary sound perfection status by pounding their listeners with super loud, dense, lifeless mixes year after year. Check out Gaucho or the Wall or Brothers in Arms on a real good system. Sure...the RMS is much lower and the musical density is completely different but some of these recordings still stand today as the greatest of all time - fact is it doesn't need to be "LOUD" to be good, my friend.

For me personally - it's about clarity and restraint...let the song breathe and build on it's own....I don't know about you - but loud, dense music just tires my ears out...it's like a constant buzz that never lets up. If I need it louder, that's what my remote is for...I don't need the record company to ruin my party by pinning every track to -0.01 db with no room to breathe.

I would rather have a user turn the volume up slightly and maintain the integrity of the track and it's dynamics instead of letting a series of cheap plugins max the track RMS and change the source mix so radically that people are going to be reaching for the volume knob to turn the damn thing down...as is the case on every track on Vapor Trails.

Cheers,

Cuzin B

Subject:RE: Compress & Normalize

Reply by: SonicSounds

Date:5/21/2003 11:04:57 PM

You've got a point about why tracks are louder now, but don't you think a lot of it has to do with the newer technology? Things are louder and clearer because we have the technology to do it.

I won't comment on current Rock recordings since that's not my bag, but in the realm of electronica and hip-hop, I think most listeners prefer a "louder" mix. I hate popping in an old A Tribe Called Quest CD and having to crank up the volume because it was mixed so low. I'd rather lose what little bit of dynamics I might and have a louder mix. The average person won't even notice the loss. I know this might tick off the technophiles out there but it's the truth. If anything, having a lower final mix is worse IMO. You're not hearing the full clarity and range from the track. It's just like recording a track to low...it's not being recorded at it's full clarity.

I guess it's all personal taste.

Subject:RE: Compress & Normalize

Reply by: fresnog

Date:5/21/2003 11:51:12 PM

In the past, I would take most or all of my tracks (electric guitar, bass, vocals) and normalize them in SF before mixing together in Acid. However, I have recently quit that and concentrated on getting the best sounding original signal into Acid - not necessarily the LOUDEST signal, but the best I can get using a little compression and EQ on the way into Acid.

The last song I just finished - I took the 24 bit wave over to SF and applied some slight EQ and that was it. I was so happy that I didn't have to tweak the hell out of it or worse......Ultramaximize it.

Subject:RE: Compress & Normalize

Reply by: dkistner

Date:5/22/2003 4:26:34 AM

Rip Rowan's use of the analogy that overly "maximized" (squashed) mixes of the latest Rush ilk are like TYPING IN ALL CAPS ALL THE TIME SO THE EYE (E.G. EAR) FATIGUES AND PEOPLE STOP WANTING TO OR BEING ABLE TO PERCEIVE WHATEVER IT IS YOU'RE TRYING TO COMMUNICATE BECAUSE THERE ARE NO PEAKS AND VALLEYS FOR THE EYE (E.G. EAR) TO HANG ONTO AND EVEN DELIGHT IN BUT AT THE LEAST WOULD ALLOW THE SENSES (EYES, EARS) TO BE ABLE TO GET A CONSCIOUS OR EVEN UNCONSCIOUS GRIP ON THE MATERIAL AND NOT SLIDE OFF SO THAT SUDDENLY YOU JUST GLAZE OVER, TUNE OUT, TURN OFF BECAUSE, GEEZ, THIS IS JUST REALLY HARD TO READ (LISTEN TO) AND IS IT WORTH IT EVEN TRYING TO SIT THROUGH THIS WHEN YOU COULD BE READING/LISTENING TO SOMETHING ELSE is right-on and explains the degradations of "louder is better" quite well, I think.

Over the Limit by Rick Rowan (from discussion at kvr). This is an extremely instructive article, and he's painted some pictures, too.

Matter of personal taste? Some people prefer to type and read everything in all caps, I guess. Probably because they don't understand the effect all-caps has on readers. The rest of us just delete it or go right past it without even bothering to try to take it in.

Subject:RE: Compress & Normalize

Reply by: braulio

Date:5/22/2003 7:29:41 AM

Diane, you've made a very good point. I believe that compression can be used to bring a certain track of a song to the forefront, or to correct dynamic problems with vocal parts. I don't use any compression if I can avoid it. I like to think the balance of instruments that I've used creates the dynamics I intended without the use of compression. I always normalize using the peak method to avoid messing with my dynamics. The examples shown in Rowan's article clearly show that compression to the hilt is destructive to the quality of your mix.
The greatest composers in the world always leave some holes between the notes to let the silence shine through. It's kind of like little rest areas for the ears on the interstate highway of music. It's true that Beethoven never got a record contract, but there are still herds of artists and orchestras doing covers of his pieces, and they still rock.

Subject:RE: Compress & Normalize

Reply by: Vocalpoint

Date:5/22/2003 10:17:26 AM

Sonic,

You would think that newer technology should only enhance your listening experience...not destroy the mix to the point of being annoying. Granted - it is dependant on the type of music being examined.

But Diane makes a great point on the Rip Rowan article. (Which really started the whole Rush issue). The most damning part of his article (which is very well done) is the audio samples displayed in SF software that clearly shows that today's newer technology really ruined what could have been an outstanding Rush project. The loudness factor of that CD is so over the top that I cannot tolerate that disc at all.

And you are right - it is personal taste - but I am a technophile...and I still do not subscribe to the notion that my mixes need to "compete" on some imaginary volume level with other product out there. I have never had any of my clients complain about my stuff. They can always turn it up if they want.

Cheers,

Cuzin B

Subject:RE: Compress & Normalize

Reply by: Spy

Date:5/22/2003 10:47:02 AM

"I hate popping in an old A Tribe Called Quest CD and having to crank up the volume because it was mixed so low. I'd rather lose what little bit of dynamics I might and have a louder mix. The average person won't even notice the loss."

It all depends how the 'average person' is listening to the track. Commercial radio has (IMO) a nasty habit of compressing the hell out of everything, including the announcers/DJs, with the net effect being that if you play an old CD or vinyl LP on your home stereo it sounds 'quiet' in comparison.

But that's why amplifiers have volume controls. Besides, 'loud' isn't necessarily best even in Hip Hop and Rock. Dynamic range can make the difference between a good tune and an all-time classic. See http://sound-on-sound2.infopop.net/2/OpenTopic?a=tpc&s=215094572&f=884099644&m=7203098917 for the arguements.




One Love, Spy!

Subject:RE: Compress & Normalize

Reply by: RasKeita

Date:5/22/2003 11:00:24 AM

The origanal post seems to have been lost. First taste is subjective the (loud/not loud arguement). Comparing Rush and Steely Dan productions to settle fidelity arguement is like saying taste this orange and tell me how a grapefruit tastes. Simply put what sounds good to me may not sound good to you. Second Compression/Limiting, Maximazation,Normalization are all very different processes, one cannot be substituted for another, and shouldn't. Third it was stated "Normalize would actually alter the source material permanently and that is a no no in my workshop." this is a misconception, if you use Sound Forge, take the original file copy it, normalize the copy name it, "normalized copy" now how is the source material even affected. Compression is a tool somewhat like say a flanger,wah wah, ect. it can be very useful or very damaging. The best way to find it's true structure is take an audio clip you can use Acid for this experiment bring the same loop up twice use compression on one notice the differences, do the same thing with, Maximizer, Normalizer. The thing is, none of these tools should be avoided, they should be used to enhance your project. Sometimes less is more...Level Vibes

Subject:RE: Compress & Normalize

Reply by: Vocalpoint

Date:5/22/2003 1:04:14 PM

RasKeita,

Yep - you are right. The original post vibe has morphed into something else.

Regarding your "Normalized Copy" process...sure that would work as long as your are just copying mixes. But generally copying each element from an entire project is not really desirable (or necessary). All I was saying is that I feel that normalizing is not a natural thing to do to a recorded bit. If it's a case rescuing some bad audio, I may resort to it but artifically raising all the levels (and noise floor) in a mix so everything is more or less at the same level does nothing for me.

The vibe I was getting from the original poster was that he was going to start normalizing every individual track in the project. After permanently altering the original parts (unless you copy everything) what's the point of that? - I guess you could normalize a complete mix but then your mix is not really your mix anymore...

Cuzin B

Subject:RE: Compress & Normalize

Reply by: SonicSounds

Date:5/22/2003 1:32:04 PM

Well, I'm not arguing the fact that compression and making everything loud is better. Actually compression can be a touchy issue and can be annoying. I just don't think you get that presence that you may or may not like unless you normalize or maximize.

In the case of compression being annoying, I think the opposite can be said somtimes. Listen to Daft Punk's "Discovery" LP. The whole album has been saturated with compression but it adds an aesthetic quality to the music.

I haven't heard the above-mentioned Rush CD but how does the audio info look in waveform view? Is it pushed way past clipping? I just want to know how some tracks can be within the -3db range and be loud while other tracks are in that same range but aren't very loud.

Subject:RE: Compress & Normalize

Reply by: Vocalpoint

Date:5/22/2003 2:46:47 PM

Sonic,

The Rush CD has no waveform at all. It basically looks like a solid brick with a shaved top on most of the tracks. While band members revealed later that there were "small techincal issues" with the recording - I believe that the final mastering team did the most damage on this when they actually pushed the track boundaries past digital zero for most of the CD into the very annoying digital distortion realm. As noted in the Rip Rowan article....going past digital zero is not pleasant like going past analog zero.

The Rush CD is filled with nasty digital noise that is very unpleasant and downright tiring to listen to. It's one of the first CDs I own that made me jump for the volume knob to turn it down. On my first listen, I thought my system was acting up since the sound was filled with a constant hard to place grinding noise. It was only later that I discovered the CD mix itself was to blame.

For the mystical answer as to why "some tracks can be within the -3db range and be loud while other tracks are in that same range but aren't very loud", try and find out all you can about RMS values in recorded material. This is key to making the "loudness" world go round.

Cuzin B

Subject:RE: Compress & Normalize

Reply by: spongebob

Date:5/22/2003 3:38:41 PM

Following this thread has been enlightening, even if it has wandered from the original question.

I originally asked the question because I made a track in Acid, burned a CD, and gave it to a dj at a club with a midiocre sound system. When the dj played my track, I noticed a big difference in volume between my track and other tracks.

So I figured that I should pump the volume somehow. The volume meter in Acid was already approaching 0db, so I needed a way to add more without clipping. Otherwise whoever is at the controls must play with the volume from one cd / record to the next, and that someone would rather be focusing on the mix / fader as opposed to re-adjusting the volume each time they mix.

Now I realize that compressing and normalizing may compromise the dynamics and may not be a good idea when creating music that is meant to be listened to in the comfort of your own home. But in some situations, it may be the way to go.

Subject:RE: Compress & Normalize

Reply by: SonicSounds

Date:5/22/2003 4:01:04 PM

Wow, a solid brick?!! I'll have to check that CD out. I need to get a list of songs that I've actually seen the waveform onscreen. I've never seen a waveform on a commercial CD go past zero.

So what was their excuse for this? That's ridiculous.

The volume I want to attain would be comparable to 4db at the peaks. That might seem a bit much but I think it's a decent volume. I just haven't been able to achieve that unless I turn up the master volume while mastering so that it peaks at 4db. In other words I want the loudness of 4db at the 0db range. Is this possible? Is that where maximizing comes in?

Come to think of it. Slum Village's new CD is mixed way to loud. On their song "Tainted" you can start to hear their vocals crackle a bit. I think it's due to the excessive highs in the mix. That's when it starts to get annoying.

Subject:RE: Compress & Normalize

Reply by: dkistner

Date:5/22/2003 4:44:28 PM

Another good link is Bob Katz' CD Honor Roll, which includes some tips for how to gauge loudness/effectiveness. The whole site has good information. I haven't bought the book yet, but I plan to.

Subject:RE: Compress & Normalize

Reply by: Vocalpoint

Date:5/22/2003 6:05:34 PM

Sponge,

In this case for a club where pounding it out is the standard of the day - I would think that maximizing would be a requirement to stay competitive to the next cut coming up.

I am not advocating to stay away from the tools..because in certain situations - they can really make a positive difference...like a club track or for special effect. That's the fun part about computer/digital audio - there are no rules.

I just try to apply some solid techniques across the board and go from there.

Cuzin B

Subject:RE: Compress & Normalize

Reply by: SonicSounds

Date:5/22/2003 11:18:00 PM

I've had the same experience. A couple of years ago when I wasn't as fluent in my mastering abilities, I had some of my material played at a local club. It just wasn't loud enough and the DJ had to compensate a little with the volume.

Cuzin B, I've been looking at the waveform onscreen of various singles. You're right in the respect that most of them go WAY past clipping. It's almost like engineers have redefined where 0db is.

My question is how does metering differ in the digital domain versus the analog? It seems like you can boost the signal way past 0db and still not get a ton of noticeable distortion, but in the analog world you can't go past 0db without some noticeable distortion.

In other words, is the definition of where audio clips different in the digital realm?

Subject:RE: Compress & Normalize

Reply by: RasKeita

Date:5/23/2003 12:49:07 AM

There are two schools of thought in the digital distortion debate. Some say never never allow clipping, while others say tear the roof off. Once again it's subjective, your ears your taste. I've found the VU meters in Acid to be inaccurate, when I pound the red I get no discernable distortion, wheresas if I pound them in Sound Forge I can hear clipping. As far as increasing over all volume. I normally use busses, that way I can push the individual source material and let the master control over all volume. Sometimes depending on the project and the production needs I use Waves Mastering Suite. That might be cheating...

http://www.waves.com/htmls/prods/masters.htm

Subject:RE: Compress & Normalize

Reply by: stuffedspacedog

Date:5/23/2003 5:09:22 AM

CuzinB, I guess it depends how the L1 is used... I don't have a straight line, I have a nice range of pointy peaks. If it is a loudness fixation, then it is one that's shared by almost everyone producing popular music these days... load any piece of music that's not classical or jazz into sound forge, and you see exactly the "straight line" you described... I'm actually arguing for preserving more of the dynamic range!

Subject:RE: Compress & Normalize

Reply by: Vocalpoint

Date:5/23/2003 6:40:05 AM

Sonic,

In digital - zero is it. It cannot be refined. If you go past 0.0db you are in nasty distortion land and you are actually losing data since anything past digital zero cannot be properly interpreted by the computer. If you are seeing clipping of any kind in your software - it would be best to back it off until NO red is present.

In analog - most guys WANT to go past zero. Properly done with good gear - +1 +2 +3 etc from analog tape has a compression and sound that is actually very pleasing most of the time. However sound scale wise....digital zero 0.0db and analog 0.0 are not the same thing...I believe that analog zero is actually -20db or something on the digital scale. I never bothered to do the math...I just stay out of the red at all times with my digital mixing.

Stuffed,

I hear ya on the L1. It can be used (and should be used) to fill things out a bit. But if you are hitting the "red brick" syndrome with this plugin...dynamic range is out the window.

For both of you guys - pop over to this site : www.digido.com and read up on Bob Katz and how the industry has gone from great sounding records up to around 1991 to today where the volume wars have pretty much killed a ton of promising recordings.

I learned a lot from here.

Cuzin B

Subject:RE: Compress & Normalize

Reply by: spesimen

Date:5/23/2003 9:51:33 AM

>>I just want to know how some tracks can be within
>>the -3db range and be loud while other tracks are in
>>that same range but aren't very loud.

Like CuzinB (or somebody) said, you should maybe try to read up on the science behind RMS values. This is the 'root-mean-square' and basically is an average of the volume of the track over time. Because of the psychoacoustics in our brains, the average strength of a signal is what we percieve as 'loudness', not the peak values which you actually see.

Another good article from the site dkistner recommended is "an integrated approach to monitoring, metering, and leveling practices." (also at www.digido.com)

and i think that the 1st tribe called quest album sounds awesome. as a dj, i prefer tracks that still have some dynamic range - they slam way better on a good system than ones where the kicks are all limited out and you just get that grinding hum.