Normalize: Peak or average?

[st3333ve]

1. When EMC 7 "normalizes" a CD as part of the burning process, is it simply normalizing the peaks or does it normalize in a more complicated way that involves average volumes (or something similar)? My understanding is that the former has a negligible impact on sound quality and preserves the dynamic range within each track, whereas the latter, while it can be more effective in terms of the real "perceived volume" of the tracks, also involves compressing dynamic ranges and can have a more substantial impact on the sound quality.

 

2. If EMC's "normalization" is "peak" normalization, does it normalize the peaks to 100% (i.e., 0db) or to a slightly lower level (e.g., 98%)?

[st3333ve]

I'm replying to my own post because I subsequently created a WebTicket and received an answer today from Roxio. In case anyone is ever searching this forum about this issue in the future, I thought it would make sense to post that answer here.

 

According to Roxio Technical Support, Roxio's "normalize" feature does what, as I understand it, is the simplest and most common kind of volume leveling (which is what I'd been hoping). For each separate audio file, Roxio (to quote Tech Support) "will calculate the difference between the peak volume level for that file and 0db, and then amplify the entire audio file by the amount of that difference."

[cdanteek]

st3333ve

 

Thank you for that information!

 

cdanteek

[st3333ve]

Today I burned my first audio CD using Roxio's "normalize" feature, and I regret to report that the response I got from Roxio Technical Support (as summarized in my previous post in this thread) is not correct.

 

The audio CD was burned from .WAV files with "Normalize audio (applied at time of burn)" selected (in the Project Settings dialog). After burning the CD, I ripped 7 of the tracks from the new CD to new .WAV files (using Exact Audio Copy) and then compared the new .WAV files to the old .WAV files.

 

Here are the before and after "peak" volume levels (in db) for each track:

 

1. -3.0 --> -3.6

2. -0.0 --> -4.2

3. -0.0 --> -5.5

4. -3.3 --> -3.3

5. -5.8 --> -0.6

6. -4.0 --> -4.8

7. -4.6 --> -0.7

 

As you can see, Roxio "normalized" Track 3 by cutting its peak volume level almost in half (from -0.0db to -5.5db), while "normalizing" Track 5 by almost doubling its peak volume level (from -5.8db to -0.6db).

 

Now, I suspect there's a method behind this apparent madness, because the resulting CD does in fact play with what sounds like a reasonably consistent volume level -- so I assume Roxio's "normalization" has probably somehow leveled the volumes with a focus on something more like the "average" volume level of each track (rather than the peak). But whatever method Roxio used to achieve this, it's clearly not the method described by Technical Support in its response to my WebTicket (not even close).

 

So I'm afraid I remain in the dark as to my main concern -- that is, I don't know if Roxio's "normalize" method (whatever it may be) potentially modifies the dynamic range of the tracks (which I'd rather avoid) or otherwise manipulates the tracks in a way that has a greater potential to degrade the sound quality than a simple uniform amplification of the track's volume level.

 

If anyone has any further information to offer, I'd love to hear it.

[sknis]

Today I burned my first audio CD using Roxio's "normalize" feature, and I regret to report that the response I got from Roxio Technical Support (as summarized in my previous post in this thread) is not correct.

 

The audio CD was burned from .WAV files with "Normalize audio (applied at time of burn)" selected (in the Project Settings dialog). After burning the CD, I ripped 7 of the tracks from the new CD to new .WAV files (using Exact Audio Copy) and then compared the new .WAV files to the old .WAV files.

 

(deleted some text)

 

Now, I suspect there's a method behind this apparent madness, because the resulting CD does in fact play with what sounds like a reasonably consistent volume level -- so I assume Roxio's "normalization" has probably somehow leveled the volumes with a focus on something more like the "average" volume level of each track (rather than the peak). But whatever method Roxio used to achieve this, it's clearly not the method described by Technical Support in its response to my WebTicket (not even close).

 

So I'm afraid I remain in the dark as to my main concern -- that is, I don't know if Roxio's "normalize" method (whatever it may be) potentially modifies the dynamic range of the tracks (which I'd rather avoid) or otherwise manipulates the tracks in a way that has a greater potential to degrade the sound quality than a simple uniform amplification of the track's volume level.

 

If anyone has any further information to offer, I'd love to hear it.

 

Beauty is in the eyes of the beholder. Music is in the ears of the listener (not as poetic). You have listened to both --how do they sound? If you can't tell the difference then you have accomplished what you set out to do -normalize several music tracks until they sound about the same volume.

 

If you want to do more perhaps a dedicated sound editing program would be what you need. Download the free trial of GoldWave and see if it has something for you. You can normalize in that program three different ways.

[vid2man97]

I'd have to agree with sknis...if it sounds good to your ears (the best judge of music), then you're ok.

 

As far as "normalizing", it's always been my understanding that you are averaging each track in comparison to the other tracks. In that way, one track doesn't blare at you during playback while the next one begs for the volume to be turned up and so on. It's a compromise that few people would notice.

 

If you aren't happy, sknis suggestion to try Goldwave is a good one. have fun

[st3333ve]

Thanks to sknis and vid2man97 for your input, but your aims clearly aren't mine. There's no question you could turn 10 different sound engineers loose on the master tapes for my 10 favorite CD's and each engineer's approach would be somewhat different, and many of the results might "sound good" to me. But no, thanks, I'd still want the original CDs back.

 

The audio files I'm working with are from old tapes that will at some point become unplayable, and I want to preserve them in their "original" state as much as possible -- e.g., with their existing dynamic range, rather than with some amount of compression that may "sound good" but not be what the artist released.

 

If Roxio was willing to tell us what its "normalize" feature does, then we could all make a better-informed decision about whether (or when) to use it, regardless of our different aims, right? But it appears I may have set my hopes too high.

[ogdens]

Thanks to sknis and vid2man97 for your input, but your aims clearly aren't mine. There's no question you could turn 10 different sound engineers loose on the master tapes for my 10 favorite CD's and each engineer's approach would be somewhat different, and many of the results might "sound good" to me. But no, thanks, I'd still want the original CDs back.

 

The audio files I'm working with are from old tapes that will at some point become unplayable, and I want to preserve them in their "original" state as much as possible -- e.g., with their existing dynamic range, rather than with some amount of compression that may "sound good" but not be what the artist released.

 

If Roxio was willing to tell us what its "normalize" feature does, then we could all make a better-informed decision about whether (or when) to use it, regardless of our different aims, right? But it appears I may have set my hopes too high.

 

 

 

 

What do you want for $80 USD.

 

 

If you really HAVE to know, continue with the Tech support.

 

Sonic/Roxio is really not into audio, their main market target is DVD Video.

 

Maybe you could search the net a fined a program that meets YOUR requirements .

 

Good luck.

[vid2man97]

I can understand wanting perfection. When it comes to photos, I settle for nothing less. Obviously for you it's music.

But as Odgens suggests, you probably would be better off searching for a dedicated audio program. If you find one, ask the questions first before you put out any money. That way, if it doesn't deliver, you won't be throwing more away (money and time).

 

good luck in your quest to preserve the music.

 

drop in some time to let us know which program works out best for you...there might be someone else around here in a similar predicament and could benefit from you sharing in success

[sknis]

I can understand wanting perfection. When it comes to photos, I settle for nothing less. Obviously for you it's music.

But as Odgens suggests, you probably would be better off searching for a dedicated audio program. If you find one, ask the questions first before you put out any money. That way, if it doesn't deliver, you won't be throwing more away (money and time).

 

good luck in your quest to preserve the music.

 

drop in some time to let us know which program works out best for you...there might be someone else around here in a similar predicament and could benefit from you sharing in success

 

St333ve,

 

 

I agree there are better programs to preserve your music in as pristine condition as possible; however there is a limit. First, the original analog recordings you have were not perfect in capturing the live music. The recordings have aged. Second, there is no way you can preserve all the warmth and nuances of the analog recording as a digital recording especially with a home studio regardless of the format. Third, you want to manipulate the sound so that the volumes are all the same (normalizing). I understand that you want to preserve for posterity. I would spend my time and effort preserving the original records and tapes in the best possible conditions. You already have something that sounds as good (almost) as the original. Listen to that and perhaps in a few years technology and computers will improve to the point where you can do more in capturing the analog music.

 

For images, you can capture the scene with an film or digital camera, you can transfer it to a computer and adjust the images even to the same brightness. There is no substitute for being there and those images will never be as good as the original --actually being there.

Save your money and spend it to go to live concerts. Did you try GoldWave as I suggested?

[vid2man97]

For images, you can capture the scene with an film or digital camera, you can transfer it to a computer and adjust the images even to the same brightness. There is no substitute for being there and those images will never be as good as the original --actually being there.

 

So true...photography with any medium is just an attempt at a representation of reality. You can only hope to duplicate it as faithfully as possible, be happy with it, and move on to the next challenge.

You're a wise man sknis.

[sknis]

So true...photography with any medium is just an attempt at a representation of reality. You can only hope to duplicate it as faithfully as possible, be happy with it, and move on to the next challenge.

You're a wise man sknis.

 

Compliment graciously accepted.

[st3333ve]

Yow. Thanks for all the philosophy, fellas. It doesn't really apply to me, but maybe some of the other readers will benefit from your wisdom.

 

I'm not looking for "perfection" or anything close to it. (That's certainly not what's on the tapes I'm converting.) As I understand it from my web-surfing, different audio programs take different approaches to leveling the volume on a mixed CD. Some preserve the dynamic range at all costs, even if it means some of the tracks end up sounding quite a bit softer than others. Others reflect an approach that says it's worth sacrificing some dynamic range (and maybe a bit of sound quality) to make the volume levels more even. Since I'm definitely in the former camp (and I think Roxio is, too, by the way, judging by some further examination of my "before" and "after" samples), I'd want to avoid Roxio's "normalize" feature if it was going to sacrifice any dynamic range (or sound quality) for the sake of leveling the loudness.

 

I'd say it's a matter of prioritizing between two possible imperfections, not seeking perfection.

 

As they say in Morocco: If you see him riding on a bamboo cane, say to him, "Good health to your horse."

[d_deweywright]

Yow. Thanks for all the philosophy, fellas. It doesn't really apply to me, but maybe some of the other readers will benefit from your wisdom.

 

I'm not looking for "perfection" or anything close to it. (That's certainly not what's on the tapes I'm converting.) As I understand it from my web-surfing, different audio programs take different approaches to leveling the volume on a mixed CD. Some preserve the dynamic range at all costs, even if it means some of the tracks end up sounding quite a bit softer than others. Others reflect an approach that says it's worth sacrificing some dynamic range (and maybe a bit of sound quality) to make the volume levels more even. Since I'm definitely in the former camp (and I think Roxio is, too, by the way, judging by some further examination of my "before" and "after" samples), I'd want to avoid Roxio's "normalize" feature if it was going to sacrifice any dynamic range (or sound quality) for the sake of leveling the loudness.

 

I'd say it's a matter of prioritizing between two possible imperfections, not seeking perfection.

 

As they say in Morocco: If you see him riding on a bamboo cane, say to him, "Good health to your horse."

Well, let me wade in here. You're priority is to retain the closest digital reproduction of the original that you can. I'll assume that your tapes are not "originals" that is, they are "mixes" from other tapes and or LPs? If they are original, purchased tapes, then I wouldn't think you would want to do any normalizing, and you would want to keep the original levels as they were engineered.

 

So, my solution is a two part approach. First, capture what you have, as it is, and make a copy to CD either as an Audio CD, or as .WAV files on a Data CD, or both. Now, if you still want to normalize, I would pick up something like GoldWave, as previously suggested. Now you can manually adjust levels, without any compression, to balance the sound as you see fit. Remember, if you have to lower the level of any tracks, you're reducing the dynamic range, but proportionally, without any compression. Get the tracks balanced as you see fit, and then write your Audio CD. (Of course, all of this would be done using .WAV files, not .MP3 files to avoid loss during the conversion to the compressed audio format.)

 

You still have the original copy to work with in the future, if you want, and you have a copy with balanced tracks, knowing how it was done.

 

Hope that helps!

[st3333ve]

Thanks for the input. I was already planning to create a 2nd set of "archival" data CDs (with the .WAV files) as a backup. And I have Audacity, which will let me manually "normalize" on a one-file-at-a-time basis -- although fellow Audacity users should note that you don't want to use Audacity's "normalize" feature for that, since Audacity's "normalize" simply cuts the peak to 70% and is intended to prepare a track for mixing with other tracks, not for burning to a CD. (Use Audacity's "amplify" effect instead.)

 

But I was hoping to save time and effort by using a more automatic "normalize" effect (like Roxio's) that took care of the whole disc at once without any file analysis or fiddling on my part.

 

As a final note, when I talk about preserving the dynamic range, I'm talking about the proportional difference between different moments within a track. My understanding is that if you amplify an entire track by the same db amount (positive or negative), the dynamic range (in this sense) is preserved. Based on my examination of some "before" and "after" samples, I believe Roxio's "normalize" feature does preserve the dynamic range (in this sense), but I'm no sound expert, so I'd still appreciate a brief (and accurate) description from Roxio (or any Roxio Community member who knows) of what Roxio's "normalize" feature does to the audio files.

[d_deweywright]

Thanks for the input. I was already planning to create a 2nd set of "archival" data CDs (with the .WAV files) as a backup. And I have Audacity, which will let me manually "normalize" on a one-file-at-a-time basis -- although fellow Audacity users should note that you don't want to use Audacity's "normalize" feature for that, since Audacity's "normalize" simply cuts the peak to 70% and is intended to prepare a track for mixing with other tracks, not for burning to a CD. (Use Audacity's "amplify" effect instead.)

 

But I was hoping to save time and effort by using a more automatic "normalize" effect (like Roxio's) that took care of the whole disc at once without any file analysis or fiddling on my part.

 

As a final note, when I talk about preserving the dynamic range, I'm talking about the proportional difference between different moments within a track. My understanding is that if you amplify an entire track by the same db amount (positive or negative), the dynamic range (in this sense) is preserved. Based on my examination of some "before" and "after" samples, I believe Roxio's "normalize" feature does preserve the dynamic range (in this sense), but I'm no sound expert, so I'd still appreciate a brief (and accurate) description from Roxio (or any Roxio Community member who knows) of what Roxio's "normalize" feature does to the audio files.

For what you're after, yes, the "proportional difference" is as good as you'll get. Remember, of course, that there are rounding errors whenever you change the volume level, unless you're doing it by an integer amount (2 times, 3 times, etc.) because samples are integers. So, decreasing the volume to .9 of its current level will cause rounding to occur on most samples, as would increasing by 1.23. On average, it'll be okay, but any change (except as noted above) will cause errors.

 

As for getting a better description of how Roxio's normalization works, you probably got as good an answer as you'll get, which unfortunately, was wrong. 99.99% of the folks on these boards are users like yourself. You got an answer from a real Roxio person, and they had it wrong. But, if you think your analysis of the way EMC is actually doing the normalization is correct (where it maintains the proportional levels) and you can't hear any difference between that, and what you get with Audacity, you may want to keep using EMC for your normalization.

 

Good luck!