Normalizing mp3s

Howell

Storage? I am Storage!
Joined
Feb 24, 2003
Messages
4,740
Location
Chattanooga, TN
Is it necessary to normalize my mp3s while I'm ripping them from CD? Is there any reason why I should not normalize to 100%. Is there a difference in normalizer programs? I use Audiograbber and the built in normalizer.

I'm simply trying to avoid massive volume adjustments while playing a playlist. Does is matter what the source material is (eg. classical, bluegrass, pop,...)?
 

Mercutio

Fatwah on Western Digital
Joined
Jan 17, 2002
Messages
22,303
Location
I am omnipresent
1. Audiograbber uses a LOUSY MP3 codec. I'd suggest moving to CDex + LAME for ripping directly to MP3. Or I use CDex to do recording to .Ogg.
I use EAC for ripping to .WAVs to make mix CDs.
2. If all the tracks are from the same disc, they presumably have been engineered to the same sound level, so there's really no need for normalization. If you always use the same program, and you always normalize your output, you also shouldn't have any problem.
3. Recording sound levels have changed over time, and are recorded at different volumes depending on the dynamic range of the music. It might surprise you to know that an album recorded in 2003 by Celine Dion is approximately 50% louder at the same volume as one recorded in 1978 by Black Sabbath. In general, newer Pop/Country albums will be louder than older ones (it's thought this is a result of more people listening to music with portable devices). In general, pop/country albums tend to have fairly constant, high volumes.
Music that relies on quiet passages as well as intense volume (jazz, classical, most action movie soundtracks) will be recorded at consistently LOWER volumes, so that the high volume passages have more impact.

In any case, your ears are the ones that have to be happy with the results of your work. I'd say that the default level of normalization in most ripping program I've used is good enough about 90% of the time, for my personal needs.
 

Howell

Storage? I am Storage!
Joined
Feb 24, 2003
Messages
4,740
Location
Chattanooga, TN
Mercutio said:
1. Audiograbber uses a LOUSY MP3 codec. I'd suggest moving to CDex + LAME for ripping directly to MP3. Or I use CDex to do recording to .Ogg.

AG was recently modified (It became freeware 2/04) to understand --alt preset standard. It also accepts external encoding programs.

I use EAC for ripping to .WAVs to make mix CDs.
2. If all the tracks are from the same disc, they presumably have been engineered to the same sound level, so there's really no need for normalization. If you always use the same program, and you always normalize your output, you also shouldn't have any problem.
3. Recording sound levels have changed over time, and are recorded at different volumes depending on the dynamic range of the music. It might surprise you to know that an album recorded in 2003 by Celine Dion is approximately 50% louder at the same volume as one recorded in 1978 by Black Sabbath. In general, newer Pop/Country albums will be louder than older ones (it's thought this is a result of more people listening to music with portable devices). In general, pop/country albums tend to have fairly constant, high volumes.
Music that relies on quiet passages as well as intense volume (jazz, classical, most action movie soundtracks) will be recorded at consistently LOWER volumes, so that the high volume passages have more impact.

In any case, your ears are the ones that have to be happy with the results of your work. I'd say that the default level of normalization in most ripping program I've used is good enough about 90% of the time, for my personal needs.

My concern is mostly for mixed playlists. I'd assume that within a CD the levels would either be consistent or I'd want them inconsistent (movie soundtrack). So what is your recommendation if I wanted a play list with say '78 Black Sabbath and '03 Celine Dion.

I'm only normalizing now because the technical desription sounded good, not because I actually tested it for effectiveness. :oops: Basicly, I want to listen to music and this encoding business is slowing me down. :)
 

Mercutio

Fatwah on Western Digital
Joined
Jan 17, 2002
Messages
22,303
Location
I am omnipresent
My recommendation is to find a program and stick to it,, and to pretty much always leave normalizing on. The results sound better if you do.
 

Howell

Storage? I am Storage!
Joined
Feb 24, 2003
Messages
4,740
Location
Chattanooga, TN
I just wanted to say again, that kind of response is exactly the kind I was looking for. Now I can slop thinking about this. :)
 

Jan Kivar

Learning Storage Performance
Joined
Feb 3, 2003
Messages
410
Mercutio said:
3. Recording sound levels have changed over time, and are recorded at different volumes depending on the dynamic range of the music. It might surprise you to know that an album recorded in 2003 by Celine Dion is approximately 50% louder at the same volume as one recorded in 1978 by Black Sabbath. In general, newer Pop/Country albums will be louder than older ones (it's thought this is a result of more people listening to music with portable devices). In general, pop/country albums tend to have fairly constant, high volumes.
Music that relies on quiet passages as well as intense volume (jazz, classical, most action movie soundtracks) will be recorded at consistently LOWER volumes, so that the high volume passages have more impact.
I thought that they mastered the newer discs louder as they would be able to obtain more dynamic range by doing so. CD-DA has dynamic range of 96 dB, but one can get "wider" range by shifting the zero dB position. Say, boost ten or 20 dB. One can't really hear that the sound never stops as normally the listening volumes are so low that the lowest levels are below the hearing threshold, or it's so loud that one simply can't hear anything so subtle. And the type/style of music has also an effect, as You pointed out. Trying to listen properly mixed/mastered classical music too quietly will lead in loss of the fine, subtle tinges.

Some, if not most radio stations compress the sound (speech, music) so that it would be easier to listen them in noisy environments, such as cars, while maintaining low overall sound levels.

Cheers,

Jan
 

Gilbolt

What is this storage?
Joined
Mar 5, 2004
Messages
7
Location
Halifax, NS
I would strongly suggest Replaygain.

It is by far the most powerful normalizing tool I have found, and I'm pretty serious about audio archiving.

Replaygain does not modify the files themselves (and so can be used on lossless audio without destroying the integrity of the archive). It works by applying a tag to the file which compatible players will recognize, adjusting their volume. There are plugins for several players. Foobar2000, which is far and away the most functional Windows player I have found, has native support for Replaygain tags (it's also free). I know that the lossless aspect doesn't concern you particularly considering that you're using mp3s but there are several other features of replaygain that are tremendously useful.

1. It preserves the internal dynamics of tracks (which most normalizers are supposed to do), but if they're altering the musical data itself who knows.
2. It offers the option to preserve dynamics within albums, while normalizing between albums. So tracks that are supposed to be quieter on an album are actually quieter --instead of every track being the same volume. But if you play a quiet track for a quiet album beside a loud track from a loud album the difference isn't obscene. This is a tremendous feature that is a necessity for me (at least). It is most important (really, it is necessary) for classical and jazz recordings, but is of value for every album since it allows you to listen to the disc the way it was meant to be heard. For example, many Rock Bands make extensive use of dynamics between songs with albums.

This last feature is really crucial. If you rip albums one by one, you normally normalize :)wink:) them after you rip them. This means that all the tracks in an album are normalized to eachother, but the albums themselves aren't normalized to eachother. What you really want is the opposite: to preserve the dynamics within an album, but ensure that the actual albums aren't out of whack with eachother (as is often the case as Mercutio aptly pointed out). Replaygain does both of these things, other normalizers do neither (if you rip your collection one disc at a time, as most humans do, or you don't like to re-normalize every file everytime new files are added --something most of us would prefer not to be bothered with).

Incidentally, I would suggest HydrogenAudio.org for audio information. It is a tremendously valuable resource. There is lots of info on Replaygain to be found over there.[/url]
 

Gilbolt

What is this storage?
Joined
Mar 5, 2004
Messages
7
Location
Halifax, NS
Jan Kivar,
I have heard somewhat the opposite asserted: that the reason newer disks were louder was because they were mixed to be played on the radio. I have noticed that many albums of radio friendly bands have a compressed dynamic range on the CD itself. This may just be the style of music though. It removes an interesting aspect of the music though, relative to the Jazz and classical discs in my collection. In years gone by the tracks were mastered for record play first, not for radio.
 

Splash

Learning Storage Performance
Joined
Apr 2, 2002
Messages
235
Location
Seaworld
Jan Kivar said:
...I thought that they mastered the newer discs louder as they would be able to obtain more dynamic range by doing so. CD-DA has dynamic range of 96 dB, but one can get "wider" range by shifting the zero dB position... ...Some, if not most radio stations compress the sound (speech, music) so that it would be easier to listen them in noisy environments, such as cars, while maintaining low overall sound levels...

Some of what you are talking about is dynamic range compression and some is audio normalisation.

"Normalising" audio is simply taking existing (presumably quieter) audio and shifting all the content's amplitude upwards so that the peak is now equal to 0 dB. There is no dynamic range compression. Dynamic range compression make the softest and loudest passages closer in volume than what the original content was. Radio stations (and television stations) do a lot of compression to their audio signals. In fact, they do advanced forms of compression where they have differing amounts of compression in different parts of the frequency range. There are also interesting "effects" that you can also do with audio compression. One is fooling around with the compression "knee" (knee relates to a graph shape) where you make loud audio softer than the slightly less loud audio of a given passage.
 

Mercutio

Fatwah on Western Digital
Joined
Jan 17, 2002
Messages
22,303
Location
I am omnipresent
Jan Kivar said:
I thought that they mastered the newer discs louder as they would be able to obtain more dynamic range by doing so...
Trying to listen properly mixed/mastered classical music too quietly will lead in loss of the fine, subtle tinges.

I sincerely doubt that there's anyone else here who listens to as much properly mastered classical music as I do. :)
A couple of perfect examples of "showpiece" classical works are Berlioz' Symphonie Fantastique and Tchaikovsky's 1812 Overture. In both pieces, the range goes from pianissimo to "triple f" Fortississimo (the direction for drums taking place of a cannonade in 1812). In both cases, audiophile recordings typically record the soft bits as softly as possible to maximize the dynamic change during poor Hector's march to the scaffold or the Russian triumph over Napolean.
The result? You turn your equipment up just loud enough to hear the soft bits and are blown out of your chair when the fireworks start. When you start screwing around with the floor volume, all you do is make the ceiling less interesting from a listening standpoint; when I listen to a "budget" recording of the 1812 overture, I've been known to miss the cannons completely... and I know exactly where they are.

Jan Kivar said:
Some, if not most radio stations compress the sound (speech, music) so that it would be easier to listen them in noisy environments, such as cars, while maintaining low overall sound levels.

Radio stations do a lot of funny things to sound. At ABC in Chicago, they had special machines that would take out "umms", stammers and long pauses between words so they could fit in more commercials, usually about 2 extra in every hour. They also had machines that normalized THEIR playlists as well.

While I'm thinking of it, A *HUGE* proportion of pop music is mixed for mono, rather than stereo play, even in 2004. Why? Because a lot of people hear music over PA systems, in elevators, or under less-than-ideal conditions (e.g. radio). There's very little channel seperation in pop music.
 

Howell

Storage? I am Storage!
Joined
Feb 24, 2003
Messages
4,740
Location
Chattanooga, TN
Gilbolt said:
If you rip albums one by one, you normally normalize :)wink:) them after you rip them. This means that all the tracks in an album are normalized to eachother, but the albums themselves aren't normalized to eachother. What you really want is the opposite: to preserve the dynamics within an album, but ensure that the actual albums aren't out of whack with eachother (as is often the case as Mercutio aptly pointed out).

Procedurally, how would this work?
 

sechs

Storage? I am Storage!
Joined
Feb 1, 2003
Messages
4,709
Location
Left Coast
You rip all of your albums. You then analyse and normalise all of the songs, as a group.

If you add music, you'll have to repeat the normalisation.
 

ddrueding

Fixture
Joined
Feb 4, 2002
Messages
19,747
Location
Horsens, Denmark
sechs said:
You rip all of your albums. You then analyse and normalise all of the songs, as a group.

If you add music, you'll have to repeat the normalisation.

I don't like the idea of constantly altering the content of my audio tracks. I'm not an Audiophile (256-320kbps MP3 is fine for me), but that just sounds like a good way to screw up your tracks.
 

Onomatopoeic

Learning Storage Performance
Joined
May 24, 2002
Messages
226
Location
LaLaLand
Mercutio said:
Radio stations do a lot of funny things to sound...

I've seen 19-inch 72-U equipment racks at top-$ radio stations (KRBE FM was one I've seen, though there are others) that were labeled by the time of day, with each containing different brands of processing gear along with different EQ settings. By the way, many of these pieces of gear in those racks were as expensive as some automobiles, and some you could not buy, but only rent (for some very dear rates). The engineer in the back office would simply patch in the correct rack when it came time. And, to top it off, some of these racks were duplicated and sitting in standby mode just in case there were technical problems! Think of this crap as the equivalent to modern food, think Processed Cheese Spread.

And, just to lead the anti audio processing charge, there was one radio station here (KLOL FM) that was advertising "We Don't Process Our Signals!" Sure enough, they sounded a LOT less stressful on the eardrums, which was rather ironic since they play mostly hard rock. I suspect they got sick and tired of pissing away $300K ~ $400K each year.

Another COMMON piece of niche audio signal processing gear found in a lot of places -- including radio and television stations -- are what's "de- essers." These get rid of the "ssss" sound in speech. The "ssss" sound just absolutely drives most microphones into a tizzy for a few milliseconds and you hear this annoying distorted sound called "sibilance."

Then there are all of these other sound processors, which are very proprietary, that add some indescribable ambience, warmth, or presence to the signal, and might even do their own form of sibilance control. I own a couple of these types of units. One is an Aphex Aural Exciter and the other a BBE Sonic Maximizer.


204Face700x063.jpg



482i-front.gif




Mercutio said:
While I'm thinking of it, A *HUGE* proportion of pop music is mixed for mono, rather than stereo play, even in 2004. Why? Because a lot of people hear music over PA systems, in elevators, or under less-than-ideal conditions (e.g. radio). There's very little channel seperation in pop music.

Wellllll.......... close. They normally only do a mono check to make sure that their stereo mix is compatible with mono. Some spatial effects can have enough phase shift between channels as to cause partial or significant cancellation problems when the Left and Right audio signals are summed. So, pretty much STANDARD PROCEDURE of any audio engineer during mixing is to do a Mono Check. And, just about as common -- and ALWAYS done with top hit prospects -- is to check your final mix using a selection of different speakers, of which some are nothing more than freeking 6x9 coaxial automobile speakers in a nice speaker cabinet!

But, the most popular way to handle this time consuming chore of listening to mixes through different speaker systems is to use a particular small 2-way Yamaha speaker (model NS-10, I believe) that was specifically developed for mixing purposes. This speaker was developed in the late 1970s to approximate the average sound of nearly all speakers and have not changed much at all -- if any -- in the past 20+ years. They are still highly popular to this day in most recording studios worldwide. Usually, you'll commonly see a little pair of these speakers sitting up close on top of the mixing console. But, during recording sessions, nobody listens to the music through these speakers. The engineers, producer, and musicians will instead listen to bigger, better, louder speakers at that point in time and/or through headphones.
 

mubs

Storage? I am Storage!
Joined
Nov 22, 2002
Messages
4,908
Location
Somewhere in time.
Onomatopoeic,

How good is the Aphex Aural Exciter? ~ 15 years ago I wanted to buy one, but didn't. I have a ton of one-of-a-kind live recordings made with poor equipment - mono portable cassete recorders - and the sound is horribly crummy. IIRC, the Aphex can put back harmonics that got lost during recordings like these. Any info on them would be greatly appreciated (performance, availability, pricing). Thanks.
 

timwhit

Hairy Aussie
Joined
Jan 23, 2002
Messages
5,278
Location
Chicago, IL
mubs said:
Howell, sorry for hijacking your thread (bangs head really hard against the wall).

I think Howell was satisfied with the answer he got way up at the top. So it probably isn't a problem. Unless Howell says it is.
 

.Nut

Learning Storage Performance
Joined
Jul 30, 2002
Messages
229
Location
.MARS
mubs said:
How good is the Aphex Aural Exciter?

It can work very well in certain situations. There are (and have been) many different types / classes of Aphex Aural Exciter during the past... ummm... 25+ years. We are basically up to "Type III" these days. I also have an older (mid '80s vintage) Type "C" as well that I was -- until recently -- using in my bass rack.

Of course, Aphex makes more than Aural Exciters these days. They make a line of compressors, noise gates, mic preamps, and many other 19-inch rack mount goodies. Even some floor pedals, now!


I have a ton of one-of-a-kind live recordings made with poor equipment - mono portable cassete recorders - and the sound is horribly crummy.

An Aphex Aural Exciter will not correct all those poorly recorded Snoop Doggy bootleg concert tapes of yours!

What the Aural Exciter lineup has been all about is taking a *good* -- but dull / lifeless / sterile -- audio source and essentially adding a proprietary mix of even-order harmonic distortion to "sweeten" the sound up along with a touch of multi-spectral dynamic range compression. The algorithms that the Aphex Aural Exciter use are closely guarded secrets. The Aphex Aural Exciter is a complicated processor in that the function it performs is not a fixed function, but an adaptive one that's driven by the content of its input source. You can tweak its fixed operating parameters to add or remove emphasis to certain functions.


IIRC, the Aphex can put back harmonics that got lost during recordings like these. Any info on them would be greatly appreciated (performance, availability, pricing). Thanks.

It can do the job of adding harmonics (actually, just harmonic distortion) that may or may not reflect reality. Sometimes, to the well-trained ear, you can detect use of Aphex Aural Exciter equipment. About the most over-the-top use of the Aphex Aural exciter that I recall on a popular recording is the patent overuse of it on Linda Ronstadt's voice on her mid- and late-70s recordings. To the discerning ear, those overtones on her voice sound artificial BECAUSE THEY ARE. Yep, good ol' Aphex at work -- probably ganged units (in series) at work. Yuck-o.

Pricing? Fortunately, they aren't at the level of having to mortgage the house these days. In fact, in the beginning, Aphex Systems was so paranoid about others getting one of their units and reverse-engineering it, that they would only rent them -- at phenomenal fees. Eventually, they began to sell them, for something like $50K each (!). Nowadays, depending on the model and number of channels, a few hundred $ to under $1K.

http://www.aphex.com/

 

.Nut

Learning Storage Performance
Joined
Jul 30, 2002
Messages
229
Location
.MARS
mubs said:
Howell, sorry for hijacking your thread (bangs head really hard against the wall).

So, I guess that a bunch of part-time photographers and audio engineer hang out here, correct? :)

 

Mercutio

Fatwah on Western Digital
Joined
Jan 17, 2002
Messages
22,303
Location
I am omnipresent
Pretty much.

How is the Aphex "super proprietary wonder harmonics" in comparison to the DSP Listen modes on your average $250 receiver? 'Cause it sounds to me like they do the same things.
 

Howell

Storage? I am Storage!
Joined
Feb 24, 2003
Messages
4,740
Location
Chattanooga, TN
ddrueding said:
sechs said:
You rip all of your albums. You then analyse and normalise all of the songs, as a group.

If you add music, you'll have to repeat the normalisation.

I don't like the idea of constantly altering the content of my audio tracks. I'm not an Audiophile (256-320kbps MP3 is fine for me), but that just sounds like a good way to screw up your tracks.

The nice thing about the method Gilbolt describes is that it does not alter the waveform (the music) but it does add a flag to the file.

One downside I can see is that as the music collection grows it becomes less and less convenient to add music to it. The flag in the file can be reset without distubing the waveform. Another downside is that the player must understand the flag. If the player doesn't understand the flag it will not be normalized.
 

.Nut

Learning Storage Performance
Joined
Jul 30, 2002
Messages
229
Location
.MARS
Mercutio said:
How is the Aphex "super proprietary wonder harmonics" in comparison to the DSP Listen modes on your average $250 receiver? 'Cause it sounds to me like they do the same things.

Well, since I don't know about these receiver's "DSP Listen" modes, I'd venture a guess at those being fairly straightforward compressors. The Aphex circuitry is complex.

By the way, don't confuse Aphex Systems, Inc with the laughable nutcase artist Aphex Twin (it’s not like you possibly could):

B000002HOF.01._PE_SCMZZZZZZZ_.jpg
B000002HIK.01._PE8_SCMZZZZZZZ_.jpg
B000003MSH.01._PE_SCMZZZZZZZ_.jpg
B00000I8U7.01._PE_SCMZZZZZZZ_.jpg




apxtwin.jpg
apxtwin1.jpg


Ironically, Mercutio, here's Aphex Twin's and Philip Glass' collaboration deom about 1993 or so!
 

.Nut

Learning Storage Performance
Joined
Jul 30, 2002
Messages
229
Location
.MARS
...collaboration deom about 1993 or so!

FROM

Jeez! I've slipped back to no good with my typing lately! I *keep on* dropping 1 character every once in a while in a word (typically "n" or "a" it seems). Now this! ok...
 

sechs

Storage? I am Storage!
Joined
Feb 1, 2003
Messages
4,709
Location
Left Coast
Howell said:
The nice thing about the method Gilbolt describes is that it does not alter the waveform (the music) but it does add a flag to the file.

Generally, it's a really great idea. Problem is, it requires support.
 

Howell

Storage? I am Storage!
Joined
Feb 24, 2003
Messages
4,740
Location
Chattanooga, TN
sechs said:
Howell said:
The nice thing about the method Gilbolt describes is that it does not alter the waveform (the music) but it does add a flag to the file.

Generally, it's a really great idea. Problem is, it requires support.

Yes I'm considering getting a phatnoise phatbox for the car. I'm annoyed it doesn't support the flag.
 

Gilbolt

What is this storage?
Joined
Mar 5, 2004
Messages
7
Location
Halifax, NS
I think there's some confusion with what I said about ReplayGain. You don't have to redo anything when you add music. You just normalize the new music. If you want your collection to be normalized with other normalizers you have to renormalize everything when you add new music --if you want everything to actually be normalized that is. Replaygain eliminates the need for this.

And sechs is right. The problem is it requires support. Like I said though, Foobar2000 is terrific and has support built in and it's free. There are Winamp plug ins to.
 
Top