Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: What some audio "engineers" think about loudness normalizati (Read 4271 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

What some audio "engineers" think about loudness normalizati

So, this is a bit of an unusual post for me, because I normally frown upon name-and-shame sessions identifying ignorance on other forums. This situation, though, is different. First, if I truly spoke my mind on those forums I'd get my ass banned, and I don't want that, so I'm venting here. Second, this whole situation was truly surprising to me, so I figure it would surprise some people here too. (And please, no inter-forum wars, I'm not looking for a posse.)

On a couple different forums, I have noticed several audio professionals decrying the use of loudness normalization (ReplayGain, SoundCheck, ITU-R BS.1770-based corrections). A few people have legitimate complaints: that reducing the loudness evaluation to a simplistic computation opens up the potential for abuse; that efforts to legally mandate such normalization are wrongheaded; etc. The latter in particular is a sort of hot-button issue, with Congress seriously considering mandating LN for some media, and with Bob Katz defending, and perhaps drumming up support for, such mandates (in one of the threads below).

But a substantial portion of the opposition doesn't like LN because, in their view, it sounds bad. YA, RLY. Many mainstream, non-audiophile audio engineers believe that a digital volume control changes the sound, and methinks not in the same way that they sometimes confuse peak limiting with LN, or argue about digital mixing implementations in Pro Tools.

A related, equally popular, and equally wrongheaded and pernicious view being represented is that LN compromises artistic freedom. Because really, it's a good thing when listeners need to adjust the volume. It means they're in control.

http://recforums.prosoundweb.com/index.php/t/29101/16937/

Quote
Thought that clients had screwed up the wav's before uploading them prior to conversion, but it felt so out of question since they have been so satisfied with my results. So I digged a little deeper and then found the problem: "Enable volume normalization" under a preference menu. Deactivate it and then things were just fine. Tried to find a little information about this shit and it turned out to be a new silly attempt to make everything sound similar in level. Contacted a few clients telling this and they just went "what the f**k!?", obviously very concerned about it. Also found that many major releases which I consider really good sounding, though "hot", were being dragged down to about 60% or so in level. Big budget recordings being treated like crap. How smart is it to lower a mp3 before possibly hitting an already lowered digital fader before hitting the low-res DA... I hate this normalization, replay gain(?), K-system, RMS, whatever shit there is. They are all silly attempts to create something impossible - software hearing.  F**k everyone trying to make "technology" like this happen. Those people are stepping miserably bad into any musicians craft.


That poster has a lot more to say on that thread, and the following one - the quote above is the only one I pulled from him though, everything below is from different people. Some quotes put into a list to avoid the 7-quote limit:

http://www.gearslutz.com/board/mastering-f...udness-war.html

  • "On a funny sidenote, many A&Rs, clients etc. listen to Mixes/Masters/Roughmixes via iTunes with the LN-thing dialed in!  "
  • "I think it is a horrible idea to force anyone to limit/or reduce volume or do anything with their record. That is the whole point......that this is their record. What's next, are you going to tell Picaso that cubism isn't round enough? Or stop middle aged men from buying a shiny new red camaro?"
  • "that my point dear mister BK, IMO it is not about what *you* think is fair. it is about artists and/or entrepreneurs making up their own mind and not having their business or art interfered with by some arbitrary (yes, arbitrary) standard."
  • "And one limiter to rule them all... puh-leaze. Terrible idea."


Quote
pleeeeeeease! LN impacts the sound... a lot. This impacts the intentions of the artist.. a lot. What's next? transposing the key? Adding distortion? I would be ok if the reproduction system had some sort of loudness adjustment (it odes). t


Quote
Like Beethoven needs to compete with Metallica. And that is the album setting then. In single setting it gets even worse, because the system will mess up the balance between songs based on some loudness detection algorithm.  So, it can not be loud, it can not be quiet, it can not be dynamic between songs/movements/parts. And all that to allow poor Beethoven (artist that doesn't need ANY support) and poor Norah Jones (entertainer that hardly needs support) can "compete" with Metallica, and, of course, so the couch potato can have his music in pre-chewed, half digested, lukewarm, glorious ALL equal loudness blandness without even having to use his remote. In the name of art. Seriously.


Quote
The idea to just use the knob makes most sense. Or hire a good dj, with ears, to do it for you.  And also, the idea to not moderate everyone's music with arbitrary standards makes sense. But if you do, don't say it is in the name of freedom of expression. Say that it is because you know what is best, and will make that the default, overruling what artists freely chose for. I can see why you want it, but your reasoning is flawed. You cannot moderate art to give artists more freedom.


Quote
"In classical music, SINGLE setting would cause the second movement of a Beethoven symphony to be as loud as the first. ALBUM setting would permit the entire Beethoven symphony to compete with the rest of the world, including rock. So Beethoven would appear (nearly) as loud as Metallica." Ooops. really? And in the pauses between the movements with room sound we have an audible level jump? Please do not speak for the classical world. I think such a system is just unacceptable and unnecessary. I know several classical musicians that would send the men in black to your house, if they knew what you are suggesting here.  Why should the second movement be as loud as the first? From he point of view of a (classical) musician this is a completely unacceptable suggestion. There are producers and engineers that carefully balance a classical production according to the intended dynamic of the score and interpretation. The only thing that should be between them and the listener is the listeners volume knob. Anybody else stay out of it.


Quote
"The artist does not now nor have they EVER been able to force you to listen to anything at a certain level. The listener has been and will always be able to determine their own playback level." Of course they are not able to force anything. which is exactly why there is no need for a default on (or any) LN system. Artists are free to use the medium the way they want, and should be. And someone in the entertainment biz is free to crush 20 dB if he thinks that will help sales.


Quote
"... 16 dB of difference between 2009 and 1980 there may only be 2 or 3..." Too much manipulation. It takes a lot of skill to get something 6 or 8 dB louder and still have some definition left. I can't see how an automatic system can do this amount of "correction" without destroying it.  The real thing is that the more I think about it (and read other opinions) the less I like the idea.




Soooo.... yeah. Mastering engineers apparently have trouble understanding multiplication and division. And an automated volume control compromises their skills and the artists' intent... if they had any skill and intent in the first place, that is.

My point is: there are a lot of "real" audio professionals who do not understand their own craft, and a lot of so-called audio engineers who couldn't describe the operation of BS.1770 (much less implement it) if the specification paper-cut them them in the eye. And the misconceptions are not just in the "I don't know about ABX" or "I don't know the intricacies of lossy codecs so I think all MP3s sound as bad as 128k CBR" sorts of ways. To think that LN compromises music, in the fashion being described here, seems to betray a profound lack of comprehension as to how people actually listen to music. (You have no idea how hard I'm biting my lip right now to avoid making an incredibly insulting comment about who a "real" audio engineer is.)

I guess that's not really news, but I tend to be pretty nonjudgmental by default about the competency of most disciplines, and for better or worse, this is really making me start to question that. It's like a lot of these people don't even listen to music.

What some audio "engineers" think about loudness normalizati

Reply #2 –
Quote from: Axon on
I guess that's not really news, but I tend to be pretty nonjudgmental by default about the competency of most disciplines, and for better or worse, this is really making me start to question that. It's like a lot of these people don't even listen to music.

Please correct me if I am wrong but you seem to be taking samples from posts to web forums. How representative do you think is of the views of the industry?

In the field I can judge best (not audio), the posters are enthusiastic but inexperienced youngsters, a small number of retired people and a significant number of relatively unskilled/untrained people on the fringes. The vast majority of the professionals successfully getting on with their job do not read the forums or even know they exist in most cases.

What some audio "engineers" think about loudness normalizati

Reply #3 –
Quote from: honestguv on
Please correct me if I am wrong but you seem to be taking samples from posts to web forums. How representative do you think is of the views of the industry?

In the field I can judge best (not audio), the posters are enthusiastic but inexperienced youngsters, a small number of retired people and a significant number of relatively unskilled/untrained people on the fringes. The vast majority of the professionals successfully getting on with their job do not read the forums or even know they exist in most cases.

Good point, but my impression is that usually, the forum people are among the better-read people: they ought to be underrepresenting the people with major misconceptions, not overrepresenting.

Many of these people agree with the usual narrative of the Loudness War, so it isn't like they are just moths to the flame (war).

What some audio "engineers" think about loudness normalizati

Reply #4 –
Quote from: Axon on
Good point, but my impression is that usually, the forum people are among the better-read people: they ought to be underrepresenting the people with major misconceptions, not overrepresenting.

I am sorry but this is bonkers. In the field I can judge best, the views expressed online are at best naive and are quite unrepresentative of the views held by mature, experienced, successful professionals. There are perhaps thousands of home audio forums on the web and this web site is unrepresentative in being hard up against the "common sense" end. And yet I would suggest the average post here has quite a long way to go in terms of "common sense" if it is to be representative of the views held by trained and informed professionals producing hardware for home audio. So what does the average post on an audiophile forum look like?

Quote from: Axon on
Many of these people agree with the usual narrative of the Loudness War, so it isn't like they are just moths to the flame (war).

Can I suggest that people that are complaining about the compression of modern commercial popular music may have lost the plot. What is the purpose of modern commercial popular music? Put yourself in the position of a consumer of commercial popular music listening on a cheap mp3 player with cheap earbuds while jogging, commuting on a noisy train or whatever.

What some audio "engineers" think about loudness normalizati

Reply #5 –
Of course, that LN would sound worse is crap from a technical point of view (edit: at least when you sufficiently increase the output bit depth, e.g. 16->24 bit). Still I do understand some of their discomfort. Good 'heating up', i. e. being louder than the competition, is far from trivial. Of course I don't want some retard to reduce the dynamic range of a good Jazz or classical recording to 5 bits. But there are other genres where hot (minimal) dynamic range is part of the equation. Look at the producer Timbaland, for example; I can often identify pieces, that he has had his hands on, by his specific signature of 'fatness'. If his tracks got brutally leveled down to sound just as loud as simple conservative forms of mastering, where all dynamic range is preserved, his tracks would sound like shit in comparison. You might argue, that's what he deserves for crippling music. But that's not what he is paid for. He is paid for producing the hottest while optimally least crippled sound in a succession of music videos, and that's where he excels. Automatic LN adjustment, if it was implemented for broadcast, would seriously limit his individual value as a producer. So it is especially in the top league producers' interest to fight this.

What some audio "engineers" think about loudness normalizati

Reply #6 –
I'm not sure if everybody means the same thing when they say "Loudness Normalization".    If they are talking about what Dolby calls "Dialog Normalization", or about the compression that radio stations use to get "constant loud volume", then yes,  that alters the dynamics of the sound, and it's evil!   

If they are talking about ReplayGain, that's just a volume control and unless you're in a movie theater or an accurately calibrated home theater, there is no "correct" setting for the volume control!  (In the "track" mode, ReplayGain can mess-up the song-to-song differences, but that does not happen in the "album" mode.)

Quote
...non-audiophile audio engineers believe that a digital volume control changes the sound...
Digital volume controls affect the sound in almost the same way that analog volume controls affect the sound!  When you reduce the volume digitally, you reduce the "effective resolution" because all of the bits are not used.  Something similar happens when you reduce the analog volume.  The signal gets reduced, which means that you (potentially) get a reduced signal-to-noise ratio.

There is generally no problem when you reduce the volume, because any negative effects are at the lowest levels anyway.  (When you reduce the size of an image on your screen, you don't usually see the loss of resolution... In fact, it might look sharper, even though there are fewer pixels.) 

However, if you later re-amplify the sound, then you may notice the loss of quality/resolution.  (Just as you will notice the loss of resolution when you blow-up an image that you have previously reduced.)

What some audio "engineers" think about loudness normalizati

Reply #7 –
Dynamic compression and loudness normalization are related psycho-accoustically. When you apply a linear gain to a song and let someone ABX it to itself, depending on the amount, the listener might describe the louder version with attributes as "better bass" etc. instead of correctly identifying the linear gain. We see this regularly with non level matched tests.

So a song, which has dynamic compression applied to sound fatter than the competition, looses its "advantage" when it is attenuated by ReplayGain & Co. Worse, it sacrificed large amounts dynamic range for something, that isn't even there anymore in its intended extend. Of course, some 'fatness' will remain, but a more conservatively mastered 'rival track' may suddenly sound better to the average (loudness war agnostic) ear.

What some audio "engineers" think about loudness normalizati

Reply #8 –
This is probably a bit obvious, but some of the comments in Axon's OP don't seem to understand the difference between Track Gain and Album Gain--I'm thinking of the people talking about flattening the dynamics in a Beethoven Symphony. (BTW, Ludwig van is an interesting name in this context, what with going deaf and beating pianos to death.)

They also don't seem to be clued up on how a lot of people listen to music a lot of the time: if you're on Shuffle, you don't want to have to grap for the volume control when you suddenly come to some item that's had the MegatonMix ™ treatment applied. Especially if the volume control is on a DAP in your pocket under your coat.

There also seems to be a group who don't think of an amp that goes to 11 as a joke.

What some audio "engineers" think about loudness normalizati

Reply #9 –
Quote from: DVDdoug on
Digital volume controls affect the sound in almost the same way that analog volume controls affect the sound!  When you reduce the volume digitally, you reduce the "effective resolution" because all of the bits are not used.  Something similar happens when you reduce the analog volume.
Even today? I find this hard to believe. Given that most digital volume controls nowadays are working with 32 bit floating point you have a bit reservoir that should be huge enough for even low volumes.

Quote from: doneysr on
K-system is not a levelling device or method. It is simply a proposal to normalise the way we talk about monitoring level and headroom.
It is a protocol for setting mix and monitor calibrations in a studio environment.
From what I´ve read about K-System it seems to be a good proposal for standardizing monitoring level and headroom. I also might add that Bob Katz who started the discussion at Gearslutz.com correctly distinguished between normalization and replay-gain. Other people in the thread seemed to get it wrong all the time - shame on them. But he also talked about the Soundcheck feature of iTunes and that it destroys the sound by applying destructive compression and introducing artifacts. Is that true? I seldomly use iTunes so I can´t comment on this.
marlene-d.blogspot.com

What some audio "engineers" think about loudness normalizati

Reply #10 –
Both Replay Gain and iTunes' SoundCheck can cause dynamic compression. It happens if a fixed output gain is used, which may push formerly quiet tracks into digital clipping. Many Replay Gain implementations let you either choose a hard limiter or limited overall output gain to prevent this. Apple chose to keep it simple and applies hard limiting to tracks, that have peaks exceeding 0db after SoundCheck correction, and did not offer an option to lower the output gain value. So in practice you cannot prevent compression of very quiet tracks in iTunes as long as you have SoundCheck enabled.

What's the default in Foobar for Replay Gain BTW? I think it is also hard limiting. The gain adjustment method only works after scanning your collection for the highest peak value and then lowering your global output gain until clipping is prevented for every track.

What some audio "engineers" think about loudness normalizati

Reply #11 –
Quote from: Axon on
... with Congress seriously considering mandating LN for some media, and with Bob Katz defending, and perhaps drumming up support for, such mandates (in one of the threads below).
I had no idea about this - thanks for mentioning it Axon. I'll try and read more about it.

Quote
http://recforums.prosoundweb.com/index.php/t/29101/16937/

Quote
Thought that clients had screwed up the wav's before uploading them prior to conversion, but it felt so out of question since they have been so satisfied with my results. So I digged a little deeper and then found the problem: "Enable volume normalization" under a preference menu. Deactivate it and then things were just fine. Tried to find a little information about this shit and it turned out to be a new silly attempt to make everything sound similar in level. Contacted a few clients telling this and they just went "what the f**k!?", obviously very concerned about it. Also found that many major releases which I consider really good sounding, though "hot", were being dragged down to about 60% or so in level. Big budget recordings being treated like crap. How smart is it to lower a mp3 before possibly hitting an already lowered digital fader before hitting the low-res DA... I hate this normalization, replay gain(?), K-system, RMS, whatever shit there is. They are all silly attempts to create something impossible - software hearing.  F**k everyone trying to make "technology" like this happen. Those people are stepping miserably bad into any musicians craft.
Thank you for much for quoting this Axon - it's really brought a smile to my face!

Who knew I had such power to annoy people? And the very people who have been destroying the sound quality on so many CDs for the past decade or more. Delicious.

What irony though - someone who wants to clipress the top 12dB of all pop music thinks that an automated volume control is "stepping miserably bad into any musicians craft".   

I won't even click on those threads though - it might drive me mad. Plus I'd hardly be seen as the most objective defender of an automatic volume control, would I?


So, without knowing the background, let me comment thus: If you're going to mandate something (and of course Dolby already has this covered!) mandate the format of the metadata, mandate that devices can all respond to the metadata, mandate a way of filling in or calculating the metadata - but also mandate that users can switch the function off if they want, and (at least on media that supports it) users can tweak the value of the metadata if they think it's wrong.

Cheers,
David.