[KeithEmo]

This whole "argument" has deteriorated into an argument about semantics.

Let's just that, for the sake of argument, on average, more people bought white cars this year than fifty years ago.
But let's also agree that a white car made today is not any brighter than a white car made fifty years ago.

It is technically most accurate to say: "More people are buying white cars now than fifty years ago."
It is also correct to say: "The average color of a car bought today is brighter than the average color of a car fifty years ago."
I would consider it less concise and less informative to declare that: "Cars have gotten brighter in the last fifty years... "
Although you might reasonably argue that it isn't technically incorrect either.
(It's less concise because it fails to provide enough details to tell us if more white cars were sold or if the color of white paint has actually gotten brighter.)

I think that, at this point, when it comes to modern music recordings, we all know what we mean.

The article in SOS was making a point of what they see as a sort of misunderstanding, or perhaps misattribution (I agree with their assertion).
Many people today seem to believe that modern music is "poorly recorded" due to "incompetent engineering"....
Many others seem to believe that CDs, or perhaps digital media, or even digital audio editors, somehow "inherently" reduce dynamic range - either deliberately or unintentionally....
The point of the article, backed up by plenty of data, is simply that the technical capabilities of modern recording formats and modern equipment are just the same as they ever were....
(If anything, they're far better. For example, a 24 bit digital audio file can reproduce far wider dynamic range than a vinyl record album.)
And it is merely an artistic choice of modern artists and recording engineers to produce music that on average occupies a smaller part of that range more of the time.
(And this makes "the loudness war" a "cultural shift" or "a matter of taste" or even "artists selling out to tone-deaf producers and listeners" rather than "a technical failing or limitation".)

In this case, since there are words that define the situation accurately and concisely, it seems to make sense to use them....
Rather than to continue arguing whether words that describe the situation in more common terms, but less accurately, "are right or not"...

It's not my taste, or even about my taste. Stop playing word games! Elephunk proved louder than DMC. American Idiot proved louder than Highway To Hell and even my Kiss CDs.

Castle, with me you get the real deal. If we park our cars together and your's is white and mine is Navy, your car is BRIGHTER than mine - period. It's not my "opinion". White is a brighter color than Navy. Facts here, not Fox!

 

[castleofargh]

I know who Castle is. He is one of the knights who say Ni.

for another million I can leak photographic evidence that you're hiding a space alien in your bat cave.

It's not my taste, or even about my taste. Stop playing word games! Elephunk proved louder than DMC. American Idiot proved louder than Highway To Hell and even my Kiss CDs.

Castle, with me you get the real deal. If we park our cars together and your's is white and mine is Navy, your car is BRIGHTER than mine - period. It's not my "opinion". White is a brighter color than Navy. Facts here, not Fox!

my point wasn't that an album couldn't be louder, but that you're complaining about it as if it was a mistake. I just see different albums being different, and a youngster raised on modern stuff might actually complain that DMC is too quiet to fit with the rest of his playlist. old recordings did often have more headroom and many were more dynamic. now it's different. the real options are as such:
1-do nothing and accept that different albums are simply different the way they already are when using different singers, notes, and all that. if it annoys you like it does me, you can use replaygain, some sort of normalizers like Sony and other brands have on some devices, or play music from online services that already reduce the difference in perceived levels to some extent, or whatever option you find best for yourself. I just output my signal at 24bits because replaygain+EQ+often digital volume setting, it can become a lot for a 16bit signal.

2-try to even things out by pushing the quiet stuff up until 0dB. but in many instances(almost all classical music for starters) those quiet stuff still have some content already near 0dB. so now the actual choice is to clip such album to make it sound like the others(nobody is a fan of that idea). or to compress those albums to gain the headroom necessary to align it better with the perceived level of modern stuff. it's done fairly often with new releases and you absolutely hate that approach and made everybody well aware of that, but it's still one solution that that perceived loudness issue.

3-try to even things out by reducing the gain of the louder stuff. I talked about R128 in my previous post, that's one possibility, knowing that in practice without clipping anything and without compressing anything, there will be many track/albums(including again, a lot of classical music) that won't be able to comply and will remain a good deal quieter subjectively than the predefined loudness target. and as I said, people making albums will probably never agree to sacrifice several bits of their CD for that purpose unless they are forced to do it by law.
to solve the matter of the very dynamic albums remaining too low so they aren't clipped nor compressed, we would have to reduce the gain of everything else to match the quietest albums. but now we're talking pretty massive gain reduction for modern albums genres. we're now seriously flirting with quantization noise becoming audible even for the cardboard ears and their blind tests, who gather in this sub section. meaning that in this scenario, we're pretty much forced to use 24bit. it's doable, it's a serious solution to most issues, but even then, many people will complain. starting with all the true believers of 24bit learning about all the bits that might be sacrificed on the album they will purchase. and ending with guys like me who don't mind using 24bit to much, but don't want to have to pay extra for it. even less so when some options from 1- can give me the same result with a cheaper 16bit album.

4- force mastering engineers to do things the way you want like a dictator, even if it means that the album cannot sound the way the artist or sound engineers wanted it to sound. that would probably kill entire musical genres.


am I missing some solutions? if not, then there you have your options. good luck forcing you view of the correct choice onto the rest of the world.

 

[taffy2207]

for another million I can leak photographic evidence that you're hiding a space alien in your bat cave.

To be fair, that's ET fault for claiming he could hear differences between Audio Formats though

 

[gregorio]

if I was creating my album and someone came telling me to follow say EBU R128 recommendations for my music content, I'd tell that person to F off. first because I'm the creator and wish to do things the way I want. but also because I'm not giving up 3 or 4 of my 16bits when that loudness standard could pretty much be applied on playback or the listener could just go and turn the volume knob.

Hang on, careful here. That's not how EBU R128 (or ATSC 85/A or the underlying ITU BS.1770) loudness normalisation works. If you have a completed master, analyse it with say R128 and then raise or lower the level to achieve a target loudness, the end result is very different to if you create a mix/master to comply with that same target loudness!! To over-simplify; loudness normalization works on the principle of applying a filter (what's called a "K-Weighted" filter, that's somewhat related to the inverse of a loudness contour) and then measuring the average (RMS) of that filtered signal over the duration of the song. Because it's an average over time, there are two fundamental ways of achieving that average: A. Simply have a signal that always maintains that target average or B. Have a signal that spends half it's time at the maximum level above the target average and half it's time at a corresponding amount below the target average. For example, let's say our target loudness is -20LUFS and we have two signals, "A" and "B": Signal "A" peaks at 0dBFS, stays there constantly and measures -10LUFS. Signal "B" also peaks at 0dBFS and also measures -10LUFS but only half the time, the other half the time it peaks at say -12dBFS and measures -30LUFS. To achieve our target loudness of -20LUFS: Signal "A" has to be lowered by 10dB, it now peaks at -10dBFS and stays there. Signal "B" doesn't have to be changed at all, it already hits the target (averages -20LUFS) and it still peaks at 0dBFS (half the time). Clearly there is a significant difference between "A" and "B" but the only option available on playback is effectively "A", lower the whole signal/song to match the target loudness, however the creators of the signal/song can do either "A", "B" or anywhere in between. In practice, there are virtually no circumstances where the artists would choose "A", they would virtually always choose some variation of "B". This has two consequences: Firstly, applying the loudness standard on playback would be a significantly different (and undesirable) result to applying it during creation and Secondly, the resultant masters would still peak at or near 0dBFS and so the artists would not be "giving up" any bits (but of course they couldn't stay at 0dBFS constantly).

[1] Just to satisfy all the politicians on here, I metered the two CDs I mentioned, Elephunk and DMC: 80dB sensitivity, A-weighting, Slow response. ...Barely two bars in, the intro to Elephunk PEGGED THAT METER ALL THE WAY TO THE RIGHT! Sure, it dropped for a few seconds at a time, to about +2 SPL, or during a softer intro, like trk. 13 "Where Is the Love"?
So now, are you all satisfied that I am not imagining things? The MATERIAL on my more recent CD IS LOUDER than the material on my older one.
[2] .... This movement to me implies more range in loudness on the older DMC recording, and less on the BEP one. More dynamics? You guys tell me - you're the experts in semantics! I just deal with what IS.

1. You're joking right? OK, let's use an analogy because clearly you can't understand what's going on with audio: Let's say I make the assertion that a 1.0l VW Polo has a more powerful engine than say a 6.0l truck. Everyone who knows anything about vehicle engines would tell me I'm wrong. So I say, "to satisfy all those politicians", that I've measured and proven it, the Polo's engine goes all the way to 5,500rpm on a tacho, while the truck only goes to 4,000rpm. Unfortunately of course, a tacho does not measure engine power, it only measures RPM and RPM is only one factor in the power output of an engine. I've actually measured the wrong thing, I should have measured BHP and/or torque! The only thing I've actually proven (and "satisfied" everyone of) is that I'm ignorant about engines/engine power! However, I know I'm not ignorant, I know a thing or two about engines, so everyone else is just playing semantics/mind-games and being politicians. - What am I? Am I both ignorant and deluded (that I know what I'm talking about) or am I right and everyone else is wrong?
You've measured sound pressure level (SPL), not loudness (LUFS/LKFS), you've measured the wrong thing! You've therefore satisfied us of just one thing, that you don't know what loudness is and are ignorant that SPL is only one factor of it!

2. And the movement of the needle on the tacho implies to me more range in engine power on the Polo, and less on the truck. You guys tell me - you're the experts in semantics! I just deal with what IS.

Keith Emo, a few pages back, suggested that I "don't have to turn down" the volume for the CD that sounds louder to me. So perhaps you should clarify that with him. Of course I have to turn down the louder recording - people are leaving the room!

What people are leaving the room? You + your wife + maybe a mate or, all the people in the world? We've already ascertained that some/many night clubs are too loud for you (and presumably your wife), does everyone always leave these clubs? If these clubs were always empty, they'd rapidly go out of business and there wouldn't be any of these clubs. So how do you explain their continued existence, if it's not just your (and your wife's) personal taste about loudness?

G

 

[castleofargh]

Hang on, careful here. That's not how EBU R128 (or ATSC 85/A or the underlying ITU BS.1770) loudness normalisation works. If you have a completed master, analyse it with say R128 and then raise or lower the level to achieve a target loudness, the end result is very different to if you create a mix/master to comply with that same target loudness!! To over-simplify; loudness normalization works on the principle of applying a filter (what's called a "K-Weighted" filter, that's somewhat related to the inverse of a loudness contour) and then measuring the average (RMS) of that filtered signal over the duration of the song. Because it's an average over time, there are two fundamental ways of achieving that average: A. Simply have a signal that always maintains that target average or B. Have a signal that spends half it's time at the maximum level above the target average and half it's time at a corresponding amount below the target average. For example, let's say our target loudness is -20LUFS and we have two signals, "A" and "B": Signal "A" peaks at 0dBFS, stays there constantly and measures -10LUFS. Signal "B" also peaks at 0dBFS and also measures -10LUFS but only half the time, the other half the time it peaks at say -12dBFS and measures -30LUFS. To achieve our target loudness of -20LUFS: Signal "A" has to be lowered by 10dB, it now peaks at -10dBFS and stays there. Signal "B" doesn't have to be changed at all, it already hits the target (averages -20LUFS) and it still peaks at 0dBFS (half the time). Clearly there is a significant difference between "A" and "B" but the only option available on playback is effectively "A", lower the whole signal/song to match the target loudness, however the creators of the signal/song can do either "A", "B" or anywhere in between. In practice, there are virtually no circumstances where the artists would choose "A", they would virtually always choose some variation of "B". This has two consequences: Firstly, applying the loudness standard on playback would be a significantly different (and undesirable) result to applying it during creation and Secondly, the resultant masters would still peak at or near 0dBFS and so the artists would not be "giving up" any bits (but of course they couldn't stay at 0dBFS constantly).

not sure I get your point. if I wanted to make a modern album with low dynamic like many in existence, the kind that troubles @TheSonicTruth for being too loud, wouldn't I have to put the highest peak several dB down so the average can reach the target?

 

[TheSonicTruth]

not sure I get your point. if I wanted to make a modern album with low dynamic like many in existence, the kind that troubles @TheSonicTruth for being too loud, wouldn't I have to put the highest peak several dB down so the average can reach the target?

You, as the material provider, don't have to adhere to the loudness standard. The deliverer(broadcaster, streamer, etc) will apply appropriate gain to bring it into line.

As far as "troubling me"? I am hardly the only person on Earth bothered by excessively loud music or other content. A Google of loudness will surely not bring up "the sonic Truth". I'm just tired of the politicians and connivers on here trying to tell me what I hear, see, or feel isn't real.

 

[gregorio]

not sure I get your point. if I wanted to make a modern album with low dynamic like many in existence, the kind that troubles @TheSonicTruth for being too loud, wouldn't I have to put the highest peak several dB down so the average can reach the target?

Not necessarily, depending on the construction of the song and what the average (LUFS/LKFS) target actually is. As the SOS article demonstrates/proves, the dynamic range/variability hasn't really changed much over the decades, even though compression and limiting have significantly increased. With a typical "intro/verse/chorus/verse/chorus/outro" type structure, you could achieve a lower LUFS by making the mix of the intro, outro and verses more sparse (though not necessarily lower peak level) while the choruses are still the same level (peaking at 0dBFS, with the same amount of compression/limiting). Other genres, such as most EDM, has a somewhat less strict song structure but still has sparser (or quieter sounding) sections building up to louder/big climax type sections, even though it's all compressed to death. The only pieces where you'd struggle and probably have to reduce peak level is some thrash/death metal pieces, which sometimes have little/no structure, it's just dense thrash from beginning to end.

@TheSonicTruth - Oh, you must be right - because using big, bold fonts for your insults definitely qualifies as reliable evidence/proof.
And BTW, you did NOT measure loudness, by your own admission you measured SPL and, for the umpteenth time, neither me, nor the article, nor anyone else here is stating that there isn't a loudness war! You can't understand analogies deliberately simplified for your benefit and you can't even understand simple sentences! Honestly, what is wrong with you?

G

 

[TheSonicTruth]

Honestly, what is wrong with you?

Nothing at all. Because if there was, there would also be something wrong with my wife, and any guests we have had over, listening with us.

I just listened to CDs from different eras, and for the most part the recent ones ARE LOUDER, to anyone's ears, not just mine, than those of 30 years ago.

And all the deflections in the world, telling me I'm "measuring the wrong thing", etc. won't change that indisputable conclusion.

Like I said gregorio: Run for office! You'd be a great politician - tell people what they want to hear, tell them what's real, and what isn't. Wait: don't we have someone in DC doing all that now? I must have you and him mixed up in my mind.

 

[castleofargh]

You, as the material provider, don't have to adhere to the loudness standard. The deliverer(broadcaster, streamer, etc) will apply appropriate gain to bring it into line.

As far as "troubling me"? I am hardly the only person on Earth bothered by excessively loud music or other content. A Google of loudness will surely not bring up "the sonic Truth". I'm just tired of the politicians and connivers on here trying to tell me what I hear, see, or feel isn't real.

I'd be very happy if all the audio sources had a sort of replaygain integrated. as to politician, I write "perceived louder" instead of "being louder", you interpret that as me doubting your ability to tell if something is really louder. just now I mention your name as a very obvious example of someone unhappy with music stuck near 0dB all the time, and you're mad as if you weren't by far the most obvious reference of someone obsessed and very vocal about those stuff on the forum. all right, I'm a politician if you want.


I think I'm done trying to clarify your mess.

 

[castleofargh]

Not necessarily, depending on the construction of the song and what the average (LUFS/LKFS) target actually is. As the SOS article demonstrates/proves, the dynamic range/variability hasn't really changed much over the decades, even though compression and limiting have significantly increased. With a typical "intro/verse/chorus/verse/chorus/outro" type structure, you could achieve a lower LUFS by making the mix of the intro, outro and verses more sparse (though not necessarily lower peak level) while the choruses are still the same level (peaking at 0dBFS, with the same amount of compression/limiting). Other genres, such as most EDM, has a somewhat less strict song structure but still has sparser (or quieter sounding) sections building up to louder/big climax type sections, even though it's all compressed to death. The only pieces where you'd struggle and probably have to reduce peak level is some thrash/death metal pieces, which sometimes have little/no structure, it's just dense thrash from beginning to end.

all right so I didn't get the wrong idea. but maybe I'm expecting more albums to have to go seriously down in gain than is actually the case based on what replaygain(that now relies on the R128 thing AFAIK) does to my library.

 

[TheSonicTruth]

perceived louder"

If 100 people think something is either louder than something else, or is too loud, period, they're not "perceiving" it - it's JUST TOO FRICKN LOUD!

Stop with the play on words already - all of you!

 

[Light - Man]

 

[Yuurei]

If 100 people think something is either louder than something else, or is too loud, period, they're not "perceiving" it - it's JUST TOO FRICKN LOUD!

And what if 1000 people think it's not louder or too loud?

 

[gregorio]

Nothing at all. Because if there was, there would also be something wrong with my wife ...

Thanks for proving my point! And thanks also for the again entertaining: It must be scientific fact because my wife is the same!

all right so I didn't get the wrong idea. but maybe I'm expecting more albums to have to go seriously down in gain than is actually the case based on what replaygain(that now relies on the R128 thing AFAIK) does to my library.

Firstly, to be honest, I'm not sure what replaygain is exactly based on, it might be the same basic principle as R128 (IE. The underlying ITU k-weighted RMS) but I'd be surprised if it is actually R128. There seems to be various different flavours of replaygain. R128 is designed for TV broadcast and loudness is normalised to -23LUFS. However, that is too low for many playback devices, the target for Apple Music is (according to Bob Katz) equivalent to -16.5LUFS and Youtube (when they apply it) is reportedly equivalent to -13LUFS. Again, we come back to the problem of music having no agreed/mandated standards, this is a pretty big difference between target levels and it also goes back to where we first started, -23LUFS potentially being a bit low even for some TVs/systems and is sometimes one of the reasons why the dialogue might seem occasionally too quiet.

Secondly, even with a fairly low target LUFS, it is of course still possible to have the loudest sections at 0dBFS and heavily compressed but the duration of those sections may have to be shorter (relative to the duration of the song) and the less loud sections even quieter. Generally that would still be preferable to most artists than just lo lower the level of the whole song. What I would expect to see eventually though is still the loudest sections peaking at 0dBFS quite consistently but probably three things: Not quite so much limiting, not so much EQ (or multi-band compression + make-up gain) emphasising the most sensitive region of hearing and a gradual evolution of song structure/construction.

[1] If 100 people think something is either louder than something else, or is too loud, period, they're not "perceiving" it - it's JUST TOO FRICKN LOUD!
[2] Stop with the play on words already - all of you!

1. True, providing of course the population of the planet is 100 people, is that what you're suggesting? You haven't answered the question of how come all the night clubs still exist, where the music is too frickn loud for you but not for the countless thousands who pay good money for it to be that loud! This is the sound science forum, not the "it's too loud for thesonictruth and his wife" forum. How many times (are you going to make a fool of yourself)?

2. Yes, a simple sentence is a "play on words" and you've repeatedly demonstrated that you can't understand a simple sentence but that's your problem, not ours! "All of us" can't stop with the sentences because it's how the rest of our species communicates!

G

 

[castleofargh]

Firstly, to be honest, I'm not sure what replaygain is exactly based on, it might be the same basic principle as R128 (IE. The underlying ITU k-weighted RMS) but I'd be surprised if it is actually R128. There seems to be various different flavours of replaygain. R128 is designed for TV broadcast and loudness is normalised to -23LUFS. However, that is too low for many playback devices, the target for Apple Music is (according to Bob Katz) equivalent to -16.5LUFS and Youtube (when they apply it) is reportedly equivalent to -13LUFS. Again, we come back to the problem of music having no agreed/mandated standards, this is a pretty big difference between target levels and it also goes back to where we first started, -23LUFS potentially being a bit low even for some TVs/systems and is sometimes one of the reasons why the dialogue might seem occasionally too quiet.

Secondly, even with a fairly low target LUFS, it is of course still possible to have the loudest sections at 0dBFS and heavily compressed but the duration of those sections may have to be shorter (relative to the duration of the song) and the less loud sections even quieter. Generally that would still be preferable to most artists than just lo lower the level of the whole song. What I would expect to see eventually though is still the loudest sections peaking at 0dBFS quite consistently but probably three things: Not quite so much limiting, not so much EQ (or multi-band compression + make-up gain) emphasising the most sensitive region of hearing and a gradual evolution of song structure/construction.

things have changed a good deal over the years in foobar and probably will again. at some point we could set our own target, but I can't seem to find that option anymore. tried a few tracks and I'm getting a stable -18LUFS (using dpmeter3 VST inside foobar) for each entire individual track while track gain is applied. so hey indeed seem to have given up on the -23LUFS and picked a gentler value, at the cost of the target not coming close to most of my classical albums that I can't boost without clipping or compressing them.

about your secondly, I agree that if such standard was forced onto the music industry, they would most likely adapt to that and most would change the way they would make their album. part of my previous argument was about how things changed with new tools so I'm not going to contest that likely possibility^_^. but I was considering the case of someone who wished to make an album with the dynamic of the worst Metallica album, because that's how he imagined his tracks should sound. or even more annoying, already released albums that had really low dynamic and would be released again with a mandatory loudness normalization standard. but I guess that would lead to remastering when possible instead of giving up some dBs. so you do have a point.