Loudness compression, loudness wars.. What exactly it is and why is it happening?

[TBone]

in today's world of mass compression ... DR13 is more than reasonable ... can you please list the sku# for your CD version of Fully Completely.

 

[Guermantes]

Bach Goldberg Variations, Variation 28 - Trevor Pinnock, harpsichord - Archiv Produktion

Waveform:

DR analysis:

WLM LKFS analysis:

 

[Soniclife]

The long term (or integrated) loudness of this track is -20 LUFS with a true peak of -1 dB, so to be streamed on Spotify or TIDAL it may be need to be peak compressed before loudness normalisation to their target of -14 LUFS.

Is this what the streaming apps do for things lower than their target? I assumed they just applied simple gain, with no consideration of peak clipping, the inverse of what they would do for too loud tracks, because if you want something better you should just turn the feature off.

 

[Soniclife]

Has there ever been testing done on what people prefer for dynamic compression?

I long assumed that one of the secrets of vinyl is it's dynamic range is quite domesticated, at a comfortable listening level the peaks don't get too loud, and the quiet bits too quiet. CD being less constrained allowed people to start considering what they think other people would want.

 

[bennetng]

Is this what the streaming apps do for things lower than their target? I assumed they just applied simple gain, with no consideration of peak clipping, the inverse of what they would do for too loud tracks, because if you want something better you should just turn the feature off.

According to this article, only Spotify compresses (and definitely not clip) quiet songs to meet the target when loudness normalization is enabled.
http://productionadvice.co.uk/online-loudness/

Notice the date of posting, things may change but it's easy to verify. At least Youtube doesn't compress or normalize songs quieter than the maximum allowed level. Youtube won't recompress or clip excessive loud songs as well, they just do simple downward normalization.
https://www.audiosciencereview.com/...it-is-and-why-is-it-happening.3392/post-83062

Also this:

I don't use Tidal but according to this article the settings are highly customizable.
https://smartphones.gadgethacks.com/how-to/tidal-101-enable-loudness-normalization-0180380/

 

[TBone]

I long assumed that one of the secrets of vinyl is it's dynamic range is quite domesticated, at a comfortable listening level the peaks don't get too loud, and the quiet bits too quiet. CD being less constrained allowed people to start considering what they think other people would want.

In practice, and in general, CD's have become are much more constrained based on the tilted mastering.

Again, in general, because examples of constraint (dynamic compression) can be shown favorably - in either format.

 

[Soniclife]

According to this article, only Spotify compresses (and definitely not clip) quiet songs to meet the target when loudness normalization is enabled.
http://productionadvice.co.uk/online-loudness/

Thanks, I hadn't realised the services were so different. It's possibly a good thing they are all slightly different, will prevent or slow down someone producing a plug in that subjectively sounds louder but fools their algorithms.

I don't use Tidal but according to this article the settings are highly customizable.
https://smartphones.gadgethacks.com/how-to/tidal-101-enable-loudness-normalization-0180380/

They seem to have removed anything more than off or in that in the current android app.

 

[bennetng]

someone producing a plug in that subjectively sounds louder but fools their algorithms.

I am worrying about this as well.

 

[stalepie]

Battery life lasts longer if the user doesn't have to keep the volume as high (it requires more energy to output smaller waveforms at enjoyable listening levels). So popular music was increasingly made loud to maximize the battery life of people's iPods and phones.

 

[Blumlein 88]

Battery life lasts longer if the user doesn't have to keep the volume as high (it requires more energy to output smaller waveforms at enjoyable listening levels). So popular music was increasingly made loud to maximize the battery life of people's iPods and phones.

I believe this member is now banned. Rightly so with terrible statements like the above.

 

[TBone]

just read Best Buy has abandoned the CD. They still sell LPs, as vinyl sales gained 9%. As how this impacts you Audiophiles; well, from my perspective, I've resisted purchasing new CD's for well over a decade. Still purchase used originals versions or perhaps some early well recorded pre-war re-masters, a decision solely based on dynamic content.

CD started life just fine, majority of originals contained fine dynamic content, the same (measured) or better compared to the LP version. Never-the-less, it was panned, especially by Vinylphiles (I.inc.), many who incorrectly and ironically claimed it lacked dynamic range. Yet, within a few years, many early DDD based digital recordings were being mastered to include superbly wide dynamic range values, beyond anything prior recorded to tape.

Many examples exist, but one digital recording, 1988 Tracy Chapman, in particular, toppled the cart ...
----------------------------------------------------------------------------------------------
Analyzed folder: C:\tb1_music\tracychapman\
----------------------------------------------------------------------------------------------
DR Peak RMS Filename
----------------------------------------------------------------------------------------------
DR14 -0.80 dB -17.06 dB 01 Track01.wav
DR15 -0.39 dB -18.78 dB 02 Track02.wav
DR15 -0.88 dB -18.27 dB 03 Track03.wav
DR12 -9.06 dB -23.83 dB 04 Track04.wav
DR15 -0.20 dB -17.43 dB 05 Track05.wav
DR14 -0.65 dB -17.22 dB 06 Track06.wav
DR15 -0.21 dB -17.77 dB 07 Track07.wav
DR13 -1.05 dB -16.85 dB 08 Track08.wav
DR15 -0.20 dB -17.70 dB 09 Track09.wav
DR16 -0.76 dB -21.00 dB 10 Track10.wav
DR14 -5.83 dB -24.15 dB 11 Track11.wav
----------------------------------------------------------------------------------------------
Number of files: 11
Official DR value: DR14
==============================================================================================

Both CD & LP versions were released about the same time. I can remember it like yesterday, back in 88, less than 5 years after CD came to market. and the Audiophile circus was coming to town. Nearly every hotel room had Tracy demo/singing ... and circa 88, every room spun a turntable. Few catered to CD based audiophiles. So, what I ended up experiencing, unknowingly, was digital content, demonstrated like a hot tomato from room to room, while audiophiles clapped like hungry seals at a sardine party. A 16 bit digital recording, w/exceptional dynamic content ... played beautifully on vinyl. Analog never sounded so good.

Advance digital technology +4 years: and examples of fine dynamic content are less common, but still exist ...

Cafe Blue
--------------------------------------------------------------------------------
DR Peak RMS Duration Track
--------------------------------------------------------------------------------
DR14 -0.77 dB -19.34 dB 5:25 ?-01_What_A_Shame
DR14 -0.03 dB -18.46 dB 6:59 ?-02_Mourning_Grace
DR13 -3.92 dB -21.66 dB 4:40 ?-03_The_Thrill_Is_Gone
DR13 -3.09 dB -23.05 dB 4:28 ?-04_Romanesque
DR17 -0.09 dB -20.09 dB 5:08 ?-05_Yellow_Car_III
DR15 -4.13 dB -22.45 dB 0:58 ?-06_Wood_Is_A_Pleasant_Thing_To_Think_About
DR13 -6.99 dB -23.90 dB 5:03 ?-07_Inch_Worm
DR18 -1.36 dB -24.19 dB 5:21 ?-08_Ode_To_Billy_Joe
DR12 -0.06 dB -16.43 dB 7:59 ?-09_Too_Rich_For_My_Blood
DR18 -0.10 dB -22.92 dB 4:29 ?-10_A_Taste_Of_Honey
DR15 -0.02 dB -19.87 dB 9:02 ?-11_Nardis
DR20 -0.27 dB -23.83 dB 3:25 ?-12_Manha_de_Carnaval
--------------------------------------------------------------------------------
Number of tracks: 12
Official DR value: DR15

... but about this time ... a disturbing trend started to appear.

--------------------------------------------------------------------------------
Statistics for: 01-War Pigs
--------------------------------------------------------------------------------
Left Right
1987 Germany 832 701-2 Vertigo
Peak Value: -0.20 dB --- -0.20 dB
Avg RMS: -15.66 dB --- -15.59 dB
DR channel: 13.74 dB --- 13.57 dB
1987 USA 3104-2 Warner
Peak Value: 0.00 dB --- -1.07 dB
Avg RMS: -18.17 dB --- -17.46 dB
DR channel: 14.82 dB --- 13.04 dB
1989 Japan 23PD-134 Vertigo
Peak Value: 0.00 dB --- -1.07 dB
Avg RMS: -18.16 dB --- -17.46 dB
DR channel: 14.89 dB --- 13.06 dB
1991 Japan TECP-23893 Teichiku
Peak Value: -1.40 dB --- -2.31 dB
Avg RMS: -16.72 dB --- -18.14 dB
DR channel: 12.83 dB --- 12.80 dB
1996 UK ESM CD 302 Castle
Peak Value: 0.00 dB --- 0.00 dB
Avg RMS: -11.50 dB --- -10.48 dB
DR channel: 9.57 dB --- 8.41 dB

... consistently! Values would only get worse with time.

CD died because as a physical medium, it became redundant. Sonically, it never really mattered, if it wasn't vinyl that put it down, it was higher rez.. But in its youth, it proved capable of fooling any discerning audiophile, and that still holds true today. As it aged, it became neglected, maybe after decades worth of being blamed for bad sound quality and a perceived lack of dynamic content; it started to act the part.

The example above was noted because I've recently purchased a few new LPs and hopefully they contain some dynamic content. While the packaging is exceptional ... vinyl quality can range from really good to depressing, dynamic content varies much the same way.

Recently purchased some new Hendrix on vinyl. AAAAAArghhhh, it's mastered by George Marino. Despite George's reputation, he's butchered more dynamic content on CD than I care to remember.

this is the original 1987 CD (Hey Joe) not mastered by George ...

and this is George's handy work (~1997) ...

My recent rip of George's re-mastered LP ...

Left/Right tonal comparison; LP vs orig CD ...

If you had suggested in 1988/98/08 ... that by 2018; CD would be dropped entirely, as retailers & consumers still preferred LP ... well ...

All along, vinylphiles have suggested that vinyl offers greater "dynamics". It's easily proven wrong, but lets pretend for now ... thankfully; George Marino didn't dynamically slam this particular LP based version, like he had with at least two prior versions on CD. Why he felt the need to scrub Hendrix with such a heavy compressed brush, in the past, is beyond reason. While I can't see a return to +13DR values anytime soon ... let's hope this recent trend (with vinyl DR content) continues ...

 

[danadam]

That 1997 Hey Joe, you can reduce volume (e.g. -3dB), apply steep high pass filter at 20 Hz and you'll get increase of DR value by 3 or 4. Do you think it will make it sound better? ;-p

 

[TBone]

No, because the canopy has been clipped well beyond repair, and no amount of fiddling with filters will replace lost data.

Just one example of hundreds of clips - within the 1997 remaster ...

 

[danadam]

My point is that your remastered LP, which you seemed to be fond of, could have been done from the same master as that CD. And that the higher DR value of the LP could be because of a similar process that I described but triggered by the transfer to vinyl. I'm not an expert, so I'm not saying that this is exactly what happens, but it is commonly known, I think, that cd and vinyl from the same master have different DR values, e.g.,

Also I'm not saying that in your case the master was the same. I'm just saying that you can't say that they were not just because DR values are different.

 

[TBone]

My point is that your remastered LP, which you seemed to be fond of, could have been done from the same master as that CD.

He may have used the same "master", but certainly not the same re-mastering.

Again, I repeat; the canopy has been clipped.

One (+hundreds) of examples ...

Same relative music timeframe; notice the severe limiting/clipping, absent on the LP.
right channel, hey joe, re-mastered CD.

right channel, hey joe, re-mastered LP rip.

As you suggested, take ~3dB off the re-mastered CD, and re-calc the DR values:

version: re-master / re-master(-3dB)
right channel: 0.00 / -3.04 dB
Avg RMS: -12.40 / -9.35 dB
DR channel: 7.39 / -7.39 dB

Everything changes by about 3dB, except the DR value, which remains identical!

And that the higher DR value of the LP could be because of a similar process that I described but triggered by the transfer to vinyl. I'm not an expert, so I'm not saying that this is exactly what happens, but it is commonly known, I think, that cd and vinyl from the same master have different DR values,

Nope, been ripping for decade; it's not commonly known, and if it is, it's mistaken!

With a fine turntable, DR values will remain relatively the same ... as I've witnessed countless times over, with numerous rips.

Also I'm not saying that in your case the master was the same. I'm just saying that you can't say that they were not just because DR values are different.

Yes I can if the evidence points in that direction. In this case; I can safely state the "re-mastering" is NOT the same (even if the same master was used). The so-called vinyl process, in this case, is not in question, its Marino's choice of slamming the CD that's questionable.

 

[krabapple]

Re: history of Loudness Wars... it's one of the tragedies of audio (another being the 'circle of confusion' Toole has written often about) that loudness mastering on CD kicked in in the early 90s, soon after *another* trend got underway in the late 80s -- the 'remastered from original master tapes' trend. First-generation CDs were reportedly (no one to my knowledge has ever quantified this) often mastered from secondary or higher-generation copies, which might have been specifically EQd and processed for vinyl (i.e., 'production' master tapes). Once that was recognized as a problem -- and a source of new revenue -- companies began to cough of up their archived original masters for CD re-release. So, for a brief window we were getting remasters that promised to deliver what CD has originally promised....but soon it dovetailed with the digital compression trend...and the rest is history. Frustratingly, we get the best tape sourcing which is then compromised by faddish, least-common-denominator mastering.

 

[krabapple]

If you had suggested in 1988/98/08 ... that by 2018; CD would be dropped entirely, as retailers & consumers still preferred LP ... well ...

All along, vinylphiles have suggested that vinyl offers greater "dynamics". It's easily proven wrong, but lets pretend for now ... thankfully; George Marino didn't dynamically slam this particular LP based version, like he had with at least two prior versions on CD. Why he felt the need to scrub Hendrix with such a heavy compressed brush, in the past, is beyond reason. While I can't see a return to +13DR values anytime soon ... let's hope this recent trend (with vinyl DR content) continues ...

Those 1997 Experience Hendrix remasters were credited to Eddie Kramer and George Marino.

I would add too that the audibility of digital clipping depends on how long it lasts, and how often it occurs.

 

[TBone]

Re: history of Loudness Wars... it's one of the tragedies of audio (another being the 'circle of confusion' Toole has written often about) that loudness mastering on CD kicked in in the early 90s, ...

In my books, the greatest sin to all things audio is the introduction of mass dynamic compression, esp. to CD post 90's. CD deserved much better. As audiophiles, we are still paying the sonic price ... contrary to the "war is over" bullshit put forth above, dynamic range values have never truly recovered. But most audiophiles don't give a rats-ass anyway; today, most would rather discuss DACs and truly meaningless distortions @-90dB while they blissfully listen to dynamically stunted re-masters & streaming services.

I much prefer feeding my system higher dynamic content, in any form ...

(last night's vinyl rips)
DR13 -1.30 dB -15.44 dB 01_FamilyMan.wav
DR12 -0.61 dB -15.79 dB 02_LittleLies.wav
DR14 -0.01 dB -17.10 dB 03_Everywhere.wav
DR14 -1.18 dB -16.99 dB 04_SevenWonders(att=7.5).wav
DR14 -2.74 dB -18.21 dB 05_Mystified.wav
DR15 over -16.66 dB 06_HigherLove(att=8).wav
DR13 -3.99 dB -19.37 dB 07_NovaHeart.wav

 

[krabapple]

As I believe at least one poster here has noted (as anecdotal evidence), the amount of added compression also matters; it's not always readily audible. I would also note that the 'dynamic range' (typically the crest factor) is also affected by EQ profile, particularly the relative bass energy of two different mastering.

 

[maty]

Loudness War, an article in NYT.

They Really Don’t Make Music Like They Used To

Opinion | They Really Don’t Make Music Like They Used To - The New York Times by Greg Milner