Let Freedom Ring

In the 2014 November issue, my good friend Steve Guttenberg ("Communication Breakdown") got his facts mostly right: It's true that most listeners (including myself) accept far more distortion today than we did years ago. Many people have never heard a great stereo system—all they've heard are overdriven boom boxes, cheap stereos, and portable systems, and that's what they expect systems and music to sound like. And distortion is part of the sonic language of such musical genres as hip-hop, rap, and alternative rock.

Technically speaking, distortion is compression is distortion is compression. You can't have one without the other: compression reduces peaks, and a distorted waveform has fewer peaks than a clean one.

However, Steve missed the important reason why many music professionals prefer distorted, closed-in, compressed sound over the clean, open version: Distorted music sounds louder! A squarewave is at least 3dB louder than a sinewave of the same peak value—and these days, musicians and producers are making a lot of squarewaves.

Louder is seductive but deceptive. Producers use compression to improve body and punch, to "glue" their mixes, but they deceive themselves if they don't match before–and–after levels for an objective assessment of what they've done. Likewise, audio reviewers should match levels. I thought my new DAC sounded deeper, wider, and clearer—but the only difference was that it was (only) 0.2dB louder than my previous DAC! While real differences do exist between components, when I matched the levels, the sonic differences vanished. This is not always the case, but a difference in loudness produces differences in the sounds perceived.

Popular-music producers take advantage of loudness every day to gain (pun intended) your attention—the loudness war has been going on for decades. The median loudness of LPs went up about 4dB between 1950 and 1980. The median loudness of the pop CD rose 9dB between 1979 and 2011 (footnote 1).

The impact, punch, clarity, soundstage, transient response, and microdynamics of recordings, all of which we audiophiles consider desirable qualities, are affected by a recording's peak-to-loudness ratio (PLR). Between 1980 and 2010, the median PLR of charting pop recordings decreased from 16.6 to only 8.9dB.

To reproduce good-sounding transients, a medium needs a minimum headroom to accommodate peaks of 16dB above the loudness. But here's the rub: a recording cannot have a high peak-to-loudness ratio and simultaneously be loud (at the same position of your volume control). This is why CD sound quality has deteriorated so much in 30 years. We can't stop the loudness war by asking producers to pull back levels, because our nervous systems are wired for louder. However, an overcompressed, squashed, distorted recording sounds fatiguing and wimpy compared to a more open version—once loudness is matched. Notably, Michael Jackson's "Beat It" rose over 10dB—that is, it became twice as loud and more compressed—over four different editions, from its 1980 CD release to a 2009 compilation. Another example, Paul Simon's beautiful Graceland, disappointed when I discovered that the high-resolution file from HDtracks is not a restored original, but a distorted, overcompressed remastering.

Did Michael Jackson and Paul Simon request louder, more distorted versions of their hits? Not. These classic recordings were remastered to be louder than the originals, to keep them from sounding quieter next to more current victims of the loudness war. To my ears, the remastering producers made the wrong decision, because the original versions of these legacy recordings can't be beat. Fortunately, HDtracks also offers the original master of Thriller: the hi-rez version sounds like listening to Michael for the very first time. Play it. Compare it to the Black Eyed Peas' Let's Get It Started to hear how far we've veered toward hypercompression and distortion. I love the Black Eyed Peas' music, but I wish it were more dynamic.

No matter what you think of the sonic quality of the Black Eyed Peas, the loudness war has to stop. It has jeopardized our recorded legacy and affected the sonic practices of popular music producers. Most of my popular-music clients request a louder master if mine is 1 or 2dB lower than its competition. Only a few daring, knowledgeable producers make lower-level records—the ones you have to turn up with your own volume control. I love to produce gorgeous-sounding recordings, but today, most artists and A&R departments choose the overcompressed version for fear that a lower-level recording will not be appreciated by the public (footnote 2). Most of those fears are misguided, because radio processing always makes every recording equally loud.

Today, most music services, including iTunes Radio and Spotify, are loudness-normalized. But iTunes' file playback is not normalized (you can turn on normalization by choosing "Sound Check" in Preferences). Many producers audition music files through iTunes, but since Sound Check is not a default setting, the loudness war wages on (footnote 3).

Steve misses the point: Loudness normalization is liberating, not limiting! While the current situation tends to box producers into a corner, loudness normalization expands our creative choices. It frees us to create sound any way we wish—compressed, distorted, dynamic, open and clear, or anywhere in between. I urge Stereophile readers with friends at Apple to ask them to turn on Sound Check by default. The same goes for YouTube and SoundCloud. This change will revolutionize the sound world overnight, lead to the end of the loudness war, and give rise to the Loudness Revolution. It will be a new day that gives artists the freedom to produce the sound they desire, without competitive loudness pressuring their decisions. Let Freedom Ring!—Bob Katz



Footnote 1: The LP figure is my estimate. The CD figure is based on actual measurements of 10,000 charting recordings by Austrian sound engineer Rudolph Ortner in his masters thesis, The Evolution of Loud.—Bob Katz

Footnote 2: John Marks wrote about this distressing trend in his April 2011 column. Footnote 2 on the first page of JM's article lists prior Stereophile coverage of the Loudness War.—John Atkinson

Footnote 3: In mastering engineer Bob Katz's new book, Mastering Audio: The Art and the Science, Third Edition (Focal Press), he explains how music is recorded, mixed, and mastered today. Chapter 17 tells a vivid tale using Rudi Ortner's statistics. Loudness Normalization requires a proper loudness target.—John Atkinson

COMMENTS
georgehifi's picture

No natural sounds in life are compressed, why do it to the music we listen to.
Uncompressed music gives you space between the notes, these days everything is the same level (no space!!).
I believe it's done so Ipod etc users don't blow their eardrums or earphones up when dynamic transients are produce, because they would turn up the volume during uncompressed quite passages.

Cheers George (I hate compression of any sort, it should be banned)

corrective_unconscious's picture

Is not described here or in situ clearly and simply enough to have the average end user make it make any long term difference in music production. It's not even in some iTunes products/versions.

Also, I think another driver of these _compression wars_ is to cope with background noise in the listener's environment. This is a post purchase consideration, and it has become more of a necessity for listeners as they multi task with other things that emit sound.

Part's "Miserere" is a good example of a recording which I must manually gain ride, even in a typical home environment with just background noise. That simply won't be acceptable to someone playing a computer game or driving or using earbuds in public.

drblank's picture

First off, let's discuss the two different approaches to making a recording.
1. Performance based - These are generally acoustic instruments (sometimes combined with electric), but the goal is to capture a performance and have a recording that's true to the original sound of the musicians in the room they were performing in. They typically don't alter the recordings all that much other than the RIAA curve for vinyl, and they typically have little or no detectable audio compression for the average listener.
2. Production based recordings. This is your typical pop, rock, metal, grunge, rap/hip hop, dance, techno, etc. This is where it's heavily PRODUCED and they want a fatter, bigger, beefier sound and when played over the radio, the track jumps out. People that tend to buy/listen to these types of recordings are generally not audiophiles, they are typically anywhere from teenagers, to young adults and they are listening to these recordings either with cheap stereos, cheap earbuds in public places, cars and nightclubs where there is a lot of ambient noise, they want to just feel the bass and they typically just want to hear a loud version of the song and dynamics is not an issue. This is where audio compression is used, along side with gating, limiting, eq, special effects like Pitch Correction, chorus, delay, electronic reverb, flanger, phasers, Harmonizers, etc. etc. This is where you hear the guitar player playing, but you can't tell why type of guitar is being played because there is so much signal processing. Drums that sound fatter and bigger than normal because they don't want a natural sound, they want an artificial, "produced" sound that hopefully grabs the listener because they kind of don't like the natural sound of instruments, because in their minds that equate to "old people' music. Heaven forbid these kids to actually know what an instrument sounds like or even know where music came from. :-)

Compression does make sense if you are in an environment where there is a lot of room ambiance because you simply can't hear subtle passages unless they are loud enough. But in those environments, people aren't typically doing what's referred to serious listening, they are going to be doing more casual listening where sound quality isn't as important.

Ladyfingers's picture

The problem with Loudness War recording isn't actually compression so much as peak limiting/clipping and pumping/ducking - both of which actually detract from perceived loudness at matched volumes.

One staggeringly loud and compressed album from the early '90s was Smashing Pumpkins' Siamese Dream, but the original issue had an RMS output level of about 10dB below full scale. My Bloody Valentine's Loveless may well be the most processed album ever made and also hovers around -10dBFS.

No-one in their right mind would describe either of these particular albums as anything but cranked, but neither of them sound harsh either.

drblank's picture

Not compressed? Yeah right, both those examples are heavily compressed with God knows what else has been done to the tracks from mix down to mastering. They certainly are not examples of a good sound quality. There aren't any dynamics. I found them to be unlistenable and unpleasant.

What happens during audio compression is that it makes it where the entire track is loud, vs having dynamics. The problem is with those albums you mentioned is there is so much processing going on, it's hard to say where the problem really lies, but audio compression is VERY apparent. I found them to be horrible sounding albums that I simply can't listen to for longer than a couple of minutes before it starts to hurt my ears at even a low volume level. Too much distortion.

Ladyfingers's picture

Both albums are heavily compressed - compression is not really the problem with the loudness wars - but there is no loss of detail due to heavy-handed peak limiting. The drums and bass still sound intelligible. You can listen into their respective walls of distortion endlessly (if they're your taste).

drblank's picture

Here's an interview with Bob Ludwig about the loudness wars. He's considered one of the top mastering engineers of all time.

https://www.youtube.com/watch?v=nMWUOCbWp1M&spfreload=10

There are countless videos on the "loudness wars" by Bob Ludwig and other mastering engineers that discuss this.

Ladyfingers's picture

Compression and limiting are two sides of the same coin. Compression raises the volume of lower passages, while limiting lowers the volume of louder ones.

If you look at pre-Loudness War waveforms, you can seen a lot of whiskery transient peaks where drums jump out of the mix. In Loudness War-era recordings these are conspicuously absent.

Compression alone would not be able to raise the overall levels to where they are now, because the maximum output would be determined by the highest transient peak. You actually have to lower the peaks relative to the average output, and you can then normalise the levels to "fill" or maximise the output up against the 0dBFS brick wall.

The overall effect can be summed up as "dynamic range compression", yes, but the primary culprit of the specific wimpy, thin, fragile, congested sound of Loudness War-ed audio is the removal of percussion and other transients through peak limiting. If you tried to achieve the same effect with only compression, you would end up with the rest of the music audibly ducking under every drum hit.

corrective_unconscious's picture

Because with the peaks limited you have still lost dynamic range.

It doesn't really matter in a qualitative sense if dynamic range is being lost only by limiting the peaks, or by both limiting the peaks and boosting the low levels. The ramifications will wind up being the same in most listening situations.

Dr.Kamiya's picture

Both of those are examples of dynamic range compression techniques.

corrective_unconscious's picture

"But in those environments, people aren't typically doing what's referred to serious listening, they are going to be doing more casual listening where sound quality isn't as important."

I feel that was implicit in my post.

And my example Part recording would not need to be compressed until it has no dynamic range. It could be left in its existing, more pure state; or, some allowances could be made without utterly destroying any last vestige of dynamic range.

The main point of my post is that the article seems to focus exclusively on _compression_ wars as being necessary for pre purchase, side by side listening comparisons by the consumer to gain a sales edge. I don't really think consumers do side by side listening in that way. And all the broadcast outlets tend to equalize (in volume and dynamic range) the material, anyway.

I think the extreme compression has more to do with the bulk of the music buying public never _just_ listening to music, so there is always other background noise. So the music must never be quiet relative to its peaks.

A clear solution for disc based media would be a dual layer approach - one layer for the earbud, video gaming, car driving demographic; another layer with more natural dynamic range for the actual listener. (And obviously only for those releases which would attract consumers who even know what "dynamic range" means, of course.)

For computer files: I guess different files for sale.

Both ideas would be manufacturing and distribution inconveniences, at best.

skris88's picture

The solution is simple.

A setting that is in all DVD players known as Midnight Mode (or similar) is included in all Windows, iOS and Android devices, and default to On.

To hear the full dynamic range of the movie (or album) one would turn Midnight Mode to Off.

The average user leaves it On, and gets the compressed and "louder" sound that he or she prefers.

Those who know better can turn it Off when they want the full dynamic range and experience.

Stereophile readers can turn them Off when listening on our hi-fi systems at home, but turn it On when listening through our Bluetooth units in the bedrooms and kitchens, or when we are in our cars if need be.

Record producers and partially deaf musicians would get the compressed sounds they prefer on their Default Midnight Mode On home and portable systems, and leave the sound engineers to do the right thing.

I mean, digital recording systems have a 96dB of signal to noise ratio, far above the 40dB or so in the heyday of analogue multi track recordings, so let's use it!

Stereophile to take this up with Microsoft, Apple and Google. Funding? We'd pay, just set up a PayPal donation page.

Problem solved!

WJ ARMSTRONG's picture

The clearest & most succinct explanation of this topic that I've ever read, so well done!
My brother recently released a new record, so these issues are very much in my head because I was advising him. In the end he went through three (very well known) mastering engineers before finding his favourite! And yes, it did end up being a kind of compromise between the two positions you set out.
Anyone who is interested can give it a listen here: www.youtube.com/playlist?list=PL0kIj-iHv2AOrYIJW8TGVSA1UtiXUpF75

georgehifi's picture

Any one who is pro compression, should have a listen to what they did, to what should have been one of my favourite cd's over the last two years.
Instead I was so discussed with it once I listened to it on a good system I ended up throwing it in the garbage bin.

"Adel 21" is the cd.

https://www.google.com.au/#q=Adel+21+compressed

Cheers George

AudioMan612's picture

Compression has been around since long before the days of iPods. Many popular compressors today are of vintage descent (1176, LA2A, etc.). I love my dynamic range, but to totally ban compression would be quite a problem. It can be difficult to make certain things "pop out" at you without it.

jimtavegia's picture

I have his book and refer to it quite often, but the "business" has done us in. The kids at our High school just care about loud. Crummy earbuds or Beats turned up to 11 and what has driven this is that this business model is working for the industry and why it will continue.

Once one has heard a first generation redbook recording that is left in all of it dynamic glory know that a CD can sound very good, but once someone tries to "fix it", all bets are off. And, as Bob has mentioned, just because the marketing department throws the "remastered" label on it does not mean it was done right or with the original tapes. Too much sneaky marketing going on. With the cost of these new reissues high, reviews become important for one to not make $30 to $50 mistake.

The only way things might change is for the artists to start caring about the overall quality of their product and I'm not sure I see that happening. I sure wish it could be.

georgehifi's picture

Good for finding out who made the most dynamic version so you can hunt it down, and it's usually a s/h older version.

http://dr.loudness-war.info/album/list?artist=Pink+Floyd&album=Final+cut

Cheers George

audiventory's picture

Compression of sound of end-user stuff exists from tape compression. Somebody preffer it digital sound due the compression "improve" sound. Taste matter, me seems.

Non-extremal compression (total or for separate instruments), I suppose, will used always as art tool in music production.

corrective_unconscious's picture

Has nothing intrinsically to do with any saturation on analog tape masters.

And the sorts of compression under discussion have little to do with their being an "art tool." It is a marketing demand.

georgehifi's picture

"It is a marketing demand"

Yes it is a marketing demand, by the marketeer's that have to save money, and have the reco for the "loudest (average level) sound" which nothing to do with the end user/listener they're forcing it onto, which some of newer generation sad to say don't know any better.
I showed the difference to my 17 year old son of an uncompressed sound and compressed sound on my system, and he agreed that uncompressed was way better. Then he went back and stuck his MP3 player back into his ears and totally forgot what he just heard, shame they just don't care!!!

Just listen to Paloma Faith's "only love can hurt like this, it starts off with her singing at normal to low level, then the chorus of the same words she lifts to a very high crescendo, but the crescendo is lower in level to listen to than the beginning soft level, instead it should be much louder. No dynamic shift. WRONG, WRONG, WRONG!!!!

Now look at the compression test of it, no green at all just red red red red for every track
http://dr.loudness-war.info/album/list?artist=Paloma+Faith&album=A+Perfe...

Now look at this Here we have three copies of Dire Straits' Brothers in Arms. One of the most dynamic records of all-time. Notice that in 20 years of EQ, compression and "remastering" it has been robbed of half of its dynamic range:
http://www.thegearpage.net/board/showpost.php?p=13935853&postcount=1

The marketing guys are killing our music

WRONG WRONG WRONG!!!!!

Cheers George

georgehifi's picture

If the marketeering guys want to have a compressed versions for the pubs,clubs and mp3 crew with high ambient background noise, then do it.
But also have a non compressed version, for the millions of audiophiles that want to hear it without compression on their audio systems at home.
I bet you will sell millions of more copies with this option.

Cheers George

georgehifi's picture

Here are two versions of Sade's Diamond Life.

1984 http://dr.loudness-war.info/album/view/76652

2000 (remastered) rubbish. http://dr.loudness-war.info/album/view/17901

Remember green is good, yellow so-so, orange rubbish, and red throw it in the bin.

Cheers George

Les's picture

This is a subject near and dear to my heart, and I agree with Mr. Bob Katz and his sensible sentiments. The solution/problem at hand isn't just a simple "compression = bad" ideology. There are many shades in between.

Historically, the sound of pop music has always been about compression. And compression, in and of itself as a sound-sculpting tool, is neither good nor bad. There are plenty of examples of both. Artists should simply follow their bliss, aurally, even if it is that of a crushed sound. So be it, then. There should be room for that, as there should be for something more dynamic. (I personally much prefer the latter.)

I first started noticing this downward aural trend starting around the year 2000 mark, which I guess coincides with the rise of software-based mastering tools. But I soon realized, this ability for almost infinite control, was a case of too much of a good thing. It's as if people forgot what a good sound was, all of a sudden. And as such, many/most modern releases have this certain "sound."

But even more painful has been the destroying of musical legacy by this new aesthetic. Nothing adds opacity to sound like over-zealous compression. And there are countless more cases, in addition to the examples cited in the article, of ruined remasterings. It has gotten so bad, regular use of the Dynamic Range Database as a reference (to weed out the bad) has been a must.

It's sad to realize that we've had a lost decade or two...

georgehifi's picture

Beat it (Thriller)

Here is the original 1982 first CD pressing
http://dr.loudness-war.info/album/view/28363

And here is a 2001 CD pressing.
http://dr.loudness-war.info/album/view/67242

And here is the woeful squashed Acoustic Sounds 2013 HD download 96/24bit one
http://dr.loudness-war.info/album/view/56849

And the far better Qobus HD download 24bit
http://dr.loudness-war.info/album/view/56968

As you can see the original 1982 cd is about the same DR as the best Qobus HD download, forget the HD Acoustic Sounds version it's squashed like road kill.

Now go on ebay and see which one puts the bigger dent in your wallet.
Audiophiles know which sound better.

Cheers George

skris88's picture

Music with an overly compressed dynamic range has it's uses - in your car a 20db PLR recording would be a problem.

And, agreed, while YouTube normalisation helps keep the loudness of tracks "equal" so there are no "loosers", it's not a default setting in iTunes, thus further adding to this horrible chasing of Loudness by producers and artists alike.

The changes to "Beat It" over the years explains why some people claim vinyl has more presence than CDs, I've always suspected this to be the cause of the claimed failure of digital audio over vinyl.

Audyssey and most DVD players have a Night Mode to reduce the dynamic range of the movie from waking ones' neighbours.

Why not offer the same on all audio systems? Call it the Loudness button but instead of trying to compensate for the Fletcher–Munson curve that the failed switch of the 70s/80s did, this time let it switch on dynamic range compression for those who prefer their sounds with a 5db range - that many do. In fact with my Audyssey I use their Dynamic Loudness setting for exactly that - when entertaining and music is played for background listening, disabled of course for when I have my audiophile sessions.

Lastly reviews need to start measuring and stating the PLRs of recordings. it'll very quickly raise the awareness of this issue and bring quality back to our recorded music.

Ironic that with analog and it's 50db SNRs and 20db PLRs we've moved to 96db SNRs in digital but with only 5db PLRs!

X