Music Into Noise: The Destructive Use Of Dynamic Range Compression

Weezer.

K-ROXOR!!!

"deluded into thinking it sounds acceptable"

Arf! Arf!

"Recently, a prominent individual in the recording industry was asked to serve on a panel that would judge the best engineered CD for the Grammy's. After listening to over 200 CD's, they couldn't find a single CD worthy of a Grammy based on the criteria they were given. Everything they listened to was squashed to death with heavy amounts compression. What they wound up doing was selecting the CD that had the least amount of engineering. In reality, the winner didn't win because of great engineering, he won simply because he had messed with the signal the least. What a way to win a Grammy."

All Weezer stuff fatigues me quickly. I like their music, but can't stand the engineering...

How does that win "Best Engineered"? Whether I agree or not all these quotes are so dumb and useless.

When the Grammy committee talk of 'engineering' how are they able to distinguish what happens in the recording studio from what happens to the final stereo mix in the mastering suite, or does it not matter? Who gets the little trophy?

Someone asked for differences between 'good' and 'bad' engineered music...well, pick up some AudioQuest CDs... whose engineer prefers recording live onto 2 track...yup- 2 track... THEN listen to the over-engineered CDs of most record engineers...

Or check out other audiophile labels... compare the recordings (not on a 200 sony system either).... then come to a conclusion on your own...

In purely electronic music, well, it's difficult to compare - we're not talking about attempting to preserve some sense of people playing instruments in a real space in a real time (or *suggest* that through overdubbing and panning). Things can be squashed to buggery for creative purposes, or have ludicruous dynamic leaps (silence to full-scale in 1/100th of a second); I imagine mastering Ryoji Ikeda CDs simply involves duplicating whatever he supplies.

I think this is exactly it.. but I think it extends further than commerical pop... I think it is endemic to digital recording in general....

Hmmm, can you think of a DDD classical or jazz release which features hard-limiting and no dynamic range? You might be right in the sense that it's a digital-specific thing, simply because trying to do this with tape and an analogue desk would be (I imagine) impossible without obvious saturation effects.

The funny thing is that a lot of folks who record entirely in the digital domain, like to bounce back out to analogue tape (2" to mix, or 1/4" to master) precisely for the saturation/compression effects of tape.

The one record I have which is flat-out in the top 1-2dB throughout its length is Ninotchka's "I've Got Wings" (even Destiny's Child can't match it). But it's a very deliberate act of production and it sounds great!

'No dynamic range' - I dont know how to quantify this...but I can easily think of NEW classical recordings that are shamed to death by earlier recordings in terms of dynamic range.... Im not a large jazz fan.. so I will not comment on that...

However, I think your point is well taken- it is definitely more visible in rock music, but then again compared to classical overall the average DB level is going to be higher by nature of the genre..

The funny thing is that a lot of folks who record entirely in the digital domain, like to bounce back out to analogue tape (2" to mix, or 1/4" to master) precisely for the saturation/compression effects of tape.

Hmm, well I would most definitely argue that it has nothing to do with analogue that caused the saturation/compression but instead the recorders lack of knowledge using analogue that caused it (which was to their advantage in this case)....

oh and are you saying that there arent recordings that average in the 1-2 db range? or that there arent many records that average there? anyway... I have read about many recent CDs that average around 5 dB....

http://www.tcelectronic.com/static.asp?page=bob_katz

Ok, change that to 'severely compromised DR'.

but I can easily think of NEW classical recordings that are shamed to death by earlier recordings in terms of dynamic range....

Could that be down to recording techniques? A multi-mic derived mix vs simple stereo array above the orchestra? I can think of a few small-ensemble digital recordings of mid-90s or later vintage with dynamic range in excess of which was actually physically possible with pre-Dolby SR tape.

However, I think your point is well taken- it is definitely more visible in rock music, but then again compared to classical overall the average DB level is going to be higher by nature of the genre..

True.

Hmm, well I would most definitely argue that it has nothing to do with analogue that caused the saturation/compression but instead the recorders lack of knowledge using analogue that caused it (which was to their advantage in this case)....

Well, not quite - I mean, the euphonic effects of slightly overdriving tape (and the soft limiting effects thus attained) are well known and some artists/engineers like to very deliberately make use of this phenomenon. It might be quicker than trying to achieve the same feel in the digital domain with T-Racks or some VST plug-in.

oh and are you saying that there arent recordings that average in the 1-2 db range? or that there arent many records that average there? anyway... I have read about many recent CDs that average around 5 dB....

Oh, I'm sure there are loads of records brick-walled in this way; I just mentioned the Ninotchka tune as one which is just one block of colour under CoolEdit analysis (hilariously, this is mastered in HDCD where available on compact disc; originally vinyl only, I think), but isn't actually an example of this squashed-to-death fad, but more an act of bloody-mindedness on the part of the producer.

Looking at stuff knocking around my hard drive, I see "Bootylicious" has an average RMS power value (50ms window) of around -6dB; a B&S track more like -13dB; the meaty part of a 1970s Jarrett/Garbarek/Danielsson/Christensen track around -19dB; an Andrews Sisters recording from the mid-50s about the same (that's mono). My own recordings (and those I receive from other people) tend to hover between -11 and -15dB average RMS. Plenty of compression at source in those tracks, but still some sense of dynamics.

Thanks for the Katz link; the TC Finalizer is, I think, rather demonised amongst the anti-limiting brigade - it'll be interesting to see what Katz has to say in his spiel for the product.

Dynamic range compression and OGG/MP3 file size compression are completely different things that happen to have the same word in their name. Hopefully you understand that at least.

I don't think MP3 file compression has any discernible effect on the dynamic range of a sound, unless you get into really low quality.

As for the dynamic profile of an MP3 file being different to the WAV file from which it was rendered well, erm, maybe - you're ditching all manner of temporally masked content to squeeze the file size down, so I guess it's possible that the dynamics may change *slightly*. As Graham says, with super-low rates it's telephone-line quality anyway, most of the HF content has gone and with it a lot of transient peaks. But if all the transients have been flattened in mastering, MP3ing won't make any difference there. It's not like MP3 raises the noisefloor or anything.

Anyway, Shellac (and other albini prods.) might be an example of rock w/o compression.

i suspect the real answer was that albini wanted to hang with page -- i realised albini's grainy sound was very similar to led zeps -- bonham's was the loudest kick drum in the business and jones has commented that their whole bottom-end was different from anyone else's for that reason, but maybe not only that reason -- both page and albini make rock bands sound raw but clean, yet they're not compressing

maybe it's because the triangular relationship between bass, rhythm and lead guitar is a lost art, with stupid unison rhythmic bass'n'riff being all todays bands can do -- bass playing in particular is not the complimentary quieter element it was for AM music

i believe that there is a small bandwidth that all FM radios can re-compress (loudness button) into a marketably heavy sound -- since FM receivers now come in all shapes and all sizes, music that sells must fit into the bandwidth acceptably common and heavy to sound ok with any receiver, definitely a subset of the led zep dynamic

As one of the links from the original URL posted at the top of thread mentions, FM radio is already massively compressed (Optimod and the like), so the compression at the mastering stage to make it radio-friendly is a bit of a waste, and maybe even self-defeating (compressors fighting against each other = pumping).

Wasn't Pete Waterman's trick to roll-off everything below the lower-mid, so he could get a higher average level broadcast on radio? Get rid of the bass energy. Sting would be followed on late-80s R1 by Sonia and *bang*, it had a couple of dB more impact. (Arguably, Sonia has always had more impact than Sting anyway).

Is there a technique for cleaning out some of the overcompression on a WAV file ripped from a CD?

i am grateful for this trend in CD mastering because it means more things are listenable on it.

in other words: mixing things to sound as good as possible on radio play ISN'T INHERENTLY BAD YOU FOOLS

Non-technop0rn answer: No.

Once those transients have been squashed flat there's no way to restore them because there's no information in the signal to tell you what they were and when they were - it's irreversible.

*However*, one of the clever tricks Pacific Microsonics developed for their HDCD system (20-bit resolution on a 16-bit CD, goes the spiel) was just such an embedded code - HDCDs played back on a regular CD player had slightly compromised dynamics (but supposedly great sound due to careful mastering) due to low-level compression and soft-limiting. Played back on a HDCD-capable machine (the HDCD decoder being part of the reconstruction filter in the DAC chipset), this compression would be undone, and the transients unpacked.

I'm not sure how well this works; there was a lot of fuss recently on one of the audio newsgroups about a Roxy Music re-issue. It was proposed as a shining example of the improvements in digital technology: the 1999 HDCD version allegedly sounding miles better than the original late-80s CD issue. Someone then pointed out that the new version had actually been compressed to all hell (in the modern manner), which led to moments of stickiness wherein it was kinda implied that maybe a few audiophiles had fallen for the 'louder = better' trick. Ah, but HDCD *restores* these squashed peaks, yes? Looking at a ripped WAV isn't going to tell you the whole story - you've got to record the thing in the analogue domain to capture what the HDCD decoder is doing. Well, I had a go and it didn't look much different to the digital rip. Inconclusive. By this time everyone had moved on to arguing over cables again.

(Oh, and if yr thinking "20 bits resolution on CD? We can do that in critical narrow-bands with dither and noise-shaping from higher-res master". Well, yes you can. But HDCDs make the little green light on my CD player come on!)

We've had record labels sending us both radio cuts and 'normal' versions of singles, with the only difference being that the radio one would sound flat and horrible. As for getting our own releases on the radio... well, 'not compressed enough' was a handy excuse occasionally trotted out.

A feeble one. That's what radio stations have compressors for, surely.

Your definition of 'listenable', then, seems to approach my definition of shite.

in other words: mixing things to sound as good as possible on radio play ISN'T INHERENTLY BAD YOU FOOLS

It is when what you are doing is compressing a record to the point that you are compromising its quality- in this case its dynamic range... (the range from the 'lowest' to the 'highest' sound)..The music for your radio is going to sound like shite whether it is produced for a high end audio system OR for radio... since the output is shite.

But the trend continues because for the most part, the public listens to shite, on a shite system or in the car.. while talking on their cell phone, making reservations for their Tai Bo class....

Oh and related to another post.. the term 'compression' as it is used here has nothing in common with the way that MP3s are 'compressed"-

(I'm lazy)

http://www.informatik.fh-hamburg.de/~windle_c/e_index.html
.. such as "Warning: Pink can be dangerous for health!"

Thanks for the synopsis ....

this thread isn't very funny ;_;

serious custos is serious

IN A WAY, ABUSE OF DYNAMIC RANGE COMPRESSION MAY BE COMPARED TO WRITING AN ENTIRE POST ON, SAY, ILM, USING ALL-CAPS-LOCK. I THINK THIS SUMS UP BETTER THAN ANYTHING ELSE WHAT THE LOUDNESS WAR IS ABOUT.

How does range compression affect the way people speak? Do they start to sound like an ELO record?

If something using range compression is playing in the background, people will start to sound like James Hetfield. To be heard.

Anything that makes people sound like an ELO record should be encouraged!

I wonder how much of the compression on 00's albums was added BEFORE the mastering process?

-- sleeve, Thursday, 3 January 2008 20:41 (6 days ago) Link

that's not the kind of compression that's the issue

-- Jordan, Thursday, 3 January 2008 20:45 (6 days ago) Link

Well, it's part of the issue. Most of the time people are compressing individual tracks and then compressing the master again. The loudness war tends to refer to mastering compression, but too much mixing compression can produce bad effects too.

the right sorts of mix compression can drastically reduce the need to overlimit tracks just for reasons of volume.

I recently made a driving mix for CD and was frustrated by the huge range of levels. My remastered version of The Who's "Out In The Street" sounds great, especially in the car, on a shitty boombox, not so much on headphones. I loaded it into Audacity and had a look at it, and it had such extreme clipping, it looked like a solid block. Given the originally distorted sound of that particular song, it kind of worked. I went ahead and compressed and clipped the hell out of most of the other songs on the mix to try to match those levels. I couldn't bring myself to go to that extreme, but at least the rest are just about as audible in comparison. Kind of micro-model of how the compression trend snowballed I'm sure.

If in doubt, just download the free software and look at the waves to compare different music:

http://audacity.sourceforge.net/

My hope would be that as more stuff becomes available for download, they'd start to introduce decent masters of uncompressed .wav and lossless codecs along with the .mp3 and .aac for mass market. However, I don't know that the "audiophile market" (those who listen to music at home on decent speakers rather than on noisy trains and in their cars) will ever be big enough to be catered to overall.

there are some classical companies that are offering uncompressed downloads in .wav form...huge files obviously but i think the market is starting to develop.

I've been trying to get most of the stuff on my ipod normalized, so that I can listen on shuffle at whatever volume I want, without massive shifts from track to track blasting my ears off. In doing this, I've been opening more or less everything in a wave editor. So fucking appalled by the clipping/compression on a lot of stuff I've downloaded/ripped, that I'm just basically trashing everything that looks like a solid block. Shit sounds hellish shrill, anyways.

there are some classical companies that are offering uncompressed downloads in .wav form

There are rationally mastered CDs being sold. They just aren't in the US mainstream pop rock market. Most of my import remasters of classic rock are OK. Plus there are still a lot of undermastered CDs from the early Nineties still floating around, particularly at budget price.

In contrast, everything that comes out of Nashville is set to blare at any volume. I always have to remember to turn the stereo down when I go from the former to the latter, say, like Sugarland's new one.

I don't listen at all on earbuds. MP3's sound noticeably inferior to me. Most of my stuff still comes out of a nice but not extravagant stereo, only about one percent sitting at the computer, a terrible way to listen to the music I like.

I don't know that the "audiophile market"

Incidentally, I'm hardly an audiophile. Since the standards have been lowered (or twisted, perhaps) so much, it would seem anyone who doesn't listen on iPod or computer is deemed an audiophile.

MP3s seem to be decent at maintaining the dynamic range of the source recording, at least at reasonable bitrates (192 or so).

Incidentally, I'm hardly an audiophile. Since the standards have been lowered (or twisted, perhaps) so much, it would seem anyone who doesn't listen on iPod or computer is deemed an audiophile.

Yeah, that's why I put the quotes in. I'm audiophile by mainstream standards, but real audiophiles would laugh at my rig. I've enjoyed upgrading my speakers, but can't yet bring myself to buy pre/pro and amp that costs as much as a car.

I'm hardly an audiophile either, though for listening at home through decent speakers, I'd rather have a better source than MP3. Though for listening in the car, I think MP3s are fine, since the road noise drowns out the fine detail anyway.

High bitrate MP3s don't sound atrocious to me. In a quiet room, on a nice system, you notice the loss, but even then, it's not so profound as some make out. Then again, maybe I don't have the ears/gear. Either way, I don't like the sound of hypercompressed audio. Cuts through the chatter, but so does a car alarm.

No, they're not terrible by any means. Sometimes the MP3 actually sounds better than the original - depending on the recording. They tend to round off the upper treble a bit, which can improve some harsh trebly CDs. Sometimes reducing the density of detail helps the ear pick out the important elements more easily - so I think they can actually be more pleasant to listen to for certain types of music. You do lose small subtleties of texture and detail though.

let's not confuse compression in mixing & mastering with audio format compression (mp3s)

Yeah, maybe the mp3 discussion belongs on a different thread.

Yeah, though I was referring to MP3s, I was talking about dynamic range compression & clipping in mastering (at least I presume it's in the mastering, and that these things weren't actually mixed for bricklike sound).

http://www.nme.com/news/metallica/39816

nme website readers comments don't see what the fuss is about

MASTERER of puppets LOL

DEAF magnetic

https://www.youtube.com/watch?v=3Gmex_4hreQ

now it makes a lot more sense to me (though i could hear the problem with certain releases, i couldn't really figure out what was going on in regards to the mastering process)

no doubt you'll tell me this aint the crux of the issue, but hey ..

Coo - Jarvis Cocker just quoted a bit of Nick's Stylus essay on Radio 6.

Someone else told me this; what was the context, which bit did he quote?

i genuinely have still never noticed this

though i know someone who swears down he can't tell the difference between a 320kbps and 128kbps mp3, which is just completely o_0 to me, so maybe it's like that

I think he was reading a bit that was quoted in Perfecting Sound Forever?

It's about an hour and 10 mins from the end of the show if you look it up on iplayer anyway.

How very bizarre to hear Jarvis speak my name on the radio. My mum will be thrilled; she's from Sheffield and knows who Jarvis is!

I can't say I've ever consciously had a big problem with this but that Iggy experiment is pretty blatantly obvious.

Link?

http://web.archive.org/web/20060612221324/http://www.stylusmagazine.com/articles/weekly_article/imperfect-sound-forever.htm

Oh, thanks! :-)

I've read the article more than once; I have the issue of Best New Music it's in. Just looking for the Jarvis bit. I'm going to go searching.

http://www.bbc.co.uk/iplayer/episode/b00qhrx6/Jarvis_Cockers_Sunday_Service_31_01_2010/

55 minutes in

The Loudness Wars: Is Music's Noisy Arms Race Over?

For genres like pop and rap that already used heavily-processed sounds, this wasn't a big problem, and some say limiting has been a productive tool. For music that uses live recordings of drums, guitars, and piano, however, such processing arguably ruins the experience of listening to music made by humans. The biggest furor surrounding loudness centered on Metallica's 2008 album Death Magnetic, a piece of music so loud that some fans called it "barely listenable" and prompted one person to complain that "to hear this much pure damage done to what was obviously originally a decent recording, in the mistaken belief that it sounds good, is hard to stomach." At the time, the outlook seemed bleak. If there was no impetus to get quieter but every advantage to pushing volume to the maximum level technology could achieve, why wouldn't the trend toward increased loudness continue forever?

To counter this seeming economic inevitability, some critics of loudness turned to legal remedies. Audio engineer Thomas Lund has been working in Europe to lobby for governmental regulations on a standard loudness limit on all CDs and digital music. (The limit has so far been adopted as a universal standard by the International Telecommunications Union, which describes itself as "the UN agency for information and communication technologies.") You already have something like this at home if you use iTunes: Just check the box that says "Sound Check" in the preferences menu and the volume level on all of your songs will be equalized. Lund's proposal would do the same thing for any music you could buy.

Taking advantage of the trend towards listening to music from the digital "cloud"—via services like Pandora, Spotify, and Apple's forthcoming iCloud—the proposal would institute a volume limit on any songs downloaded from the cloud, effectively removing the strategic advantage of loudness. "Once a piece of music is ingested into this system, there is no longer any value in trying to make a recording louder just to stand out," said legendary engineer Bob Ludwig, who has been working with Lund, in an email. "There will be nothing to gain from a musical point of view. Louder will no longer be better!"

But while the proposal has seen some success in the EU, it seems unlikely that audiophiles could rely on the US government to take a similar stand, in large part because it isn't a matter of public concern. "I don't see it happening," wrote Greg Milner, author of Perfecting Sound Forever: The Aural History of Recorded Music, in an email. "I think the general increase in awareness regarding the issue is more than counter-balanced by the fact that, by and large, nobody (in a sweeping, generalized sense) cares about music sounding 'good' in some sort of rarefied way. It's more important that it be heard above the noise of everyday life, since we hear so much of our music on the go."

Indie songwriter Owen Pallett went so far as to record all of the vocals for his 2006 Polaris Prize-winning album He Poos Clouds without compression, a step not taken since the early days of sound recording.

this is a weird and out-of-place detail. I'm no expert on sound recording technology, but surely recording without compression and mastering without compression are two completely different processes. and applying dynamic range compression to individual vocal tracks is different from applying a uniform level of compression to the final mix (vocals, instruments, and all). the loudness wars brouhaha is only really concerned with the latter practice.

besides, it's not even true, according to Owen:

He Poos Clouds is uncompressed, except for one note. (The timpani hit right after "I'm just made" on the title track).

― Owen Pallett (Owen Pallett), Tuesday, August 1, 2006 12:06 AM (4 years ago)

Whoop. I lied. We did compress the vocals. But everybody compresses the vocals, it sounds weird without it.

― Owen Pallett (Owen Pallett), Tuesday, August 1, 2006 2:24 PM (4 years ago)

yeah, vocal compression is almost necessary.

http://arstechnica.com/apple/news/2012/02/mastered-for-itunes-how-audio-engineers-tweak-tunes-for-the-ipod-age.ars

That's one for the "if you can't beat 'em, join 'em" files.

Format conversion, dithering and compression are different beasts than "dynamic range compression". Still, an interesting article.

I had a WTF moment when I ripped Youtube audio for a DJ set and decided to tweak the EQ in Logic. I was surprised at how muffled the track sounded compared to the other songs. Flipping on the frequency analyzer, it seems that Youtube audio contains NO audio information above 15 kHz.

ha, i did the same thing recently. did you use it? i overdubbed some tambourine and lasers.

Smart! No, nothing as cool as that, I used a gentle plug-in called Vintage Warmer, which simulates tape saturation. It didn't *really* do the trick, but I went with it.

Then I e-mailed the friend who'd played the track for me originally and asked him for a copy of the CD version.

(The track was "Jon E Storm" by Dog Ruff. Good track! I don't even know where it came from, some German electroclash compilation.)

here's the one i used: https://www.youtube.com/watch?v=wzQ_xGsxgvs

but since the mix isn't yet, you inspired me to get a legit copy, so thanks!

WHOA! Sounds muddy as anything, what a mess. (20% suspicious that the problem might be in the mix entire.)

I mean, his ssss's are all there but Magnolia sounds like she's shouting in the basement; the drum machine and high end on the sawtooths are non-existant, etc. Youtube audio! Fuggedaboutit.

i think it might be a radio rip too - i downloaded an mp3 that sounds waaaay better.

https://gist.github.com/kylemcdonald/1fab024c9878106b486deb1a26bc2079

http://i.imgur.com/UoQIzG8.png

the tempo plot is super interesting btw

http://i.imgur.com/0wNMcw9.png