Music Into Noise: The Destructive Use Of Dynamic Range Compression

There have been several times where I've wanted to comment on this in reviews, but it's still not entirely possible -- you can't waste your word count explaining the issue, and anyway I'm pretty sure my editor would send back an email that said "OKAY, SOUTHALL."

But you can still get it in there: plenty of reviews will point out that something's "very dense" or "blaring" or "tiring" or whatever, which is a perfectly everyday way of pointing out the issue. And you can go into it directly with re-issues, obviously: I've put short paragraphs in reissue reviews basically saying "they've had to raise the levels to modern standards, and it's made this sound (a lot worse) / (not too much worse) / (just fine)."

bob weston from volcano suns/shellac has opened a new mastering studio in chicago...this thing from their website gives people that don't want to read this whole big thread a good summary of the whole debate:

http://www.chicagomasteringservice.com/loudness.html

If i have my ipod on shuffle and a Sleater Kinney track plays, a little bit of wee comes out of me

I agree with the idea that iPods, car stereos and such should just have built-in compressors. But I do sometimes like the sheer (well, apparent) bigness of that compressed sound; a lot of the time I probably miss the special frequencies, the sense of air, etc. on a well-mastered CD. While from an audiophile perspective something like ZE's Mutant Disco reissue is probably way over-compressed, I prefer its sound to the more modest remastering of the Kid Creole and the Coconuts reissues (the Universal ones).

On the other hand: the reissue of Laurie Anderson's Big Science seems inappropriately loud to me — not that it has no dynamics at all, but recalling the vinyl version it seems like it was, for the most part, a relatively quiet record; there was something modest about its electronics, which now seem in-your face. Maybe the older CD is really better.

The fact that I got acquainted with Big Science on vinyl and Mutant Disco on CD probably has something to do with these perceptions, of course.

Tissp - go to Plastic People near Old Street in London - best-sounding club I've ever been to.

Better than Room 1 in Fabric?

My ears are usually in better condition when I'm there.

I've never been blown away by the sound at Fabric, but I've usually been somewhere else first, and I always seem to end up in room 3 anyway 'cos I can't take the crowds after a while.

I'll have to go entirely sober one time, and check it out properly.

Well, bad things can certainly happen there, too. But also good things. There are many variables which interact with compression (all of the compressor settings, i.e. attack, release, ratio, threshold; the specific compressor's characteristics; the attributes of the track being processed), and sometimes it gets even more complex with parallel compression (mixing a compressed an an uncompressed version of a track) or serial compression (using multiple compressors on a track). Lots of things can happen

Yes, obv there are all the variables you mentioned. But, why are any choices 'good' or 'bad' if they give you a sound that you want (for that individual instrument).

Vinyl to make a comeback and kill the CD?

http://www.wired.com/entertainment/music/commentary/listeningpost/2007/10/listeningpost_1029

Oh, it's on Slashdot too. Sorry about that.

Another reason for vinyl's sonic superiority is that no matter how high a sampling rate is, it can never contain all of the data present in an analog groove, Nyquist's theorem to the contrary.

Aaargh. Still this nonsense persists. That's not why vinyl sounds different/better!

exactly. vinyl is lowpassed weeeeeell below 22k

high a sampling rate is, it can never contain all of the data present

Yeah, how is this even possible if the sound was recorded digitally initially?

Before or since, labels will slow down their tendency to use compression.

But then, first of all, portable CD producers need to stop "protecting" the customers hearing and let them pump up the volume again.

And those who prefer vinyl does so because it does not sound perfect. They are usually the same people who tend to favour live music instead of studio recorded music, and prefer "soulful" playing ahead of more skilled playing.

Master digital recordings often use a faster bit rate and higher-precision resolution than the cd redbook standard.

xxp

sampling rate, i mean

All Scandinavians are baby-raping fascists.

Oops, did I make a gross and unfounded generalisation there?

Master digital recordings often use a faster bit rate and higher-precision resolution than the cd redbook standard.

Yes, and you can noise-shape and dither down to 16/44.1k and preserve much of the extra detail where the ear can actually perceive it.

I love vinyl but I get a bit miffed when the old audiophile mantra about "sampling missing stuff out" gets bandied about as fact and an underlying reason for vinyl's apparently unquestioned superiority. I think it's more to do with vinyl's shortcomings in a fidelity (or information) sense and how those technical failings manifest themselves (mostly) euphonically that explains a lot of the medium's sonic specialness. Not all of it, but a lot of it.

I've probably said this before but I absolutely adore the way ATRAC (minidisc) compression makes things sound. It's like everything's been slightly brightened, thickened, and then coated in a barely visible translucent candy shell. Vinyl recorded to minidisc is my favorite sound.

(Unless it's a Sean Kingston 12" of course)

I'm not sure anyone completely understands how the senses and the brain process sound. Signal processing theory is a useful starting point, but only gets you so far. It could certainly be true that the problem with digital is information overload. My experience argues otherwise -- my limited exposure to SACDs, which contain more info, has been positive, and mp3s, which have less info, sound like crap. Admittedly that's all anecdotal. It's hard to understand , though, why sound engineers would go to the bother of improving on 16/44.1k if all that extra information makes for a less enjoyable listening experience.

Haha they cited Shakey Mo.

i have a flying saucer attack cd single where he apologises that it isn't available on vinyl because they physically couldn't cut it due to some problem with the bass frequencies. take that vinyl lovers, up yours nyquist... 8)

It's hard to understand , though, why sound engineers would go to the bother of improving on 16/44.1k if all that extra information makes for a less enjoyable listening experience.

Who says it does? As you say, it's not just about data volume but many things intrinsic to the replay/coding process; vinyl has this warm, front-to-back depth thing going on (that CD can actually emulate, if tailored to do so), SACD's smoothness and openness could well be down to its lack of decimation filter (it's high-speed 1-bit delta-sigma rather than pulse-code modulation), CD's oft-cited coldness and sterility may down to filter ringing or just the lack of analogue euphonics that a certain listener may expect. I certainly don't have a problem with CD "sound"; I don't think there really is a CD sound, which could be the problem for some people.

I've always felt there's slightly more going on with vinyl - not necessarily "content", but some confluence of ever-present background noise and the physicality of the playback ritual. Neil Young once said something about (paraphrasing from memory) "you play a CD, you hear everything - you have no reason to go back to it; with vinyl you keep hearing new things". Which is a bit fanciful and not really my experience but could be literally true (vinyl deteriorates with each play, things are masked and unmasked).

Koogs: yeah, most vinyl is summed to mono below about 80Hz and then rolled off pretty sharpish below that; if you get out-of-phase low frequency content, then you can't guarantee keeping the needle in the groove. You do hear stories of d'n'b producers insisting that cutting engineers bypass the filters to get that sub-bass down, risking their £20k Neumann lathes in the process. I think of Scotty in Star Trek for some reason.

I'm v hazy on the technicalities, but isn't there, in least in theory, detail missing at 16 bits that you could get with analogue?

So if you mix digitally at 24 bits and then cut to vinyl you will preserve more information than if you put it on a 16-bit CD?

(In addition to the "warm" way in which you lose some of the information.)

We're off-topic, but also what about upsampling CD players, that go back to 24 bit - is that just snake oil?

A friend has one and claims that although he sees the absurdity (you can't bring that information back) it means that the digital-analogue converters can do their stuff better. I'm sceptical.

Well, first of all, you wouldn't simply truncate from 24-bit to 16-bit - you'd use noise-shaped dither and actually retain a lot of that additional info so it's perceptible below the regular -96dB noisefloor of redbook CD. (Sounds a bit counterintuitive and I'm not the best at explaining it - have a Google!)

If you were going to a theoretically perfect analogue medium then, yes, you're obviously retaining more info than if you dumb down to a shallower bit depth/lower sampling rate. But you're not - you're going to a medium which has its own physical limitations on just how much master-derived info it can bear. In vinyl's case, the theoretical 24-bit dynamic range of 140dB+ is crammed into about 65-70dB (at best) and a 96k master's flat 48kHz bandwidth is rolled off at both ends so that we're left with, perhaps, 60-15k. In pure data terms, that's like a 12-bit/32k digital system.

Before anyone jumps down my throat, I KNOW there's more to it than that but Jamie did say "in theory".

xpost - upsampling, in information theory terms, is related to sampling rate and has nothing to do with bit-depth; CD players using 24-bit chipsets are not unusual and it's just a case of doing the maths with more headroom.

Filter ringing -- I sometimes wonder if maybe that's what's going on, that some of that filtered data is folding back into some maybe not precisely audible but still somehow perceptible zone and causing the much-cited listening "fatigue."

OK, I'm struggling here, and I should just go and read some stuff, but that's what ILM is for, eh?

So the limits of vinyl are in the dynamic range (the volume) and the higher and lower frequencies. But, given that, is what's left still not higher resolution than CD (as it's sampling rate is, in theory, infinite)?

xpost

No - infinite sampling rate would mean infinite bandwidth, i.e. no upper frequency limit at all. In sampling theory, that's what the sampling rate determines - you can theoretically capture everything as long as you sample at at least twice the highest frequency contained in the music. So, with CD, you filter out everything above 22.05kHz (which has its own problems - you need to design a very abrupt low-pass filter) and then sample at 44.1kHz and you capture everything that's left.

Vinyl is a continuously-varying system, not a discrete-sample system, so it's probably not very useful to talk of sampling rates with vinyl. It remains true though, that CD's hard 0-22k limit is typically wider than what you get on most LPs (even if you get cartridges with 40-50k upper limits).

It's kinda hard to grasp the sampling theory thing, I know - people often say, "but what about the gaps?" There are no gaps - what you lose when you sample is everything above a certain frequency limit and nothing below it.

Again, in theory.

Ha, yes "what about the gaps" was exactly how I was conceptualising this.

I was also confused by bit depth and sampling rate, but I think I've got it now.

Have you got into all this theory just through being into hi-fi or do/did you work/study in this area?

Well, I did a physics degree and then I got into hi-fi in my mid-20s. I've barely used my graduate study in my working life but it's handy for hobbies... Something similar has happened recently with photography - quite keen to retrieve the optics textbooks from my mum's house!

(I could quite easily be found out if a proper electrical engineer wandered into this thread).

where is shakey mo qutoe

I've just been googling this a bit, and came across these dudes:
http://studio-central.com/phpbb/viewtopic.php?t=10705&highlight=24bit+16bit "> http://studio-central.com/phpbb/viewtopic.php?t=10705&highlight=24bit+16bit

They get on to the theoretical maximum volume of any sound ever, on Mars, and using sound as a weapon, and to cool ice cream. Nerds!

Actually they just linked to his band's myspace page, Jon.

over on my audiophile LP thread I found a neat interview with cutting engineer Stan Ricker if anyone wants to really get into DETAIL...

http://www.positive-feedback.com/Issue1/ricker1.htm

he talks a lot about frequency limitations while cutting vinyl.

http://www.rollingstone.com/news/story/17777619/the_death_of_high_fidelity

Let's bump this one, too, just cos it's the bigger thread overall.

It's about time to impose a ban on all mp3 players.

Also, about time Sony et al start producing headphones that bring a volume loud enough to damage your ears again, or those CDs that do not use that awful compression will sound way too tame.

Another thing: I guess in 10-20 years there will be a bunch of remasters of 00s albums released, to make the listener able to hear all the nuances that were lost in compression on the original release.

Good to see my stance vindicated.

I'm sick of having to knock down the reproduction level of MP3s just so they won't distort when I play them. Not that that fixes the clipping/buggered waveforms in the originals, but it's a start.

This has actually become way worse since 2002, when this thread was started. No?

of course it has.

Another thing: I guess in 10-20 years there will be a bunch of remasters of 00s albums released, to make the listener able to hear all the nuances that were lost in compression on the original release.

maybe. but the remasters will be from mp3s

Definitely worse. Even my wife complains about it now: 'you stuffed up the new (blahblah) album on my ipod, what did you do to it' etc.

"I guess in 10-20 years there will be a bunch of remasters of 00s albums released, to make the listener able to hear all the nuances that were lost in compression on the original release."

yep, there'll be a remaster of the arctic monkeys debut 'as heard on the masters'.

"I'm sick of having to knock down the reproduction level of MP3s"

how dyou do this? can it be done on itunes?

http://mp3gain.sourceforge.net/

MP3s have an internal setting that instructs the player at what volume to reproduce the content. You can change this without re-encoding. If you have an ipod, you NEED to use this thing.

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

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

Yeah, it's about ramming the levels of every component up so high that it all just mashes together, and mastering the final mix beyond saturation point. In order to have some of this stuff sounding reasonable, it'd have to be re-engineered.

ah gotcha, I think I already knew that but forgot (xpost)

yeah it seems like a total ground-up remix from the original tracks is the only way you could fix this kinda thing.

Not at all -- I've seen plenty of bitching from people who finalized mixes they were happy with, and then had the mastering engineer squash it all too loud. It'd be incredibly simple to just master them again. To, umm, "re-master" them, as it were.

They're all different.