I think the increasing importance of the car as a place where we listen to music may be a contributing factor. I have some old, well-mastered CD's which I love to listen to at home but they're hard to listen to in my car. The quiet sections are too quiet to hear above the engine/road noise and if I turn the volume up, the loud sections are too loud.
I'd much rather have a well-mastered CD and a 'Loud' button (as many car stereos have nowadays) to reduce the dynamic range for in-car listening. However, the record companies know not everyone has a 'Loud' button or knows how to use it so perhaps this is just one more reason they're killing dynamic range in their products.
Hmm. A more cynical explanation is that they all want to have the loudest CDs in the jukebox. In noisy public environments (pubs, parties, etc.) the volume of whatevers playing music is likely to be set according to the loudest tracks that are played. As a result anything quiet become hard to hear over the background noise of folk chatting. So you end up with a sort of prisoners dilemma scenario where it seems like you have no option but to master your CDs loud.
Similarly, adverts on TV are usually much louder than the shows they're interrupting to try and grab your attention and they all compete with each other for that attention. Very annoying.
Or you could pay jukebox manufacturers to put in a chip that does the same thing the radio broadcast techs do--flattens out the average volume-per-track to whatever the volume knob says it should be.
If the volume knob on my CD player actually said "mean track SPL" and had real measurements on the ticks (calibrated for a certain set of reference speakers, obviously), rather than just going "to 11", I'd be a lot happier (especially at night when I might have adjusted my volume while listening to a quiet song, and then the next song wakes up my neighbours.)
The car is one, but I would have thought the iPod was the obvious example (and other portable music players, of course).
People use them to listen to music in noisy environments with bud-style headphones. If the dynamic range is too high, you simply can't hear anything too quiet in the mix. And if you do turn it up, you'll be deafened when the loud part kicks in. I consider myself a bit of an audio nut but even I choose "hot" mixes when I'm sitting on the train.
Not sure what the solution is other than providing two mixes of each track. Perhaps a feature for future media formats.
It shouldn't be too difficult for such players to analyze the range of a track in advance and compress it appropriately in real time, perhaps to a user-specified degree.
1. Chicken and egg: Until such a feature is ubiquitous, labels can't rely on its presence, and won't release in a suitable format. Until there is demand, player manufacturers won't include it. And I don't see much independent consumer demand.
2. There is no one algorithm or processing technique which will give optimal results for all inputs. The process is still an art requiring a lot of skill and, well, taste. Mastering is still a real necessity. There's crude automated compression, sure, but for the real end product I don't think anyone wants that.
3. The sound is already mixed at the consumer end. For successful remastering (which is basically what you're doing on the fly) the separate tracks would be needed. For example, in an otherwise uncompressed track, a distorted electric guitar is pretty much naturally compressed. Drums are the opposite. Naively just compressing the whole thing will keep the guitar and lose the drums. I'm really oversimplifying here but hopefully you get the picture.
I do think such an approach could work, but you'd need to supply the mixable tracks, or relevant groups of them at least, and then supply mixing/compression instructions to the player. If the instructions were standardised with fairly predictable and consistent results, that could be a way to implement such a scheme.
I used to rip my music losslessly, and then process tracks using an unholy mixture of iTunes, AppleScript (shudder), perl (semi-shudder), and ecasound (command line sound processing) to add audio-domain compression for use in the car. It made a difference, but not enough to bother fixing my script when a new version of iTunes broke things.
Cars are definitely a reason for this. In my own experience, where I do use the car as a place to check how things side, you have to remind yourself that it's not the ideal environment. Or maybe producers today are just lazy
It always puts me a bit off when I see people trying to apply science to an aesthetic issue like this one (this article is way less guilty than many others on the subject).
There is no doubt that engineers are mastering pop music "louder" now than in decades past, but it's too easy to jump to the usual conclusions of there being some kind of Volume Arms Race, with each successive album attempting to be louder than the others surrounding it.
There's a hard limit of 0dB that digital recordings can reach before digital distortion, and it's not like professional audio engineers are trying to push the envelope there, curious to see what happens.
The real reason pop music (and we're only talking about pop music) is mastered so loud these days has much more to do with the general trendinesses of pop music than some kind of unscientific and unjustified commercialism.
In other words...
In 2009 Kanye West is a popular musician. When asked why they like Kanye West's music, people often say things like "it's got a good beat to it," or "I like dancing to it." People think Kanye West's music has a good beat because of the overall mid-scooped equalization of the track, and the effect of compressed and "loud" mixes coming out of large speakers. It makes people want to dance, and the dancing makes them happy.
But, when they get home from the club, exhausted and tired of being on their feet all day, they put on a relaxing Sufjan Stevens (or whatever the kids are listening to these days) album. The mastering is far lower in perceived volume, and the same person who was dancing to Kanye West because it "had a good beat" is now saying how much Sufjan Stevens lets him "mellow out."
Think of the difference between top-selling pop music today versus 20 years ago. I think that alone does a better job justifying common engineering practices than conspiracy theories of being the loudest song on the jukebox, or on a radio segment or whatever. Especially considering that there's a volume ceiling that is already being reached.
No way. It's not just a matter of people liking to listen to both Kanye West and Sufjan Stevens. I do like people listening to different types of music, but I don't think this is something that you can just attribute to "trendinesses" of pop music.
Reducing dynamic range is a problem. Try and listen one of the albums notorious for overcompression, like Red Hot Chili Pepper's Californication. You get "tired" listening to the music at such a high level of compression and volume, and the dynamics prevent the songs from having the impact that they could have. For example, on that album, the track "Parallel Universe" has essentially no change in volume between the verse and chorus and so the whole song sounds hollow and empty through a good pair of headphones or speakers, yet it sounds great live (even on tape). The songs sound fine if listened through cheap iPod earbuds or the radio, but if the album sounds exactly the same on the radio and in a home CD player, I think there's something wrong there.
The problem is that if it takes overcompression and loudness to get people to notice your music, music will continue to become louder and more compressed. If only those songs get peoples' attention and get radio play, pretty soon every song you hear will essentially have the same volume. That's not only terribly limiting to the artist, but a disservice to the fans who listen to the music.
Now, most music producers are aware of this and take pains to make a good mix, but raising awareness about this certainly would help stop this process in much of commercial music, especially as recorded music gets more popular.
I agree, and the newer Red Hot Chili Peppers albums are a perfect example of this.
Ever look analyze a Muse CD? The audio looks like a solid block zoomed out. Not that Muse isn't awesome.
The article seems to simplify the clipping issue a bit much, and it looks like some users here are confusing compression with maximization.
Compression is the relatively simple process of reducing dynamic range above a certain threshold by a certain ratio. If the ratio is 2:1 and the threshold is -10db, then 0 db would end up at -5, -8 would end up at -9, etc. Then you'd just make up the loss in volume by boosting it an equivalent amount. There's also the attack and release, where attack allows the volume to rise above the threshold for a given amount of time (measured in milliseconds) before it is compressed, and release will raise the volume up over a period of time after the source has returned below the threshold.
What's going on here, it's important to note, isn't quite compression. It's a much more complicated process involving potentially lots of different plugins and rack-mounted magic boxes that all sell for thousands of dollars. They focus specifically on the qualities of sound that make us perceive loudness, not just simple dynamic range.
(Though, to be fair, one of the most interesting mastering tools is the multiband compressor, which splits the audio based on frequency, compresses each band separately, then mixes them back together.)
Anyway, the focus of a 'loudness maximizer', as opposed to a compressor, is pure... loudness. That's where the clipping comes from, and that's actually how you get clipping without the sound actually sounding distorted and clipped. It's pushing the edge of distortion a little bit, and you can kind of hear that on poorly mastered loud recordings.
It's a lot harder to get the same effect with a compressor and a hard limiter (which prevents the audio from going above a certain level no matter what - basically a compressor with an infinite ratio).
edit: By the way, because of imperfections in the original compression hardware (and modern software simulations of said imperfections), the compression process isn't actually as straightforward as I said. In theory, that's how it works, but compressors are generally known to 'color' the sound a bit while they compress. In other words, the compressor that the Beatles ran vocals through is not capable of also giving you that pulsing, compressed sound you get from One More Time by Daft Punk. For example.
Just so happens I was digging up another old article on this subject yesterday by a guy called Rip Rowan. In his article he looks exclusively at how Rush's albums have been mastered:
I make hip-hop beats as a hobby. I've had industry people listen to the same track, one with a wide range, and the other "hot" and its always the "hot" track that grabs the attention. Loud is definitely the desired outcome in the music industry. I found the chart on this link very interesting.
Another possibility is that a lot of production and composition has shifted to virtual instruments and computer mixing, often times done by folks who didn't have the engineering training that producers had 10-20 years ago. The computer has allowed a lot more people to create music. Mixing a song with a wide range that sounds professional or even "commercial" takes a lot of time and skill. Making a mix hot takes some level of care but not nearly as much as the other side. If you're trying to make it as a producer, are you going to spend your time mixing amazing tracks or making more tracks?
I think there's a difference between hip-hop music and rock music, though. It's less important for hip-hop to have the dynamic range because the majority of its emotional expressiveness comes from the words, rhythm, the way they interact and the changes thereof. This is why having a great beats is so important and so good producers attract a lot of attention (e.g. Neptunes, Dre). Volume is not as important a factor. You really only need two levels of dynamics: one for the verses, one for the chorus/hook.
On the other hand, I think changes in volume are more important for expressiveness in rock music, for instance, in an unexpected bridge just when you feel the song is about to end. Now, most bands these days seem to compensate by having simpler dynamics, but I think it would be nice for them to have more options.
Also, wouldn't it be easier to have a good mix now that we can do most of these things on the computer instead of having to manually twiddle knobs and overdub live?
I find it interesting that the author refers to record companies. It's almost always the producers who make these kind of decisions. Most producers go with the current trend in terms of sound. If a producer wants to give me an 80s sound à la Bryan Adams, I'll go somewhere else. So it's the musicians as well. Everyone wants to sound "modern". Currently, this means loudness (and clipping, apparently).
Great article! Expect more "loudness wars" both in music, film, and other technology. My favorite example of this is all of the program icons that appear in your system tray, each with its own popup. It's just a visual example of the audio problem.
If you really care that much about dynamic range in music, learn to play a few instruments yourself. Quit whining about shrink wrapped commercial products, because they're never going to be up to the standards of a connoisseur.
I'd much rather have a well-mastered CD and a 'Loud' button (as many car stereos have nowadays) to reduce the dynamic range for in-car listening. However, the record companies know not everyone has a 'Loud' button or knows how to use it so perhaps this is just one more reason they're killing dynamic range in their products.