Why Music Really Is Getting Louder

Teksty Piosenek writes "Artists and record bosses believe that the best album is the loudest one. Sound levels are being artificially enhanced so that the music punches through when it competes against background noise in pubs or cars. 'Geoff Emerick, engineer on the Beatles' Sgt. Pepper album, said: "A lot of what is released today is basically a scrunched-up mess. Whole layers of sound are missing. It is because record companies don't trust the listener to decide themselves if they want to turn the volume up." Downloading has exacerbated the effect. Songs are compressed once again into digital files before being sold on iTunes and similar sites. The reduction in quality is so marked that EMI has introduced higher-quality digital tracks, albeit at a premium price, in response to consumer demand.'"

Comments

It is not too loud!

[mi]

You are too old!

Also, the Loudness War

[antdude]

VideoSift mentions an one minute and 52 seconds YouTube video showing big-name Compact Discs (CDs) [and other audio sources] manufacturers are distorting sounds to make them seem louder. At the same time, sound quality suffers.

Terminology confusion?

[megabyte405]

Is it just me, or does that article (intentionally?) confuse the two meanings "compression" can have with regards to digital audio? The loudness bit is audio compression: reducing dynamic range (which they do talk about). Then, they bring in the bit about data compression and the EMI iTunes Plus downloads, which is entirely different (admittedly, it also introduces artifacts, but of a completely different nature). The bit about the Los Lonely Boys album "compression-free" could easily be free of either (or both!) kinds of compression.

While the logical part of me chalks it up to confusing terminology being misunderstood, part of me wonders if those meanings are being intentionally conflated to make the article more impactful... it would sound less impressive if EMI wasn't "admitting there is a problem with compression"

Peaking

[zerocool^]

We always called it "peaking", and it's something that everyone who's recorded an album in the spare bedroom of their band mate's house can attest to - if you record with fewer peaks (places where the sound wave maxes out at the top of the available volume area), it sounds better. It just plain sounds better.

But, take songs off that CD and slam them onto a mix-tape style rotation or an iPod, and you'll be reaching to turn up the volume every time your song comes on.

From what I can tell, recording engineers are responding to the bands who don't want people to have to turn the music up (in particular record execs). It's one of those terrible problems - if everyone would agree on such-and-such date to back off the recording volume and get less peaks (say, no more than 7 per album), everyone's music would instantly sound better. But the fact that everyone's competing, and you don't want your copycat pop punk band to be the quiet one, means it's a self perpetuating problem.

~X

Re:Peaking

[varkatope]

The issue isn't the peaking itself. A peak happens when a signal has surpassed what the receiver of that sound was designed for. If you pump a really loud signal into a preamp on a mixing console (even a small cheap one these days) and the "peak" light has come on, it means the signal is too loud for the equipment. It results in audible distortion and you should turn it down. What a compressor would do in this case is take the full spectrum, from lowest to highest point of the sound frequency and compress in a way that in effect, makes the highest and lowest frequencies squish into a tighter waveform. It's like trying to fit a square peg in a round hole, but then shaving off the corners of the square peg to make it fit. In effect, you're avoiding actual peaking or overloading of the equipment so you can turn the signal up louder without overload. Therein lies the problem.

Compression is a necessary part of recording. Judicious use of compression can make a mix really come together and fit everything into it's right place. Notice that I said judicious. It's unfortunately a very useful tool which can easily be abused. OVER compression starts to result in the degradation of the signal. Sometimes you can hear it "pumping and breathing." Over compression is nasty, plus it destroys dynamics. Forget that crescendo on the second movement. Your violin solo is now exactly as loud as your entire orchestra. Are you excited yet? It's also extremely tiring to listen to. Take a pure square wave and pump it through a speaker. Look at that speaker and notice how fast and constantly the speaker cone is vibrating. Take your newfangled over compressed rock/pop CD and extract audio into some sort of multitracking software like Pro Tools or Ardour even. Expand the view a bit and look at the wave form. Looks a lot like a square wave the way the tops and bottoms of that wave form are chopped off doesn't it? Extract audio from a cd you really like from say, the mid 60s. Look at the wave form. There are peaks and valleys and quiet parts and loud parts, the tops and bottoms of the waveform are not chopped off. Now imagine what that new over compressed pop/rock record is doing to your eardrum even at low volume while keeping in mind the speaker cone. Your ear works a lot like that speaker cone. It's vibrating exactly as fast as that speaker cone. It's a mechanical part. There is fatigue involved. Plus it's just boring to listen to. It sucks out emotion and excitement.

By the time you hear your average top 40 hit on the radio, it has been compressed during recording twice (on individual sound sources and probably again when a stereo mix is produced), during mastering, then again at the radio station. Radio stations want their station to catch your ear, plus it helps in keeping signal strength over long distances. Labels want louder songs to compete with the other loud songs, bands want their record to sound like this other loud record, mastering engineers are asked by either the band or the label to make it as loud as possible. You know who's paying the bill so they do it. Recording engineers can be pressured to over-compress by the band or label or just by wanting to have a job in the next year and they might do it as well. Even if they turn in a good balanced mix for mastering, it's a crap shoot whether their mixes will sound the same when the record actually gets put out.

It's a shite state of affairs all around.

Good audio example

[Guanine]

Here's a great audio and visual (narrated) example of the "loudness wars" and the way that reduction in dynamic range reduces the quality of the recorded sound. Keep in mind, this isn't audiophile mumbo-jumbo... this is a very real and very unfortunate trend in what the engineers who master albums (specifically pop albums) are required to do to keep their albums "competitive" with all the other loud albums.

it's called normalize-audio

[Erris]

Couldn't we just add a tag to every track with a floating point number by which to multiply the magnitude of all the samples in that track by default.

You already have a built in upper limit, normalizing the range to that limit fixes the problem.

Normalize-audio is a package that does this. Here's what the Debian repository says:

normalize-audio is a tool for adjusting the volume of WAV files to a standard volume level. This is useful for things like creating mix CDs and mp3 databases, where different recording levels on different albums can cause the volume to vary greatly from song to song.

The package also works on ogg vorbis and mp3. You can do it on ripping, or playback. Each song can be normalized individually or as a collection. The result is that you don't have to reach for the volume knob all day.

You are SOL if the record company has already applied dumb techniques to the CD before you get it. Peak "compressing", where all of the peaks are maxed out is a real distoriton of the original sound. When you add a heavy handed turn up that clips as well, you get Californication as mentioned. As the article also notes, it's difficult to digitize clipped audio. A clipped wave is like a square wave - it has all frequencies and takes lots of bandwith.

Cranked up to 11

[tbo]

It's tough being able to hear.

I know what you mean, and I'm not even old and wise. I went to a concert for the first time in a few years, and was reminded of why I stopped. I had to wear ear plugs most of the time, which, since they don't attenuate all frequencies evenly, totally messed up the sound.

Imagine if, when you entered an art gallery, they stabbed out one of your eyes. That's how much sense it makes to destroy people's hearing when they go to concerts.

That is partially right, but there's more to it...

[yroJJory]

WHY are albums mastered so damn loud?

It's a vicious circle and it is caused essentially by one feature: shuffle mode.

Here's how the problem reveals itself:

Band A decides they want to have the "heaviest, loudest album ever made", so they tell the mastering engineer to make their master louder.

Band B is hears Band A's album and wants to be louder (or at least AS LOUD) as Band A. So they tell their mastering engineer to pump up the volume, too.

Assume the same thing happens with Bands C through L.

Now Band M comes along and they've had these other 12 albums playing on iTunes while they're mixing their album. Band M isn't so concerned with being "the loudest", but when the put their ref CD into iTunes and are listening in shuffle mode, their songs get completely drowned out by a factor of 6-12 dB of amplitude difference.

So Band M now asks their mastering engineer to make their master louder so they'll match up with everyone else's.

And Bands N-Z follow suit.

It's a very difficult domino knockdown to break out of, since no one wants to make the album that is super quiet and requires intervention with the volume knob. (Yes, I'm aware of the "Sound Check" feature in iTunes, but that's just a lousy attempt to solve the problem with technology.)

In 2005 I recorded an album for a Hawaiian band. It was gorgeous and I convinced the band to master the album at Universal because I knew the main mastering engineer and was adamant that he was the ONLY guy who could do the record justice. I was also adamant that the album did NOT need (and would avoid) any compression.

We only boosted the overall level of the album by 4 dB and that was purely using a limiter to ensure no overs.

I then sent the first ref CD to the band member who couldn't be present. He was thrilled with the mastering but had just one question: Do they make it louder when the CDs get pressed?

I told him that it was at the level I was recommending and that Mastering was the time to change levels, but that we really wanted it to sound good, not loud. His response? "Oh. But it's so much quieter than every other CD I own."

And he's right. Compared to every CD that has come out in the past 5 years, his album is seriously quiet. Possibly as much as 8 dB quieter than current albums. And maybe we did it TOO quiet. But it matches in amplitude to CDs that came out in 1989, back when some dynamic range was still an OK thing in music. Nowadays we don't like ANY dynamics.

So who is right? And can we go back?

I've been a HUGE advocate of dynamic range and NOT destroying our months of hard work at the last step in the process. But I can only do what my clients want. And I was really hoping we had a chance with DVD-Audio and other surround formats, but the over-compressors are winning out there, now, too. And it's a bigger problem on that format, since you are now forcing people to change levels between movies and surround music, when both are calibrated identically.

I don't think you know what "compression" is...

[Joce640k]

When an sound engineer talks about "compression" he means compressing the dynamic range to make the music sound louder.

This is NOT the same thing as compressing sound to save disk space.

Re:A good video explanation

[philicorda]

"What's different now that the video shows is the peaks are not getting clipped anymore, instead they reach 0.0 db but the entire mix is digitally volume maximized so almost every single peak is that loud."

No, they are getting clipped. Have a closer look for flat topped peaks.
The damage is being done by look ahead limiters like Waves L2, which are the last process in the mastering chain.
These limiters work on psychoacoustic principles, employing some of the temporal masking ideas used in lossy audio compression to make the artifacts of very fast peak limiting as inaudible as possible.
It's known that humans cannot hear short periods of clipping distortion (less than 2ms or so), so these limiters allow that to happen, clamp down a millisecond later, and increase the subjective loudness of the signal without losing 'punch'. As this kind of limiter incorporates a delay line in the audio output path, but not the side chain, it's always looking a few milliseconds ahead and so knows how to react to a peak in advance.

The problem is that if you push a limiter of this kind really hard, it cannot keep it's artifacts inaudible, the clipped periods get longer, and the music starts to sound harsh and tiring.

It's a shame as they a beautifully clean limiters if used correctly, you can knock 4-6db of pop material without the kind of artifacts a traditional analog limiter would produce.