Monday, January 4, 2010

NPR Story on the Loudness Wars

NPR just ran a new story on the Loudness Wars in which engineer Bob Ludwig does a pretty good job of explaining the phenomenon in terms anyone can understand.  A good article to share with your non-musician friends.

7 comments:

Anonymous said...

Hi Tom,

I've noticed the volume is maxed out on Storm and Meta after a previous post on the same topic. Do you control this during recording or is this done my your mastering engineer, etc? Should we expect to see more dynamic recording from you at some point? Granted, I have not examined Compass yet.

Best wishes and Happy New Year
~w~

Will C. said...

Actually not much of an article, in large part because the psych prof has NO IDEA what he's talking about--it's not that he simplifies things, it's that he makes factually inaccurate statements.

"Now, the only way you can make square blocks look like a smooth curve is by using very, very small blocks so it ends up looking as if it's smooth. Now using lots and lots of blocks means lots of storage, so we end up using [fewer] bigger blocks. Which means we end up not representing that curve very smoothly at all."
Not true. You don't play the "blocks"--you smooth them via a lowpass filter. When you smooth the "larger" blocks you end up with lower frequencies--large curves mean large wavelengths mean low frequencies. When you encode data with a lower samplerate, you end up with a loss of HIGH frequencies, resulting in a SMOOTHER curve. High-frequency artifacts in mp3s are a result of entirely different compression processes.

"The difference between the smooth curve and the rough edges you end up with in the digital recording, you can think of as noise because that is perceived as noise," Oxenham says.
This is where you realize you're getting your lecture from a professor who's totally unqualified to speak in this area--as opposed to, say, an electrical engineering professor, who would see what nonsense this is. If you actually played the "rough edges" the sound would be unlistenable. But the curve can be easily smoothed via filtering. As for the "noise" he mentions, this is noise caused by quantization, not sampling, and its level is affected not at all by the samplerate (number of samples per second). It is controlled by the bit depth, the number of bits per sample, and for those bit depths used in commercial audio equipment (CDs use 16 bits to a sample), the quantization noise is quite inaudible.

Anonymous said...

Who finds it funny that a radio station who is probably over compressing it's own signal to begin with is doing a story on this subject?

Tom said...

@-w- Those albums are certainly compressed and limited, but nowhere near to the same degree as most other scene releases (look at a Combichrist song in a waveform editor some time!) and they're definitely WAY less squashed than most mainstream releases. It's a matter of balance, I think. A release needs to have some degree of loudness, but you don't want it to be fatiguing either.

@Will C - Yeah, there are some technical inaccuracies in the way he explains it, but I think it still gets the basic point across in a way that would make sense to a layperson who might not understand why the Loudness Wars are a bad thing.. Try using the words lowpass filter, quantization, and bit depth to a non-technical person and see how far you get.

Anonymous said...

Some of my CD's bounce off the scale. Nirvana, RATM, Infected Mushroom, Mathew Dear etc. It's true yours do not do that. I think the only albums that I have that don't appear limited or show clipping are Bauhaus and an early Clannad. A23 still rocks regardless.

~w~

Will C. said...

Tom: The problem I have is that in this way he makes it seem as if he's giving an understanding of the ideas when in fact he isn't at all. Several of the things he says are in effect the opposite of the truth. And what happens is that people read these things and believe that they're read up on the subject and never realize how little they know.

Tom said...

Indeed!