MP3 Compress This
posted by on August 14 at 13:25 PM
It’s bad enough that I have to keep my technophobic parents away from things like antivirus software installations, printer setups and CD burning. Looks like I have to keep them away from the Seattle PI as well.
An article in yesterday’s edition examines the world of compressed music, and it’s full of complaints about the sound of MP3s being inferior to “uncompressed” CDs. The basic premise: Those MP3s hold only 10% of the actual song! You ain’t hearin’ shit! Blow up your computer and shop at Borders! The author even goes so far as to get a doctor’s opinion:
“Poorer-fidelity music stimulates the brain in different ways,” says Dr. Robert Sweetow, head of the University of California-San Francisco audiology department. “With different neurons, perhaps lesser neurons, stimulated, there are fewer cortical neurons connected back to the limbic system, where the emotions are stored.”
It’s as close as Joel Selvin gets to research and analysis in his journey of misinformation. Sadly, the people most likely to be confused and scared off of MP3s by the article are the ones who hate computers already and read the paper edition. They won’t enjoy the article’s bustling, online comments thread which clarifies many of the article’s wrong points.
I didn’t need to scroll down 16 comments to remember how MP3 compression actually works. The concept isn’t even full of high-level scientific mumbo-jumbo. Bottom line, MP3s can be so small because they cut out what the human ear can’t hear. Strip out super-high and super-low frequencies, and our ears still receive the exact same information.
Would’ve been good to include that basic fact, let alone ask audiology expert Sweetow whether the brain actually processes those impossible-to-hear frequencies (though the word “impossible” leads me to a certain conclusion). It might’ve also been good to remind readers that CDs use compressed digital audio as well, perhaps scaring everybody back to the clear, analog world of vinyl. And it might’ve been great to follow this unsubstantiated claim—“For digital audio to substantially improve, several major technological hurdles will have to be cleared”—with the ways that it already has.
I’d continue with conjecture about this article being part of a conspiracy plot by the confused, idiotic RIAA to sell more CDs, but Mom just called in a panic. She says she can’t find the Start button anywhere on her keyboard. UGHGHGHQ#W$%GKGHGH.
It's been shown that the high frequencies beyond the range of the sense of hearing affect the frequencies being played in music that are within our range of hearing.
It's fine if people enjoy low-fidelity music...no skin off my nose, but there is ample evidence that humans can tell the difference between a CD and an mp3.
More relevantly, humans can hear the difference between a low-quality compressed file and a high-quality compressed file. Discerning the difference between high-quality compressed files and CD audio gets into audiophile territory.
It's better not to comment if you don't understand.
MP3 does more than throw away frequencies you can't hear. If that's all it did, the files would be much much larger. MP3 also combines frequencies that it thinks you can't tell the difference between. Is it right? Sometimes, sometimes not. It also throws out any differences it finds between the left and right channels. After all that, it then applies some math mumbo jumbo (DCT) which throws away even more information that you can't totally get back.
CD's are not compressed in any way. They are sampled. Sampling != compression. CD's are cleverly sampled to include everything (with 16bit precision) up to 22.05 KHz. To achieve this they sample at the Nyquist Frequency of 44.1 KHz.
The important thing to keep in mind is bit rate. As we've just shown CD's have a bit rate of 1.4112 Mb/s. Your basic MP3 has a rate of 128 Kb/s. Now, imagine if you will a section of music so complex that even throwing out everything you can't hear (allegedly) and combing frequencies that are not perceptible, you still require 160 Kb/s. MP3 simply starts throwing more away until it fits. Unless you are using variable bit rate encoding.
The other thing to remember is that once you throw out the information, it's gone for good, you can't get it back. First everyone used 128 Kb/s MP3. Then it was decided that 160 Kb/s was better. Until Apple said we should all have AACS and Microsoft said we should use WMA. And then Apple said perhaps 192 Kb/s is better. The problem is, that if you had your music stored in one format, you can't convert it to another without losing even more data!
The obvious solution is to keep all your music in a format like FLAC, which compresses, but loses no information. Of course no one sells in that format, so if you want to keep your digital files like that, you buy CD's and rip them yourself. You are buying your music aren't you?
Of course that's what you should be doing anyway, as the online stores are full of evil DRM.
If you still don't believe how much lossy compression can affect quality, try a little experiment. Take a really good photo with a camera in TIFF format (uncompressed). Now open it up with your favorite photo editor (I like GIMP) and convert it to JPG with 100% quality. Then do it with 75%, then 50% and so on. Now compare all the images. It's easy to see which one is best. Jpeg compression is very similar to Mpeg compression. If your eyes can see it, don't you think your ears can hear it?
Yes, Andrew, there's plenty more going on with MP3 compression than mere frequency shaving. And the word "sampling" is the more proper term in terms of CD audio, though it's still relevant to consider that CD audio isn't the pure, original sound wave, either.
Of course, badly encoded audio will sound worse than the original, but when you reach a high enough bitrate--192 for older folks, 256 for audiophiles--the differences are pretty much indistinguishable...certainly nothing like looking at a compressed image. If the article really wanted to be important or relevant, the Seattle PI could've at least blasted iTunes for having such a low default bitrate for the songs it sells.
I don't think that Andrew's photo compression analogy is exactly correct -- if only because our eyes are probably better than our ears. A better experiment that everyone can test this for themselves is to take a CD, encode your most beloved track at a few bitrates (96, 128, 192, 320) and try to tell the difference. My guess is that unless you're using amazing speakers and are extremely perceptive, anything above 128 will be indistinguishable from the CD.
If you don't want to do the encoding yourself, PCWorld has an inline test of 128 vs. 256:
http://blogs.pcworld.com/staffblog/archives/003993.html
http://en.wikipedia.org/wiki/Hypersonic_effect
Red Book audio (the kind on CDs) tops out at 20 kHz...the Hypersonic effect concerns frequencies above that. Arguing any further about that, of course, delves into the CD vs. vinyl debate, and that stuff burns my throat like bile.
lossless compression for the win.
Comments Closed
In order to combat spam, we are no longer accepting comments on this post (or any post more than 14 days old).