I'm not sure how many people who read this blog are interested in this, but I am, so....

For a month or so now there has been a great furor at eMusic.com, an online music store that sells mp3 files, about an acquisition of some major-label music that seems to be encoded at a slightly lower bit rate than has been standard at eMusic. Never mind for the moment what that really means--the important point is that it represents at least a potential decrease in audio quality. Some people say they hear a huge difference, others say that if there is a difference it's too subtle for them to notice. I'm in the latter camp. I did some searching online and found this interesting test. See if you can hear the difference. I got it right but was not 100% certain of my answer.

All the eMusic files in question, by the way, are encoded at higher rates than the 128k used for the lower quality file here, so any difference among them will be more subtle than what you're hearing on this test. Some kinds of music are more affected than others. I found that I have two copies of an album of the solo piano music of Phillip Glass, one encoded at 128k and the other at a variable but generally higher rate (the encoder uses a higher sample rate in more complex passages, a lower in others--the point is to optimize the tradeoff between sound quality and file size). I could pretty easily distinguish the 128k and VBR versions--the notes of the piano sounded noticeably more alive in the VBR one. 

I also did some testing with a CD, comparing it to mp3s of the same music encoded at different rates. I think I could distinguish the cd, though it was hard to be sure, because the mp3s were being played back through my computer and at first were noticeably quieter. Once I had matched the volume on the cd and mp3 setups, it was harder to tell the difference.

The only way to know for sure, of course, is with a completely blind test, in which the listener doesn't know which recording is which.  I thought about drafting my wife to do a blind test for me but decided I wasn't really that interested. I am very skeptical of the people who claim to hear a huge difference between, say, a 320k mp3 and a 192k VBR mpr, to the point of emitting cries of pain when exposed to the latter and declaring them unlistenable.  But then I don't have the most sensitive ears in the world, and my ability to hear high frequencies is far from what it used to be (a common effect of aging, especially in men).

Supporting my suspicion, here is an article about such a blind test. The "uncompressed WAV" they refer to is, for listening purposes, identical to the audio on the cd. Summary: people didn't do as well as they expected to.

And speaking of the ability to hear high frequencies, here is another interesting listening test, in which you can find out how far your high-frequency hearing extends.  I pretty much stopped at 12k. I had to turn the volume up a little from 10k to hear the 12k at all, which makes sense because it's around 10k that my sensitivity drops off, according to a hearing test I took more than ten years ago. I had my wife try it and she was still hearing something over 20k, which is pretty amazing. Note, though--if you're listening on your computer and/or with cheap speakers/headphones, the equipment may not even be capable of reproducing the very highest frequencies, almost certainly not at the same volume as lower ones.

By the way, I've mentioned before that I have tinnitis: the 12k sample on this test is pretty much what it sounds like inside my head all the time.