FYI --
I've started an audio blog on the science of sound recording and reproduction. The first 3 articles are summaries of an AES paper I wrote in 2003, "Differences in Performance and Preference of Trained Versus Untrained Listeners in Loudspeaker Tests: A Case Study." You can download the original AES paper for free at the bottom of the article, courtesy of Harman International. Enjoy!

Cheers
Sean Olive

Sean Olive? My goodness, they'll let anyone get an account here

So the results of trained listeners of loudspeakers can safely be extrapolated to the untrained population, eh?

Does the same thing hold for trained listeners of audio codecs? Seems like we assume that the results from the small group who participate in the public internet listening tests should somehow represent a more general preference.

Good question, for which I have no answer. CODEC testing is not my area of research. From what I've read, without listener training, most of the listeners will not reliably detect differences among high quality CODECS. The general approach seems to be that if the experts are satisfied then the naive listeners will be really satisfied.

So the real issue is whether trained and untrained listeners respond differently when dealing with lower bit-rate quality CODECS that produce clearly audible artifacts. I suspect that trained listeners will be more fussy, and less accepting (reflected by lower scores) of the CODEC artifacts compared to untrained listeners. That is generally true with loudspeakers.

I'm surprised someone hasn't already published some research in this area.

Why don't you search this forum about ABX (double blind test) applied to CODEC testing. That will give you some interesting subjects to talk about on your blog.

No, actually, I can't. It's $20 for non-AES members, and $5 for members. Or is there some rather less intuitive way of doing it..?

Try this link: http://www.aes.org/e-lib/download.cfm?ID=1...amp;name=harman (271 KB PDF)