Misleading? (Forum).

`Are Speaker Curves Misleading?' is quite interesting. I used to enjoy reading Julian Hirsch's reviews. The first several and last several paragraphs were the most interesting. Technical data in the middle was akin to Greek, and I think Hirsch ignored most of it in favor of his ears. I also got chuckles out of his last paragraph summaries. They were never negative (can't make the advertisers mad), generally positive in some manner (e.g., "if you...., then XYZ deserves a....). He was a master of vagueness and euphemisms at the end. But, we still love him.

Question: what formula did you use to develop your curves? On the Stereophile test disc number one, Harley and Atkinson used 1 kHz @ -20db as a reference point. Your points at 1 kHz vary from speaker to speaker. I assume the numbers on the X axis are dB variances from some reference point. Please explain what the point is, so I can replicate the testing on my system.

Your article prompted to test my Wilson MAXXs. Using 1 kHz as a reference point, I have found great variations. I also switched my front end outputs (l to r and r to l) to determine if the dCS CD system varied by channel. It didn't. Then I switched speakers (again, l to r and r to l) to determine if the room positions influenced the numbers. They didn't.

I will be interested to read your second article and your conclusions about the curves. Do the variations make an audible difference, or not?

 
Wally Marx 
via e-mail 

Howard Ferstler replies: I use uncorrelated pink noise generated from one of two sources: either the one-minute segment on the Delos Surround Spectacular test disc or else the somewhat longer track on the disc produced by the Boston Audio Society. The BAS disc is a CD-R item.

I generally try to get as much of the curve as possible to overlay the zero line (for reasons explained below), and do not aim to get a reference point at any particular frequency, such as 1 kHz. I want to get the curves to be comparable in terms of total output and not in terms of an arbitrary zero frequency. For example, if I had selected 1 kHz as the zero-line starting point with the EOSS curve (Issue 94, Figure 1) most of the plot would have been below that zero line, with only the bass peak extending much above it. That would have made for an odd looking curve that could not be properly compared to the other curves.