|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
In Reply to: Measurements/Subjective Reviews posted by Jim Austin on May 24, 2004 at 07:49:08:
Jim,I thank that you raised this topic, because the objective reviews in the Stereophile are becoming a source of angst for me after a couple of incidents that have occurred in recently.
Firstly, for a magazine for whom most it’s reviewers regard turntables as sonically superior to most digital equipment, it is no small irony that this category of equipment is not subject to any form of objective measurements. All claims to its sonic superiority are arrived at based solely on objective listening. This policy is justified on the basis of costs and time constraints and that is very unfortunate.
More importantly, recently I was at the Speaker asylum following a thread where the Dynaudio Contour came up for discussion and some of the replies made some comments based on the speaker’s measurements in Stereophile. Doug Schneider must have felt duty bound to post the actual anechoic measurements of the Contour because of the margin of difference between Stereophile’s pseudo-anechoic results and the actual anechoic measurements. The actual measurements gave a totally different picture of the performance of the speakers in question. Now if measurements are not reasonably correct then they are of no value at all.
Finally, what is the value of published 1/3-octave analyser measurement for digital systems, I fail to see why JA persists in using this method when the industry relies on FFT analyser measurements. For me, I think this was probably brought to head recently when Keith Howard published a series of measurements in Stereophile using an FFT analyser, what can be gained from the departing from the industry standard especially when the alternate method is not peer reviewed nor any credible industry support that I am aware of. I note that not a single AES member critic of SACD/DSD has yet even referred to Stereophile’s 1/3-octave analyser measurements and I suspect it is because they are to put it bluntly of no value whatever, because the results produced simply do not stand up to public scrutiny.
Measurements are a source of objective reference, a standard by which all products can be compared with reference to any bias. Therefore if they are wrong as in the case of the Dynaudio Contour or as in SACD/DSD at odds with the industry standard with no valid justification, what value do they really provide? I rarely see a restatement of measurements, so does that suggest that they are always right?
In the light of the comments above, a response to your questions
<< Do you read the measurements section? >>
Yes, sometimes with much more attention than the subjective review.
<< If the measurements section is critical, and the subjective review is highly complementary, how do you respond? Do you dismiss the measurements? Dismiss the subjective review? Weigh them both? Dismiss them both? >>
I look for a second opinion from one of the European magazines that also does measurements, if the measurements to do not agree, then it is dismissed. That said all 1/3-octave analyser measurements are dismissed outright because of their dubious value, to the best of my knowledge no other magazine uses a 1/3-octave analyser, therefore it cannot be verified, so it fails on the first count.
Follow Ups:
> The actual measurements gave a totally different picture of the
> performance of the speakers in question. Now if measurements are
> not reasonably correct then they are of no value at all.
Agreed. In the past when I have compared my quasi-anechoic
measurements with those taken at the NRC, there has been good
agreement, other than the "2pi" bosts in the bass in mine, which I do
explain in the reviews. I will check the NRC Contour graphs.
> what is the value of published 1/3-octave analyser measurement for
> digital systems, I fail to see why JA persists in using this method
> when the industry relies on FFT analyser measurements.
As I have explained both in the mqgzine and on the Asylum, I use this
technique to allow comparisons of current products with earlier ones.
For the purpopses of the tests for which I perform 1/3-octave
analysis, the resultant graph serves its purpose, I feel. And please
note that I also use FFT-derived spectra.
> all 1/3-octave analyser measurements are dismissed outright because
> of their dubious value...
With respect, this is way too simplistic a view. For example, if you
look at my use of 1/3-octave in-room speaker spectra, these give an
excellent correlation between what is measured and the loudspeaker's
perceived tonal balance.
1/3-octave spectral measurements have a role to play, I feel, as long
as it is not assumed that they tell the whole stroy about a component.
John Atkinson
Editor, Stereophile
> In the past when I have compared my quasi-anechoic measurements
> with those taken at the NRC, there has been good agreement, other
> than the "2pi" bosts in the bass in mine, which I do
> explain in the reviews. I will check the NRC Contour graphs.
Okay, I searched the Loudspeaker Forum and couldn't find any postings
comparing NRC and Stereophile measurements of the Dynaudio Contour.
However, I did find a thread at http://db.audioasylum.com/cgi/t.mpl?f=speakers&m=161686
regarding the Dynaudio Confidence C4 that exactly illustrates
my point about measurements of a speaker's bass. A speaker that
measures as having a flat LF response in an anechoic chamber, like
the C4, will indeed have a tilted-up bass in a room. My nearfield
assessment of a speaker's low frequencies is, I believe, closer to
being representative of how a speaker will osund in a typically sized
room.
In that sense, despite them being different in the bass, both the NRC
measurement of the C4 at "http://www.soundstagemagazine.com/measurements/dynaudio_confidence_c4"> http://www.soundstagemagazine.com/measurements/dynaudio_confidence_c4 and my measurement at http://www.stereophile.com//loudspeakerreviews/794/index5.html
are both correct. They show the same speaker in different
acoustic environments.
John Atkinson
Editor, Stereophile
Sorry about the poor HTML coding. The reference to the Soundstage
measurements to the Confidence C4 measurements should have read
In that sense, despite them being different in the bass, both the NRC
measurement of the C4 at http://www.soundstagemagazine.com/measurements/dynaudio_confidence_c4
John Atkinson
Editor, Stereophile
This post is made possible by the generous support of people like you and our sponsors: