Quote from the video:
"Sadly they took this page of their website, because people were too stupid to understand what was happening... "
Here is
website in the web archive.
Such people certainly exist, but perhaps there are other reasons why this site is no longer online.
For example, because of statements like this:
"The accuracy score is perhaps the single most important tool for earphone design...
Consumer Reports reported that it was possible to predict listeners' loudspeaker ratings within 8% from a calculation based on one-third-octave frequency response measurements converted to loudness in sones. The average error in loudness from a perfect system, subtracted from 100%, gives the accuracy score. Etymotic Research extended this 21-band calculation to a 25-band calculation and routinely uses the 25-band accuracy score in all earphone designs."
So all of Etymotic's IE headphone designs are based on the Consumer Reports accuracy score. They have transferred this type of rating from speakers to headphones.
Let's see how reliably the accuracy score predicts the rating of speakers.
An obvious model for comparison purposes is that used by Consumers Union (CU) in the loudspeaker evaluations published in their magazine Consumer Reports over the past 30 years. It is based on 1/3-octave measurements of sound power that, after manipulation, yield an accuracy score out of 100, indicating how far the tested loudspeakers deviate from their notion of an ideal performance. Apparently, no formalized subjective evaluations are involved.
From Sound Reproduction, Toole
So, the only difference seems to be that Etymotic uses 25 frequency bands instead of 21 to evaluate headphone quality, I guess in terms of deviation from the specially created target curve by Etymotic (this is my guess, no details provided by Etymotic) - much like Harman, only without listening preference tests.
Now how was that with the accurate score for speakers:
Using the collection of anechoic measurements described in Figure 18.6, Olive (2004a, 2004b) undertook an evaluation of 13 loudspeakers that had recently been reviewed by CU. It began with a fully balanced, double-blind listening test (every loudspeaker was auditioned against all others the same number of times) by a group of selected and trained listeners.
From Sound Reproduction, Toole
I think
this could be a reason to change the website to no longer be associated with the accuracy score.
The buyer should buy a headphone that will sound good to them. If they have to EQ the headphone they bought to make it sound good to them, I consider it broken for them because they could get a headphone that measures how they want.
If the headphones really sound like crap, then I would agree with you.
In all other cases, we should look more in detail at how such a target curve comes about to which a headphone should be optimized.
Target curves on Etymotic Research graphs indicate 100% accuracy: The open ear diffuse-field response of the KEMAR® manikin modified to compensate for the high frequency boost added to high-quality recordings. This modification (approximately 5 dB at 10 kHz) is necessary to avoid earphones sounding too bright on commercial recordings.
Wow! That sounds like exact science and
@Sharur also presented us with Etymotic's target curve as a problem clearly solved by Etymotic twenty years ago.
What do the individual measurements of the HRTF look like? Do they have to be heavily averaged to get a target curve?
Individual HRTF for different angles, look like this:
If a diffuse field target curve was created from such individual HRTF, i.e. twenty years ago, then it should be clear that, due to the complexity, this very probably did not solve the problem of headphone target curves - the same applies to manikin with artificial average ears.
The differences in ear canal length and volume alone make it clear that a target curve will never provide optimal reproduction for all, only an approximation.
In contrast, individual HRTFs also do not appear to provide optimal results and may not be superior to artificial target curves, as this recent study "
A Perceptual Evaluation of Individual and Non-Individual HRTFs" shows, that comes to the following conclusions:
To understand the meaning of the statements, one must realize that the KU100 is "only" a diffuse field equalized recording head, without an upper body. All this suggests that twenty years ago Etymotic has not solved everything and of course probably Harman's target curve today neither.