High Efficiency Speaker Asylum: I read it again, and.... by Jon Risch

Most of my concerns are still not addressed, nor did you see fit to answer any of them here.

I did find the answer to ONE of my questions, namely, the number of listening test subjects. However, I found that information in a place, and ONLY that place, that I did not expect to have to look: the abstract.

I find this paper to be a very poorly written one, as there is critical information that is contained ONLY in the abstract, namely, the number of listening subjects (27). This is very poor practice, and hardly ever seen in formal acedemic papers, the abstract is not supposed to be the sole source of critical information, but a summary of the paper's contents.
This answers one of my questions, and still does not address all the other issue's I have at hand.

Regarding the rest of my points, after re-reading the entire paper, abstract and appendix, I still find all of my questions and concerns valid and in need of response.

In addition, I find that one of the claims made in the engineering report, namely that the opinion of Alexander Voishvillo in his paper "Nonlinearity in Compression Drivers - Where the Distortion Comes From?", published in the JAES Nov. 2002, supports your conclusion that: "non linear distortion in a compression driver is simply not a factor in it's sound quality" as being a very big stretch of the facts. Voishvillo said that: "the propagation distortion from the horn's mouth to the microphone was quite comparable with the distortion generated in the horn driver."

I take this to mean that the distortion of the horn and driver were showing up in his models (and measurements) at approx. equal levels, but this does not mean the same thing as the "the horn's distortion dominates", which is what you are saying.

I also saw another procedural flaw, in that the distortion of the listening test sound source, namely Etymotic ER4B insert type earphones, was not measured or checked to be sure that it was below that of the distortion levels in the compression drivers. The authors merely "listened to them" to determine if they were suitable. I find this absolutely amazing from someone who constantly discounts his own subjective impressions, making a strong public statement about the discounting of his own personal subjective impressions, and then turns around and uses those very same subjective impressions to pass judgement on a major component in the listening test chain, without ANY objective back-up or verification. Isn't this a bit hypocritical, to take one stance publically, and yet ignore it when it is convenient for the purposes of a listening test experiment?

I also note that no information concerning the playback levels was provided, nor any of the other related listening test operational details.

It seems that you would have us all take your listening test results on faith, rather than supply the basic information for others to judge your efforts, or even to attempt to replicate them.

I remind you that it is common practice for these kinds of test results to be replicated by others before they are taken as some sort of solid evidence, much less when they violate standard statistical procedure (by assuming a null result equals a solid negative result), and do not provide concrete controls for what CAN be detected.

As it stands now, no one who reads your engineering report could even begin to replicate the listening tests, not enough information is given in the report to do so. This is sloppy, to say te least.

If you do not wish to respond to any of the points I have raised here, so be it. But do not expect the folks who frequent these boards to be bowled over by your dazzling brilliance, when you refuse to reply to the most basic and straightforward concerns and comments.

The points I raised previously that were not clarified or responded to:

who the listening subjects were (background, hearing accuity, etc.),

what amount of training they had,

how the listening test data was analyzed (total subjects averaged, or individuals analyzed for independent positive results, etc.),

why the extremely high power levels of 100W (at full power) were used, rather than more reasonable levels,

why the power level steps were so small, only about 3 and 6 dB down from full power, leading to the very high probability that there would be no sonic differences between the sound of the drivers at the three 'different' power levels, since they were not that dissimilar, and they were all into large amounts of distortion,

why were no real controls provided for the listening tests in the form of a determination of exactly what the listening test setup and subjects COULD detect and hear reliably,

why only one 15 second segment of a pop rock song was used, and no other music at all,

why were the listening subjects exposed to so much sheer material, possibly leading to listeing fatigue,

why was the signal run through a Turtle Beach Santa Cruz sound card three times (out and in and out again) rather than a higher quality sound card or audio playback and recording device, one widely respected for it's sonic transparency,

why was a 5000W 'subwoofer' amp used to drive the compression driver tweeters, instead of an amp more suited for the determination of subtle listening issues,

and why was what is comnmonly refered to as a null result (a simple failure to achieve a statistically significant positive) treated AND stated explicitly as if it were in fact a negagtive result, i.e., that the distortion levels of a compression driver do not matter, when this is patently and demonstrably WRONG to do so.

Jon Risch