wrote in message
...
John Matheson wrote:

On a slightly unrelated matter, given that: 1. noise induced
hearing loss is a V-notch centred around 4kHz; 2. nearly all
conventionally designed speakers have a power response hole
in this region due to driver and crossover design;

I would instantly challenge you to support this assertion: that
"nearly all" conventionally designed speakers suffer form this
problem. Having seen a LOT of conventionally designed speakers,
the mere statistics of your claim are rather easy tp test. Have
you, in fact, done so, or is this simply assuming this to be
axiomatic without once challenging the assumption?


It is easy to test and I have tested it hundreds of times. But even if I
hadn't:

1. It is a consequence of the laws of physics (which all manufacturers
share - none get exemption) of conventional cone mid-range drivers &
high-order crossover design;
2. There is a gazillion published polar plots and off axis frequency
responses in the public domain that prove the assertion.


Having actually measured it at one point, I find that my copy
of an 18th century fenhc double haprsichord has some pretty
serious holes in the power response at a number of frequencies.

Why have the last 5 centuries of harpsichord makers not experienced
the joy of filling in these serious defects? What's wrong with them?


It doesn't need fixing - it's clearly part of the character of the sound of
the harpsichord. It does mean that the room the harpsichord is in has a huge
bearing on how the harpsichord sounds - but that should not be news to
anyone.

I'm not being in the least facetious: You basically assert that
an even power response is requisite to the proper enjoyment of
music.

I am not sure how you drew the conclusion that I asserted any such thing.

I counterassert that such a view is extremely narrow and
overly constrained to the point where it ignore some of the most
fundamental drawbacks at the level of first principles in terms
of recreating a realistic sound field of a musical event. No
doubt even power response is a great specification to crow about,
and one that, really, is not all that hard to achieve. But in doing
so, precisely WHAT problem have you fixed?

The problem that is fixed is better described by some of the references I've
made in other posts than I could do justice to here. Basically it's about
perceived sound quality. I think it is a sufficiently well scientifically
established to be accepted for good practice in speaker design.

Back to the original question: Why is a speaker with a perfect power
response, with a uniform, frequency-independent radiation pattern,
requisite to the production of a realistic sound field of an
instrument in the same venue?

I have not suggested such a speaker system. I suggest now that it needs a
flat(ish) on axis response and a declining power response (that is a
frequency-dependent radiation pattern). Why a declining power response?
Because nearly all real acoustic sources that we want to listen to have
declining power responses, where by power response I mean total radiated
energy versus frequency. I accept that doesn't mean the speaker is capable
of replicating the sound of any individual source in that space. To do that,
clearly it would have to have the SAME power response (and much more
importantly) the SAME directional response as the original source. Clearly
that is impractical for a hi-fi.

If you say the uniform directivity and power response is good, then
you are also asserting that the lack of such in a real instrument
is bad. If you are saying that the radiation pattern of the instrument
is good, then maybe what you are saying is that uniform power response
and radiation pattern is not necessarily bad, but maybe irrelevant.

If I said it - well I didn't! I had not until this post made any comments
about the power response of real sound sources. Of course that causes a
dilemma in recording real sources - where do you put the microphones. I
maintain that the best option for a speaker is a power response that is free
of major discontinuities and I think that should be fairly self evident from
your own essay. After all why would you want a speaker to overlay IT'S power
response irregularities on top of your recordings? The challenge is to find
a reasonable average power response that mimics sources such as, say
orchestra and voice. Funnily enough (or maybe not) the power response (not
directional response) requirement for these two disparate sources is not
dissimilar.