the double edged sword of loudspeaker ....
bfeng
rjs...@calcube.com wrote:
(snip)
> The term accuracy means exactness, correctness, or freedom from
> error. In the context of further discussion, this is the definition
> of accuracy I'm going to use as a *frame of reference*. (snip)
>
> According to Miriam-Webster (dictionary), the term loudspeaker means:
> "a device that changes electrical signals into sounds (snip)
> I'm buying into this definition for further discussion.
>
> So we've now established a *frame of reference* regarding the terms
> "loudspeaker" and "accuracy."
My interpretation of what you just said is that loudspeaker accuracy
is that degree to which a loudspeaker can:
changes electrical signals into acoustic pressure waves which
are representations of the electrical signal with "exactness ..
.. or freedom from error." Obviously, there is a translation from
one form of energy (electrical) to another (pressure in air), but
aside from that, isn't this valid concept from the frame of reference of
=> a accurate transducer <=
If you measure the sound pressure that a ldspkr produces (as some
defined point in a defined enviroment) and if that pressure signal
very closely matches the input electrical signal in shape (and I'm
talking about the raw time signal) then the ldspkr can be said to
have some degree or level of accuracy as a electrical to air pressure
transducer. Yes?
> Why is it that two words with concrete definitions
> lose all significance when combined?
As I said above, I don't think they do. You've defined terms
which have, in my opinion, perfectly clear meaning when used together
-- from several frames of reference (one of which I stated above).
> (snip) THERE IS NO SUCH THING AS
> LOUDSPEAKER ACCURACY! That's right, there's no such thing as
> loudspeaker accuracy because the *frame of reference* for loudspeaker
> accuracy is a dynamic one, not a static one.
But, for a particular frame of reference, the concept of ldspkr
accuracy IS valid. Right? If one allows the frame of reference, for
any application of the word accuracy, to vary then all uses of the
word "accuracy" can be said to be meaningless.
Even if we are talking about a single listener. The perception of
accuracy will vary from time to time and day to day. Does that mean
there is no valid concept of accuracy? I believe not. Accuracy is
still a valid concept in a statistical, or 'average' sense. Take me
for example. I feel my current ldspkrs are less accurate than a pair
of B&W 801 Nautilus ldspkers. Somedays I may feel the difference is
less than on others, but generally I feel there's a pretty big gap
in accuracy. It doesn't have to be a precise, deterministic,
invariable value to be real and meaningful.
> Loudspeaker accuracy is
> a chameleon that changes colors between measurements and realism.
Accuracy wrt to subjective realism is a different thing than
accuracy wrt to objective signal reproduction. Yes?
If you pick one frame of reference, then the term has utility and
meaning.
> Measurements being the objective color while realism the subjective
> color.
Sure, and subjective accuracy is a different beast from objective
accuracy. Both are valid with respect to the accuracy of a ldspker.
> (snip) Let's take a closer look.
> Measurements. As a loudspeaker designer, when I think of
> measurements I think of objectivity. I think of measurements as an
> objective frame of reference to consistently improve upon existing
> iterations leading one to theoretical perfection. (snip) Measurements
> are simply experiments performed in controlled environments. So why
> are loudspeaker measurements performed in controlled environments
> rather than the 'real world' environments that we listen to them in?
Why can't you perform your measurements in a real world environment
( e.g. a listening room of your chosen design).
> Let's go back to the original definition of loudspeaker: "a device
> that changes electrical signals into sounds loud enough to be heard
> at a distance." The operative phrase in this definition is "changes
> electrical signals into sounds." Loudspeaker accuracy in the context
> of measurements means designing "a device" that does not alter the
> characteristics of the "electrical signals." Is this possible? Not
> yet. Can we get close? Maybe, but it must be realized that every
> piece of material used to build such "a device" deviates from the
> theoretical perfection initially calculated, thereby never realizing
> real world accuracy (defined as freedom from error).
The fact that every piece of material used to build a ldspkr deviates
from any current notion of theoretical perfection only means that
all ldspkers cannot exactly replicate the input electrical signal
(as a snd pressure wave). It doesn't mean one cannot compare the
input and output and judge how they deviate from one another.
> It is this deviation from perfection that has allowed so many interpretations of
> loudspeaker design to proliferate. I.E., every loudspeaker designer
> thinks they have *the answer*.
That and their individual built-in, subjective impressions of accuracy.
> (snip) there is no perfect "loudspeaker accuracy" as it pertains to
> measurements, only close approximations.
The fact that no ldspker is 100% accurate doesn't mean that the
concept of accuracy does not apply -- unless you believe that
accuracy is a purely binary concept (e.g. one that can only have 2
values).
> Realism. (snip) No matter what
> the venue, we demand that our loudspeakers transform our listening
> rooms into the real event, albeit with a little imagination (close
> your eyes for reality). This imagined participation with the
> *original event* is the other definition of accuracy. Realism. The
> difference between this accuracy and the accuracy derived from
> measurements, is that accuracy in the form of realism takes on many
> different personalities that conform to the eye/ear of the beholder.
> Realism can be timbre, ambient expanse, presence, localization, etc.
> Is it possible for any loudspeaker system to conformally approach
> everyone's definition of realism? No, but some seem to come closer
> than others.
So long as individuals have individual preference then there will
never be one ldspker that is the most accurate (wrt "realism") to all
people. Since we're not all clones from a single individual, with
identical life experiences and values, this will never come to pass.
So, what's the point?
> Is it possible for multiples of loudspeakers (more than
> 2 channel stereo) to recreate realism? No cigar yet, but getting
> better.
Yes, and this progress seems to be made despite the screaming
protestations of most audiophiles (who claim stereo is the only way).
> Again, the digital age may yield breakthroughs towards an end
> of electronic tranparency, but the fact remains that perfection is
> elusive in the real world.
Perfection is unattainable in the real world. Even when we can
engineer materials one proton and one electron at a time, there will
be uncontrolled variability in the end result. A lack of perfection
does not mean improvement in accuracy is an irrelevant concept.
> So where are we now? If there is no perfection, and the term
> loudspeaker accuracy can be loosely translated to fit multiple
> definitions,
And if you pick which ever one is suitable for the purpose, objective
or "realism," then you have a definition that has meaning, purpose,
and utility. Yes?
> where do we draw the line for a *frame of reference* of
> good or bad?
If it's objective accuracy, then the frame of reference is the input
signal. Simple.
If it's "realism" accuracy, then the frame of reference is one or a
group of individual's perception.
> Do these multiple definitions render magazine reviews
> worthless or priceless?
Not at all. Magazine reviewers make attempts to judge objective and
subjective accuracy. Although they do the former with dismal results,
if one's own preferences correspond to that of a reviewer, then the
reviewers opinions will predict one's own opinions to some degree of
accuracy.
> How as industry professionals or consumers
> can we delineate what is the correct *frame of reference* for product
> judgement?
There is, perhaps, no correct or perfect frame of reference when we
consider the end user -- a human consumer. As you stated previously,
subjective preference is a variable quantity.
However, in a practical sense, the ldspkr designer can assume a
particular target market (with some understanding of that market's
priorities and preferences), and use a measure of accuracy based on
that frame of reference. If you want to do a ldspkr with very
spacious, open sound stage, and sharp imaging (for me this is ProAc)
then you may choose to discount low frequency extension (accuracy).
Where's the problem with this approach? You decide what's important
to your target market, and you try to design your product to best
appeal to that market.
> Are realism and meeasurements mutually exclusive entities,
> or are they tightly interwoven?
Why do you try to paint the world in such black and white tones?
They can be interwoven to a degree that depends on the specifics
of the task at hand.
>
> 'Realism' and 'measurements' have often spawned debate within the
> audio community as to the possible notion that both theories
> represent separate camps of ideology.
And this leads to a lot of acrimonious debate, unfortunately.
> (snip)
> Equilibrium defined as that point in which realism and measurements
> carry equal weight (in the mind of an individual) in the
> determination of loudspeaker accuracy.
Interesting definition. Why would an individual not just use the
point where realism carries all the weight? What's the point of
objective or measured accuracy if it does not add to realism (from
the point of the end user)?
> On one side of the scale we
> have 'measurements' (conservative, objective?),
Why are objective measurements conservative?
> and on the other side we have 'realism' (liberal, subjectivist?)
Why is 'realism' liberal?
(snip)
> How viable are magazine reviews in determining loudspeaker accuracy?
Interesting question indeed.
> Well, most magazine reviews do present an objective and subjective
> profile of a loudspeaker system and they employ competent individuals
> to effectuate the review. Competent in the respect that the reviewers
> have been exposed to a much greater cross section of products than
> the average individual, so they do have a pretty good feel for a good
> or bad product.
> And for those reviewers that perform measurements,
> they're knowledgable about their measuring equipment and the
> interpretation of their results.
Although:
they may not be knowledgable about what measurement most closely
correspond to subjective impression, and they often do not exhibit a
comprehensive understanding of the powers and limitations of the
analysis methods they employ.
> So is a magazine review the last
> word in loudspeaker accuracy? Nope.
I know few people who would disagree with this!
> (snip) So if we can't trust measurements or realism, what do
> we trust? Our OWN ears. After the opinion and/or after the math, we
> (as designers or consumers) ultimately choose what we like or deem to
> be 'right.'
Given that final arbiter of the product judges it's performance
subjectively, your statement certainly makes sense to this consumer.
Of course, that the ultimate measurement is a subjective perception
does not mean we can't put a certain degree of faith in
measurements.
Here's a trivial example. If a ldspker exhibited a strongly rising,
measured high frequency response, in a real listening room, then one
might think it prudent to heed this measurement. If the degree of
this rise were sufficient, I would think it perfectly acceptable for
a designer to skip the listening and accept the measurement for what
it says (too hot on top).
Personally, I would think nothing wrong of designing a spkr to
exhibit a "good" measured impulse response -- at least as a starting
point. Yes or no?
John Feng
Thomas Nulla
<snip of very good points>
Another fine, thoughtful post.
>Personally, I would think nothing wrong of designing a spkr to
>exhibit a "good" measured impulse response -- at least as a starting
>point. Yes or no?
A starting point...yes, indeed!
Here you touch on a very good reason to design a speaker that does as
good as possible a job of reproducing the input signal: such a
speaker will be more amenable to working in a predictable fashion
with external signal-processing equipment (equalizers, DSP room
correction, etc.)
Such equipment, user-adjustable and reconfigurable (even as
recordings themselves vary) is going to be much more versatile than
choosing a speaker with a designed-in set of colorations...which may
match a particular room, or set of components, or recording style in
a satisfying manner, but lacks versatility.
I'm leery of getting a system where some components must have
significant colorations in order to offset the colorations of other
components; this is a house of cards where any change will likely be
for the worse. And what happens if you move to a new listening
room...?
>John Feng
Thomas <now playing: Ozric Tentacles, "Jurassic Shift">
http://www.io.com/~nulla (high fidelity and miscellany)
Ozric Tentacles page, Dunlavy Audio r.a.* archive to 6 December 1998
"The Universe is the Practical Joke of the General at the Expense of
the Particular", quoth Frater Perdurabo, and laughed.