Though not truly "accurate", the best way that I've found was to make my own binaural, and non binaural recording as a point of comparison.
Binaural on headphones represents *more* accurate, then sample the non-binaural on loudspeakers.
It's still utterly subjective however, but I live in the real world and don't require absolute precision.
Earl I'm not sure why you dwell on "expert opinion" in this. I shun experts myself (much to my wife's annoyance) but that's beside the point. In the matter of audio I can't remember ever buying anything significant based solely on anyone's opinion and certainly never set out to convince myself that it was an objective quest to have the "best". I guess it just isn't that important to me to make it an almost moral issue out of it.
I remember Ayn Rand feeling it important enough to spend one of her speaking engagements proving objectively that all folk and primitive art was "trash" (probably evil too).
You have to admit you certainly sound very convinced you are right
Yes . . . a flat DI does (tend to) sound very bright, and air absorption may be part of the reason (I've noted elsewhere that the absorption curves in McCarthy's "Sound Systems: Design and Optimization" look a lot like the "high frequency shelf" that Linkwitz found necessary with the dual-tweeter ORION). But there's got to be more to it than just that, because it doesn't seem to apply (as much, anyway) to flat-on-axis in the near field. And it does seem to depend (a lot) on the recording . . .
And then there's below 1000 Hz . . . where the transition from a "controlled beam" to essentially omni can be quite abrupt, right in the range of most instrument and vocal fundamentals.
But there we are at it again . . . if a "flat" DI sounds wrong how do we jigger the "objective" standard to reflect the "subjective" reality?
Yes, this is what I was getting at. I said behind the speaker to make the point of total bandwidth power response. SL's theory is to raise the high frequency power response as much as possible. this was the reason for the rear tweeter on the Orion and the concept behind the Pluto.
He has stated several times that box speakers are omni at low frequencies and front firing at high frequencies.
I was wondering if you beleived as he does if this is detrimental to reproduce a convincing AS.
By using the old B&K falling response. It's been posted a few times and is easy to find around the web. I use it, I think Geddes uses something close. It does seem to work well in most domestic rooms.
I'm not sure why it sounds right, unless the mastering engineers are using it, too.
I think this is simply the result of the average mixing/mastering studio acoustics and the speakers they use.
Falling high freq response has been talked about ...
From one of Thorsten's Ultracurve EQ setup articles (2005):
GoodSoundClub - Romy the Cat's Site - Suggested target curves and setup techniques for Pro Audio Digital Equalisers....>
I would recommend the following EQ Applications for "pleasant" sound with most modern (post mid 1960's) recordings or re-masters of older recordings:
1) Boost the range below 125Hz uniformly by 2 - 4db (depending upon taste - I use 2db) and take the sliders above 125Hz so the form a falling slope back to 0db, with 0.5db (1 step) per slider or 1.5db per octave. I personally also have 20Hz at -1db compared to 125Hz and 31.5Hz at -0.5db. This is simply to slightly limit the LF Boost.
2) Apply a similar slope (0.5db per slider / 1.5db per Octave) from 2KHz upwards, meaning -0.5db @ 2.5KHz and then on to -5db @ 20KHz. You may experiment with increasing the point where the roll off begins somewhat.
3) Put a 4db notch into the response around 2.5 - 3KHz, returning to flat at 1Khz and 6.3KHz. This is the classic "BBC Dip".
The resulting curve is what I use on a daily basis and was arrived at based on the study of the various literature and a noting down of the most often applied EQ settings. It offers a good compromise between neutrality, sweetness and pleasant sound. But feel free to vary this basic recipe to taste and experiment.
(Not sure about the 'dip' ... but the bass and high roll off ...)
Actually he talked about various high freq rolloffs from 2001 when he got his Ultracurve EQ.
http://www.enjoythemusic.com/magazine/equipment/0101/behringer8024.htm
I think he connected it to some JBL or pro documents that talked about appropriate high freq rolloff...
From one of Thorsten's Ultracurve EQ setup articles (2005):
GoodSoundClub - Romy the Cat's Site - Suggested target curves and setup techniques for Pro Audio Digital Equalisers....>
I would recommend the following EQ Applications for "pleasant" sound with most modern (post mid 1960's) recordings or re-masters of older recordings:
1) Boost the range below 125Hz uniformly by 2 - 4db (depending upon taste - I use 2db) and take the sliders above 125Hz so the form a falling slope back to 0db, with 0.5db (1 step) per slider or 1.5db per octave. I personally also have 20Hz at -1db compared to 125Hz and 31.5Hz at -0.5db. This is simply to slightly limit the LF Boost.
2) Apply a similar slope (0.5db per slider / 1.5db per Octave) from 2KHz upwards, meaning -0.5db @ 2.5KHz and then on to -5db @ 20KHz. You may experiment with increasing the point where the roll off begins somewhat.
3) Put a 4db notch into the response around 2.5 - 3KHz, returning to flat at 1Khz and 6.3KHz. This is the classic "BBC Dip".
The resulting curve is what I use on a daily basis and was arrived at based on the study of the various literature and a noting down of the most often applied EQ settings. It offers a good compromise between neutrality, sweetness and pleasant sound. But feel free to vary this basic recipe to taste and experiment.
(Not sure about the 'dip' ... but the bass and high roll off ...)
Actually he talked about various high freq rolloffs from 2001 when he got his Ultracurve EQ.
http://www.enjoythemusic.com/magazine/equipment/0101/behringer8024.htm
I think he connected it to some JBL or pro documents that talked about appropriate high freq rolloff...
I don't, because I've seen (heard) it with flat mics and direct (unmixed, unequalized) capture.
Which is not to say that a lot of bad stuff doesn't happen "in the mix" as well . . . where that applies. But a "straight wire with gain" doesn't get the sound you want in your listening room (with constant DI loudspeakers) . . .
Eh, your speakers are not CD either. Actually your speakers have more uneven directivity in the midrange than an average 'conventional' speaker has. All objective.
If there's no theoretical justification then the B&K curve exists because someone, somewhere, decided it "sounds" right. And you agree. But there we go again . . . with "objective" measurements referenced to a "subjective" standard.
My own experience suggests that there are a number of cooperating factors, including (matching system response to) air loss, but also including other effects of microphone placement (relative to the orchestra layout) and some particular artifacts of the way sound is reflected in small rooms . . . all things which rather generally tend to boost the high frequencies as delivered to our ears. And of course recordings are often deliberately mixed to the bright side because it is assumed that's what the (average) customer wants.
To your first point if your being that restrictive in your definition then CD speakers do not exist - there are none. Mine are CD where it matters most.
Your second point is incorrect as the data on my web site shows.
B&K response - see Figure 5: http://www.bksv.com/doc/17-197.pdf
As an ex recording/mixing engineer for 10 years, this was/still commonly used in control rooms where at least one set of monitors was tuned/eq'd to this response.
For example, mastering engineer, Bob Katz, uses a fine tuned version of the B&K curve (no pun intended), where the response is flat to 1 kHz, and using 1 kHz as the hinge point, plot a straight line to -6 dB at 20 kHz.
I use this on my old school 3-way box with a 15" woof and mid/high horn/compression drivers. The measured 1/6 oct response looks like this at the listening position some 9 1/2 feet away:
I find the tonal balance to be just about right for any program material. Assuming that the rooms decay time has been sorted with bass traps, absorbers, and diffusers as required.
In my experiments with AS, I find that the closer together the speakers match in FR, across the audible range at the listening position, the better the AS. This assumes that the speaker setup follows the same guidelines as how it was mixed (i.e. equilateral triangle per attached guidelines).
I also found that response to 20 kHz, albeit -6 db down also made a difference in AS. i.e. it took me 3 different HF compression drivers to find one that actually could do this without being -15 dB down at 20 kHz at the listening position.
Anyway, lots of fun
Cheers, Mitch
Earl I'm not sure why you dwell on "expert opinion" in this. I shun experts myself (much to my wife's annoyance) but that's beside the point. In the matter of audio I can't remember ever buying anything significant based solely on anyone's opinion and certainly never set out to convince myself that it was an objective quest to have the "best". I guess it just isn't that important to me to make it an almost moral issue out of it.
You have to admit you certainly sound very convinced you are right
Yes that is certainly true!
I do not mean to dwell on "expert opinion" as much as the point that objective measures are superior.
That's commercially biased.
When was the last time you created a design that maintained constant directivity down as low as 80 Hz, and then sufficiently tested to see if it matters or not?
I do not see the widening at LF as an issue as long as it is not narrowing above say 500 Hz as any direct radiating speaker does. CD is the goal, but CD to ever lower frequencies is basically impossible unless it is fairly wide directivity. Below some frequency the need for CD goes away and clearly in the modal region it is meaningless.
In your subjective opinion perhaps.
Commercial agenda shines through on everything and not so hidden.
That's a familiar looking curve
I find the tonal balance to be just about right for any program material. Assuming that the rooms decay time has been sorted with bass traps, absorbers, and diffusers as required.
. . . what it doesn't explain is what is happening in our recording/reproduction chain that makes it necessary to roll off speakers like that to make them "sound right".It does seem . . . peculiar . . . that the asserted unimportance of "constant directivity" between 100 and 1000 Hz does seem associated with the (in)ability of the product to accomplish it. This especially so since "baffle step" is something that all of us are able easily to hear and "objectively" measure, and it is commonly "corrected" for (really that should be in double or triple quotes since where it exists it cannot actually be "corrected") . . .