Hi Dave
This is likeley true, and I would never advocate for a "flat" power response - it should fall somewhat at HF because thats what acoustics does - but I think that "smooth" is the key. a smooth and nearly flat direct response with a smooth and slight flalling power response - thats what I am after. I think that this is in agreement with Dr. Toole. We disagree on the "very early reflections" - I don't want them and he seems to like them - beyond that we are on the same page.
Sounds like good advice. Like a cookbook that says, "... a pinch of this and a little that and then add salt to taste....."
What does "thats what acoustics does" mean? Something about equal loudness contours?
Speakerdave - thanks much for writing such a clear and full explanation. I always like Lipschitz's stuff and not just because he is (was?) part of our local scene, eh, or an advocate of ESLs. Mostly Professor Stan is a jolly debunker of fads, I like that, but folks with tube amps or many golden-ears wouldn't like his studies. Or people who think horrible looking dips in freq response curves matter. He's was a great advocate of A-B and blind testing... not a lesson all have learned.
I join the quest for proper physical measurements. But you can't pretend your eye is your ear even in color and 3D.
Last edited:
It means that, as they travel through the environment, higher frequencies (direct and reflected) are attenuated more than lower frequencies.
Sheldon
As used in the original paragraph, if the "acoustics" is cutting high tones, shouldn't the pure/non-room measured power response be higher to correct for that or not influenced? (The El Greco astigmatism fallacy?)
Of course, it is possible your meaning is not the same as the way Earl meant it?
I've been doing this for a few years. Done well it's great but it's a lossy method, where a waveguide solution keeps the efficiency high. A foam collar to enhance a waveguide made a big difference in cutoff frequencies, 2dB of ripple reduction right below cutoff or thereabouts is nothing to sneeze at, not to mention directionality maintained lower as well (though lossy).
An externally hosted image should be here but it was not working when we last tested it.
Hello SpeakerDave,
Thank you for responding. I've let my AES membership lapse, but when I renew it I will read the sources you cited.
I think the early paradigm battle of direct response vs power response was conceptually a little bit off the mark. In my opinion we should look at direct response vs reverberant energy, and (again imo) instead of either/or, we should get them both right. What consitutes "right" for the reverberant field evidently hasn't been firmly established, so until then this is my target (with the presumption that the direct sound is "right"):
The spectral balance of the reverberant energy in the listening room should have approximately the same relationship to the direct sound as the spectral balance of the reverberant energy generated by live voices and instruments have to their direct sounds. While we may not be able to get an exact duplication of the characteristics of any one voice or instruement, smoothness of the reverberant energy, with a little top-end roll-off, serves this goal.
I'm not familiar with the Lipschitz & Vanderkooy paper, so I'll have to read it before I can comment on it. Thank you for bringing it to my attention.
Did you work for JBL at one time?
Duke
Thank you for responding. I've let my AES membership lapse, but when I renew it I will read the sources you cited.
I think the early paradigm battle of direct response vs power response was conceptually a little bit off the mark. In my opinion we should look at direct response vs reverberant energy, and (again imo) instead of either/or, we should get them both right. What consitutes "right" for the reverberant field evidently hasn't been firmly established, so until then this is my target (with the presumption that the direct sound is "right"):
The spectral balance of the reverberant energy in the listening room should have approximately the same relationship to the direct sound as the spectral balance of the reverberant energy generated by live voices and instruments have to their direct sounds. While we may not be able to get an exact duplication of the characteristics of any one voice or instruement, smoothness of the reverberant energy, with a little top-end roll-off, serves this goal.
I'm not familiar with the Lipschitz & Vanderkooy paper, so I'll have to read it before I can comment on it. Thank you for bringing it to my attention.
Did you work for JBL at one time?
Duke
Last edited:
[Think 4430....]
That's what I was thinking.
I don't normally do smilies, but one comes to mind...
This is, of course, the $64,000 question: what is meant by getting them both "right". There seems to be a large variation tolerable in the power response (actually, better to think of early sound and late sound rather than direct and reverberent). For example cinemas are now designed to a variable X curve with an HF rolloff that is totally a function of room size. The variable reverberent field is thought to give a consistent perceived balance. No fixed curve would work.
This all seems very sensible but the question is whether a particular power response is necessary. A lot of evidence points to the direct/early sound being dominant and the power response being pretty secondary. I think CD waveguides are a great thing because they let you achieve well behaved direct field frequency response over a broad listening area, generally with less variation than non CD approaches.
Long, long ago, in a Galaxie far away.
David
Yup, seems sensible to me, esp. if they are singing in your music room. In my case, I can only get them there using recordings. Do we really need double reverberation to make it sound like the original reverberation?
While I try to present my critique in a fun way, the "source" is a serious issue which can't be dismissed.
I got it - we should buy only Neuman-kopf binaural recordings! Nah, that would make it worse because it makes the perception of the hall reverberation more exact (at least for headphone listening).
Lipschitz and Vanderkoy are two smart cookies who use simple research designs and so you are unlikely to find goofs in their methods.
Last edited:
I think Floyd would disagree with you on this, but I know that I do. "right" for the direct field is pretty widely held, and a smooth power response is also pretty widely held, so I don't see "a large variation" as tollerable - some variation in detail, perhaps, but not much more than that.
In what way are these two concepts different? To me they are pretty well indistinguishable, except that "direct" and "reverberant" are clearly defined while "early" and "late" don't really have a definition in this regard.
Could you share that?
Last edited:
You know, I almost cited the 4430 to you as a well-known and respected example of what I was advocating, but then your name sounded familiar...
From a perceptual standpoint, I agree with you. For the purpose of analyzing a loudspeaker, I find it conceptually handy to draw the dividing line between the first-arrival (on-axis anechoic) energy and the reflections (dominated by the off-axis response). Unfortunately I'm not very consistent in my usage.
Very interesting! Do they increase or decrease the HF energy with increasing room size? Is this for the main speakers (L-C-R), surrounds, or all?
Well I'm probably about to lose all credibility here, but my observation is that speakers which stand out as sounding exceptionally good to me almost always incorporate design features likely to result in a reverberant field that is spectrally similar to the direct sound. This list includes SoundLab's big curved panels, Earl's Summa, Wayne Parham's 7 Pi, the MBL Radialstrahler (which has other issues but its timbre is good), Jorma Salmi's Gradient Revolution, the Altec Model 14 and Model 19, and last but not least the JBL Model 4430 (there, I did it after all).
Okay let me try to bring in a little reasoning, rather than relying solely on my own anecdotes. Reflections contribute to perceived loudness, right? Okay if we have more reflected energy at some frequencies than at others, then those frequencies will be perceived as louder than the others. Thus what starts out as a smooth first-arrival sound can be perceptually skewed by the off-axis response. In fact, it is even possible to use an on-axis dip to reduce the audibility of an off-axis peak, or vice-versa. So I'm not ready to concede that the power response (or perhaps more precisely the off-axis response) is perceptually "secondary".
And I welcome any replies, rebuttals, or smack-talk you care to offer.
Duke
In practice it's hard to eliminate reverberation in a home listening room, so it might as well support the timbre of the first-arrival sound.
If we were able to suppress in-room reverberation such that the only reverberant energy we hear is in the direct sound of a stereo pair of speakers, then we'd have a problem: It's coming from the least-effective direction; namely, the same direction as the original sound. The reverberant energy needs to come from all around - or at least from the sides - to convey a sense of liveliness and spaciousness and envelopment (not to be confused with soundstage depth).
Normal in-room reflections do not have the detrimental characteristics that most people assume. Chapter 9 of Floyd Toole's recent book is entitled, "The Effects of Reflections on Sound Quality/Timbre" and chapter 10 is "Reflections and Speech Intelligibility". Let me do a bit of quoting:
"Repetition [is] the audible effect of the same sound being repeated many times at the ear of the listeners. Reflections create new sound events, changing the temporal pattern of the original sound. This could be construed as an error, but in Chapter 8 we found that people like reflections - music in rooms is preferable to music outdoors. Repetition has another aspect, a more subtle one, in that it gives the auditory system more time to examine a sound, more individual "looks", making some aspects of complex sound more audible, and, as will be shown in the following chapter, early reflections make speech more intelligible."
"The most distinctive timbral cues in the sounds of many instruments have been found to be in the onset transients, not in the harmonic structure or vibrato of sustained portions.... This being so, it is reasonable that repetitions of these transient onsets give the auditory system more opportunities to "look" at them and to extract more information."
Toole doesn't specifically address the topic of detection of reverberation on a recording being enhanced by the auditory system having more opportunities to "look" at it, but that seems consistent with the pattern. Let's look at some more quotes:
"In the audio community, it is popular to claim that reflected sounds within small listening rooms contribute to degraded dialog intelligibility. This concept has an instinctive rightness, and it has probably been good for the acoustical materials industry. However, as with several perceptual phenomena, when they are rigorously examined, the results are not quite as expected."
"...Investigations showed that natural reflections in small rooms are too low in amplitude and occur too soon to create problems of this kind ("disturbance" of speech)."
"In the field of architectural acoustics, it has long been recognized that early reflections improve speech intelligibility."
"In small listening rooms some individual reflections have a negligible effect on speech intelligibility, and others improve it, with the improvement increasing as the delay is reduced."
Toole doesn't address every question that can be raised on the subject (I have a few of my own), but he does indicate that reflections in home listening rooms are far more beneficial than intuition would predict.
Bentoronto, it looks like you're into electrostats. I'm a SoundLab dealer, and at one time experimented with absorbing the backwave. The soundstage depth increased, but otherwise the observed effect was it "sucked the life out of the sound". This wasn't what I expected, but in retrospect it was consistent with what Toole reports.
Duke
Last edited:
If I'm far off the mark, Earl can correct me. I think the answer to the first part depends to some degree on the recording, and depends on where measurements are taken. Most recordings are made with the mike close to the instruments. So if the system is adjusted flat at the listener's position in room, it will be pretty hot for those. Maybe different with the recording taken at a listening position - maybe. Even then, those are likely mixed to be heard at some reasonable distance from the speakers, in a room that preferentially absorbs HF. I have no direct experience in that regard, so just assumptions. But I can say that a somewhat falling HF response at the listening position sounds most natural to me.
Sheldon
audiokinesis - thanks for addressing my question. I think some of Toole's quote is somewhere between conjectural and artisnal practice. That's not damning criticism, but state of the art and conjectures change from time to time. At best, the ideal reflection frequency compass and direction compass remain elusive.
So you've addressed the sensible sounding notion about reflections matching original halls in the recording. Now how would you address the way rooms are furnished and how loudspeaker design mates with each unique room... if reflections, early and late, and spectrally defined, matter?
My impression is that making the mathematically perfect speaker leads to diminishing returns when the focus should be on accommodation to recordings and rooms. Hey, I'm only echoing Sheldon's good thoughts (and thanks for your reply, Sheldon).
Yeah, ESL, love 'em. Best is direct drive amps (2400 volt B+... direct to panels, in my former dare-devil system). Like you, I've also experimented moving around some 4 x 4 foot thick acoustic fiberglass panels. Yes, behind the speaker dulls the room sound a bit and, oddly doesn't sharpen the image. Ummm, maybe all this chat about niceties of room reflections really doesn't count for much when you have big dipoles which greatly engage the room big time - like audiokinesis's singers. And I did experiments raising panels off the floor - put that in a simulation! Dipoles are mathematically intractable but aurally delightful. Been taking the backs off speakers since my Bozak's in 1968. Go figure.
So you've addressed the sensible sounding notion about reflections matching original halls in the recording. Now how would you address the way rooms are furnished and how loudspeaker design mates with each unique room... if reflections, early and late, and spectrally defined, matter?
My impression is that making the mathematically perfect speaker leads to diminishing returns when the focus should be on accommodation to recordings and rooms. Hey, I'm only echoing Sheldon's good thoughts (and thanks for your reply, Sheldon).
Yeah, ESL, love 'em. Best is direct drive amps (2400 volt B+... direct to panels, in my former dare-devil system). Like you, I've also experimented moving around some 4 x 4 foot thick acoustic fiberglass panels. Yes, behind the speaker dulls the room sound a bit and, oddly doesn't sharpen the image. Ummm, maybe all this chat about niceties of room reflections really doesn't count for much when you have big dipoles which greatly engage the room big time - like audiokinesis's singers. And I did experiments raising panels off the floor - put that in a simulation! Dipoles are mathematically intractable but aurally delightful. Been taking the backs off speakers since my Bozak's in 1968. Go figure.
Last edited:
HI Sheldon
In nature HF absorption due to air is quite significant. For example, in a large auditoriun almost nothing above 10 Khz ever gets beyond the midpoint in the room (ie. about 6 dB down). Now consider the reflections. Because of this continuos absorption, the reflected energy is constantly getting LP'd, which creates a HF loss that is simply part of the acoustic that we hear everyday. In a small room with low damping, the reverberant field can be much greater at HF than what we are used to in nature and as a consequence a lower power response at HF is "natural".
As Dave S pointed out (somewhere) all movie theaters roll off the high end at various amounts depending on room volume. Theory has it that the smaller the room the greater the roll-off should be. But this, unfortunately, does not consider the inherent power response of the source, which is a serious limitation of the theory.
A number of papers find that the ear uses time windowing for the perception of frequency response and that the window is fairly long at low frequencies and short at high frequencies. In terms of sound reproduction in the home it amounts to the steady state response at low frequencies but the direct response at high frequencies (rejecting the later reverberent field).
Bech shows that at mid frequencies the effective window is long enough to encompass the floor bounce but not likely any later reflections.
Salmi published two papers on the subject. Kates did an excellent paper were he modeled human hearing and ended up with similar windowing that emphasised the direct sound at high frequencies. (yet let in floor bounces in the midrange and a fairly long window for low frequencies). Also Samuel Bridges and Queen.
Floyd might disagree with me but I think his data doesn't. If you look at the 20 speakers he rank orders in his early paper (Loudspeaker Measurements: Part 2) Fig 12 shows that the best ranked speakers have fairly mediocre power response (# 1, 2, and 4 for example) and some units with good power response are near the bottom of the list (#16 and 18, for example). He waffles a bit on the subject but generally states that power response is primarily useful for revealing resonances. Generally his best ranked systems have a downhill trend to their power response and holes wherever the crossover points may be.
Finally, the Lipshitz and Vanderkooy paper previously mentioned.
Although it is frequently put out there as an audiophile ideal, I can't remember any rigorous listening evaluations that concluded a particular power response was essential.
Perhaps a fine distinction, but everything I've read includes the three corner boundaries as important for low frequencies (and then some) and the floor bounce for mid frequencies, so speaking of "early" sound will include them, while "direct" wouldn't. Since the effective time window varies with frequency you can't put a hard label on it, but 20ms is used by some. As you know, concert hall designers use 50ms as a boundary between early and late sound and those interested in speech inteligibility use, I think, 30ms.
David S
I usually refer to direct and reverberent sound as well, but I think early and late is a little more precise.
The bigger the room the greater the rollof in the target curve. Of course this is typical of what would naturally happen to the reverberent field of a larger room, if for no other reason than the HF absorption of air. Is it possible that the direct (early) sound is always flat?
Glad you include the 4430 on the list. To be honest though it isn't a "flat power" speaker. Use of CD waveguides will give flatter power response in the range of a given driver but directivity will still climb through the woofers range, along with there likely being a dip in the power response at each crossover point. In power terms, systems with CD elements are not greatly different than non-CD systems, except for being more level between crossover points. In my mind the reason to pursue constant directivity is that, hopefully, uniform polar response will come along with it.
Again, the key is the delay between the direct and reflected elements.
David S