AVS Forum banner
1 - 18 of 18 Posts

·
Registered
Joined
·
21,345 Posts
Discussion Starter · #1 ·
Hey folks,


I was thinking about how to determine if a speaker was accurate or not by listening to it and I kept running into the problem that everything that I have heard has been colored by my equipment, so listening to anything else simply provides other coloring and I have no idea if it is more or less accurate.


The only test that I have come up with is to listen to a real sound that I am very familar with in real life (not a recording) and then play that back over the various systems in order to determine which reproduces it most accurately. While that might work for me, that sound would not work for anyone else not equally familiar with the same sound.


This puts the kibash on listening to ANY pre-recorded audio as a reference measure for how something should sound...and isn't that how most everybody 'tests' speakers?


The fundamental problem is that our brains adapt to what we have, so even when presented with something "more accurate", it may sound 'wrong' unless we give ourselves sufficient time to adapt, which requires at least several days in my estimation.


The other side is if we adapt to the sound system, who gives a snot if it is accurate to the last 1%, as our brains will fill that in for us. (I actually think listener fatique results from distortion so the last 1% may matter even if we can't put our finger on it in A B X comparos.)


This thinking all comes from the fact that I was contemplating organizing some sort of speaker listening session yet I couldn't come up with even a single test that would work for all the people listening at the same time, as everybody has different speakers and so have different adaptations to their systems. Getting everyone on the same page would require at least several days of listening and that isn't practicable.


The Harman/JBL team often report that it is possible to distinguish the wheat from the chaff in loudspeakers. In many of their whitepapers, they talk about 'trained' vs. 'untrained' listeners. I wonder if it might kind of be the opposite where the 'trained' listeners have lost their inherent training by listening to so many different speaker types that they no longer have a specific reference point, but something of an average reference point across all the speakers. So, if that were the model, then 'trained' listeners would prefer speakers that sound very average and that is exactly what the Harman/JBL research finds.


Does that mean that anyone else will find it sounds good? Perhaps not. And that is also what the Harman/JBL research finds...though in their arogant approach they attribute it to a "lack of training" or something to that effect. I don't buy it. If it sounds good to you, then who am I to say that you are "hearing it wrong"?


Well, that is a lot of words for something as simple as "better or worse?", but I think that most of you will get my point and I hope that it provides for a good discussion.


Thanks for reading and discussing,


LTD02 aka John
 

·
Registered
Joined
·
1,396 Posts

Quote:
Originally Posted by LTD02 /forum/post/20827070


The only test that I have come up with is to listen to a real sound that I am very familar with in real life (not a recording) and then play that back over the various systems in order to determine which reproduces it most accurately. While that might work for me, that sound would not work for anyone else not equally familiar with the same sound.

What about the polar / spatial response of the microphone used? What about the fact that recording something in a reverberant room and replaying it in a reverberant room, might lead to an overly reverberant playback? And vice versa for a dead room recording and dead room playback - will either replicate what our knowledge of the real sound is? I mean just take this video for example:

http://www.realtraps.com/video_diffusors.htm - forward to 11:40


Which of those guitars was the real guitar? All of them? One of them? None of them?

Quote:
The other side is if we adapt to the sound system, who gives a snot if it is accurate to the last 1%, as our brains will fill that in for us.

I guess it depends on how much a coloration bothers you. If a violin sounds like a viola through a speaker, the average person might not recognize it. But a violin player would consider those same speakers unlivable.

Quote:
So, if that were the model, then 'trained' listeners would prefer speakers that sound very average and that is exactly what the Harman/JBL research finds.

I guess the answer to this question is, do the best Revel and JBL speakers sound very average to YOU?

Quote:
Does that mean that anyone else will find it sounds good? Perhaps not. And that is also what the Harman/JBL research finds...though in their arogant approach they attribute it to a "lack of training" or something to that effect. I don't buy it. If it sounds good to you, then who am I to say that you are "hearing it wrong"?

Back to my previous point, if all reproduced violins sound more like a viola, then I just don't understand how that could not be hearing it wrong. The question isn't :::if::: it's wrong, but rather if a person :::cares::: or :::recognizes::: that it's wrong which I think jives with the harman research. The real question, then is, if harman's trained listeners are able to recognize correct from incorrect reliably. I can not say.
 

·
Registered
Joined
·
21,345 Posts
Discussion Starter · #3 ·
ev, thanks for the replies.


let's assume the mic isn't a prob.


"If a violin sounds like a viola through a speaker"


this is exactly my point. how do you know what a violin sounds like and how do you know what a viola sounds like if you only listen through speakers?


"do the best Revel and JBL speakers sound very average to YOU"


my ear is trained to what i listen to, so i can't comment fairly on jbl or revel speakers. that was kind of my point. i don't think that anyone can.


"The real question, then is, if harman's trained listeners are able to recognize correct from incorrect reliably."


that is harman/jbl's argument, but with it i disagree as it fails to compensate for the inherent brain-based acclimation that occurs in a relatively short period of time in our heads.


depending on how frequently you revisit the real sound, you could get a shock that makes everything sound awful, including the real sound...so i'm with you in some sense...but if you don't get the shocks of the real thing, the fiction can, in your head, become just as good as the real thing...e.g., if you have never heard pachelbel's canon in real life, the recording can become the reference point. the recording then becomes an interpretation through the mic (let's assume that is good) through the amp (let's assume that is good) then off the the speakers and room (which are typically no where near to real in representation).


in the big picture, i can see how we could agree. it is with all these other things going on that begs my o.p.
 

·
Registered
Joined
·
21,345 Posts
Discussion Starter · #4 ·
replying to my own post...totally lame...ok, let's put that aside...


some of my favorite tracks, such as enigma - sadeness, don't have any live sound, so there isn't any reference point. same thing with most of enya. in the digital world, who cares what is accurate anymore as accurate has become somes arbitrary one's and zeroe's.
 

·
Registered
Joined
·
2,405 Posts
Unfortunately, in many of those JBL tests in the listening room where speakers could be moved around behind an AT wall, the anechoic response was listed in the paper, and then the tester's opinions, IIRC.


They do nothing to address floor bounce from what I have seen, nor how different a speaker would sound from another seating position.


Anechoic measurements are just that: what the speaker will perform like in an anechoic environment. That the flattest speakers also sounded better overall than peaky ones in testing was nice, but it would have been nice to see how far off from flat the nice flat anechoic response was in the testing room...esp. at different seats.


Don Keele's measurement of a Revel Salon over a reflective floor at differing distances and off axis angles was quite telling. It has nothing close to a flat freq response in the "real" world....are you gonna put enough absorption to matter onto your floor?


In real life, anechoic is dreamland. we have real reflective surfaces in rooms, and although we can treat the room to a certain extent (and practicability), those reflections cannot be completely eliminated (nor may it be fruitful to do so), and the resultant combing will result......


EV brings up a great point. Recordings recorded in a reverberant field played back in another reverberant space will never be "true-to-life".....to hear only what is on the recording and not the specific room you are in, you must use absorption....


Which brings us back to square one. How to best integrate something that was designed to perform well without any nearby reflections in a likely reflective space? #1 would be to either deaden the front wall or even better, recess the speaker into it. Decent article:

http://www.google.com/url?sa=t&sourc...Ceq-RnQIRww8jQ


Another option would be a dead front wall and a groundplane CBT, to use a reflective boundary as an asset to obtain flat freq response....


Many ways to skin this cat, but almost all require absorption in some way, shape or form. Diffusers are simply too hamstrung by bandwidth to be of value for anything typical midrange and below, unless they are made quite deep and large. Broadband absorption is the only treatment that can come close to duplicating anechoic results, which are the benchmark, despite what some say about listening in an anechoic environment being "uncomfortable".


There's a reason people don't post their main/center/surrround freq response at their LP unsmoothed. They for the most part look pretty bad, even with a good EQ "room correction" program on board, esp between seats...that's if you have the measurement equipment to measure accurately at high freqs....Subs, on the other hand, can be measured quite easily and accurately with meager measurement equipment....at least at ear level....ever measure what your feet "hear" vs your head? If you haven't, I'd say don't.....no good results there...



JSS
 

·
Registered
Joined
·
94 Posts
Nice topic you selected, this circle-of-confusion thing. Rather than agree with most of the problems you raised, I would just disagree with one point.
Quote:
Originally Posted by LTD02 /forum/post/20827070


The Harman/JBL team often report that it is possible to distinguish the wheat from the chaff in loudspeakers. In many of their whitepapers, they talk about 'trained' vs. 'untrained' listeners. I wonder if it might kind of be the opposite where the 'trained' listeners have lost their inherent training by listening to so many different speaker types that they no longer have a specific reference point, but something of an average reference point across all the speakers. So, if that were the model, then 'trained' listeners would prefer speakers that sound very average and that is exactly what the Harman/JBL research finds.


Does that mean that anyone else will find it sounds good? Perhaps not. And that is also what the Harman/JBL research finds...though in their arogant approach they attribute it to a "lack of training" or something to that effect. I don't buy it. If it sounds good to you, then who am I to say that you are "hearing it wrong"?

I am sorry, but the situation with 'trained' listeners isn't what you think it is.

Long story short trained and untrained listeners prefer all the same stuff - sound without coloration. Just that untrained listeners are not consistent in threir ratings - they are rating the same speakers differently every time they listen. It is same in all fields, where human 'testing' is involved. It takes a large sample of people to eliminate this and other nuisances from the collected data set. Trained listeners yield excactly the same results, but get there with fewer trials - less people needed.

Ofcourse there will be someone who would prefer a mid-bass hump, midrange cut etc, there is nothing wrong with that, but individual preferences are not what is measured.
 

·
Registered
Joined
·
94 Posts

Quote:
Originally Posted by maxmercy /forum/post/20827371


Anechoic measurements are just that: what the speaker will perform like in an anechoic environment. That the flattest speakers also sounded better overall than peaky ones in testing was nice, but it would have been nice to see how far off from flat the nice flat anechoic response was in the testing room...esp. at different seats.

...

In real life, anechoic is dreamland. we have real reflective surfaces in rooms, and although we can treat the room to a certain extent (and practicability), those reflections cannot be completely eliminated (nor may it be fruitful to do so), and the resultant combing will result......

....

EV brings up a great point. Recordings recorded in a reverberant field played back in another reverberant space will never be "true-to-life".....to hear only what is on the recording and not the specific room you are in, you must use absorption....

No need to entirely dismiss the precedence effect. If you make the reflections 'same' as the direct/'anechoic' sound, it would be mostly ignored by the hearing mechanism.

Trick is that recordings (mostly classical but still) get their meaningful 'hall-sound' from the dense, late reverberant reflections.

No domestic size listening room has those kinds of reflections on their own, so it is quite easy to hear the 'hall ambiance' from the recorging even if played back in the small room. A plus is if those late reflections are coming from the sides, only possible if not using absobing materials on sidewalls, or using multichannel playback.
 

·
Registered
Joined
·
26,476 Posts
You have to pick what is important to you.


Everyone needs a list of speaker priorities and they can be different from person to person.


Its all choice even when we talk about measurements there is also a choice on what matters on different measurements.


example. I want my distortion levels to be very, very good when measuring a speaker at 115dB. I want my CSDs/wavelets to have the best decay possible. I want my off axis response to be uniform. I understand there are compromises in all designs so I have to figure out which compromise is acceptable.


Now what is more important here is that people need to own more and listen more to understand all the differences. The point of reference someone has requires lots of experience. Not only is it important to have heard many different types of designs, its important to push them beyond their know limits and listen to what happens. Listening to a speaker performing in its "comfort range" will tell you little about the speaker. Instead you need to understand what the compromise of the design is and then push it into that performance range to hear what happens.


Its not only important to listen to speakers but its also important to get some drivers and listen to them. Listen to what they sound like with and without Filters set on them.


Conclusion, its only possible to objectively review any design if you accept measurements and controlled listening tests as the objective facts. Even then your objective conclusions are going to be different then someone else's because even measurements will have different meanings to each individual.
 

·
Registered
Joined
·
26,476 Posts

Quote:
Originally Posted by LTD02 /forum/post/20827070


Does that mean that anyone else will find it sounds good? Perhaps not. And that is also what the Harman/JBL research finds...though in their arogant approach they attribute it to a "lack of training" or something to that effect. I don't buy it. If it sounds good to you, then who am I to say that you are "hearing it wrong"?

Everyone has 100% control over what they like/buy/dislike BUT I would say that if their reference point is simply one of hearing BOSE and thinking its great then really do they have enough experience to actually have a valid opinion?

Without the proper experiences/education people are only fooling themselves. Its not about right or wrong at all. Its only their money and they can spend it how the please as long as they never post unexperienced opinion as fact online no one really cares what conclusions anyone has.


Honestly an accurate system does not sound good to most people who grew up with bass heaving, treble heavy systems (> 80% of the population). It takes time and "training" to finally realize how good it sounds, try going back to a inaccurate system after that is definitely hard.


Think of it like this, you have a 5 year PC, its fine for surfing the web, doing spreadsheets, word docs, etc. You never really know its slow at all. You then use a brand new PC for a week, you go back home and man your PC is slow now. This is what experience and referential points do to your expectations so when someone says "It sounds great to me". I automatically think, what experience do they have, how did they setup the listening test, do they have any measurements for validation.


I do not see JBL's appoarch as arogance. People can choose to increase their exposure/experiences or they can choose not too. Its okay for someone to buy a BOSE system and love it. Who is anyone else to judge them on those decisions when they are not posting their opinion as fact to try and validate that purchase anywhere.
 

·
Registered
Joined
·
1,396 Posts

Quote:
Originally Posted by maxmercy /forum/post/20827371


Don Keele's measurement of a Revel Salon over a reflective floor at differing distances and off axis angles was quite telling. It has nothing close to a flat freq response in the "real" world....are you gonna put enough absorption to matter onto your floor?

I'm of the opinion that for most speakers, 4th order crossovers are simply not enough. I think that unless you've got coaxial drivers or high output is not a goal, then you need use sharper slopes to cut out the vertical lobing. At least 110db/octave if not higher. The effect on vertical polar response is outstanding. The Salon2 does not have idea vertical polars, only horizontals.


A few speakers that should have outstanding vertical polars, I would expect to be the Genelec 8260A, Pioneer S-1EX, and KEF Reference 207/2.

Quote:
In real life, anechoic is dreamland. we have real reflective surfaces in rooms, and although we can treat the room to a certain extent (and practicability), those reflections cannot be completely eliminated (nor may it be fruitful to do so), and the resultant combing will result......

Off axis response aside, as far as comb filtering is concerned, there isn't really enough psychoacoustic research done to tell you whether comb filtering is a negative. You can pad down the room completely, but all that does is make us perceive everything as being quieter, and thus we need to turn the volume up in order to get the same level of clarity that we're accustomed to in reverberant spaces, and that will only serve to stress the system.


I know Toole's research showed that at the same SPL, people prefered reflections over no reflections.

Quote:
EV brings up a great point. Recordings recorded in a reverberant field played back in another reverberant space will never be "true-to-life".....

The thing to note is that recordings recorded in reverberant fields are generally done nearfield. You still record late recorded reflections, but you don't record all of the reflections one would experience from being there. Some multichannel setups do record those "seating position" reflections but what most of us want is the sense of being there on most recordings, and most recordings are invariably stereo if not even mono. Being there means having reflections around you filling in much of the information of what you are hearing. It might be reflections off of people sitting beside you, off of chairs, off the ground. The microphone placement of 99% of recordings is not going to capture the timbre the way we would hear it.


I don't think reflections or lack thereof should be cascaded, but setups using absorption rarely sound "right" to my ears for this exact reason. Some reflections in the recording might be preserved, but "what a listener would hear" might not be, because recordings (sensibly) aren't done from a seating position.

Quote:
to hear only what is on the recording and not the specific room you are in, you must use absorption....

But the recording might lack some of the information you're supposed to hear, with the assumption that reflections will "fill it in"

Quote:
How to best integrate something that was designed to perform well without any nearby reflections in a likely reflective space?

But who says they're not designed to perform in a reflective space?


You pointed out that the Salons don't measure well in a real room, but the question isn't whether they measure well in a real room, but whether they subjectively sound better in a damped room. DO they?


As far as frequency response, it's important to search for general trends rather than higher Q peaks and valley.


As far as polar response, we want tight spacing. Anything off axis should reasonably resemble the on axis to an extent. Preferably up to 8khz where possible.


Given the above, what we hear in reflective spaces, especially at the distances many of us listen, isn't going to be the direct response but the power response, so speakers are of course designed with a focus on even (probably tapering) power response.


Quote:
#1 would be to either deaden the front wall or even better, recess the speaker into it.

This creates an infinite baffle, which creates phantom imaging sources. A speaker that has an absolutely minimalist baffle is prefered.

Quote:
Another option would be a dead front wall and a groundplane CBT, to use a reflective boundary as an asset to obtain flat freq response....

CBTs are nice. What do their horizontal polars look like, with the tweeter and midwoofers so far apart? Given the shape of our ears, we're definitely more sensitive to horizontal reflections.

Quote:
Diffusers are simply too hamstrung by bandwidth to be of value for anything typical midrange and below, unless they are made quite deep and large. Broadband absorption is the only treatment that can come close to duplicating anechoic results, which are the benchmark, despite what some say about listening in an anechoic environment being "uncomfortable".

We use anechoic results because their reliably repeatable. However they are not representative of how we hear because none of us has lived our lives in anechoic chambers hearing dominant direct sounds.

Quote:
There's a reason people don't post their main/center/surrround freq response at their LP unsmoothed.

Again, you have to look right back to how we hear. It's like measuring your voice. You can measure your voice in an anechoic chamber to get the most pleasing visual results, but unless you measure its polar response (like Harman does or do they call it the linear spacial response?) you're not capturing all the information at any given microphone point in the measurement.


So returning to the concept of reverberant recording + reverberant room, you have to remember one major factor:


We don't listen in truly reverberant spaces. We listen in small rooms, which don't have "reverb". Treatments should reduce things like flutter echo etc, but we still need immediate (but slightly delayed) reflections to fill in what we "would" immediately hear. Then the reverberation of the recording adds ambience without being cascaded.


My only point earlier was regarding the concept of recording something in a room and playing it back in the same room. I was merely pointing out that the recording methodology must be very carefully analyzed, because a farfield recording played back in the farfield might not sound the same as the original sound in the farfield, but a nearfield recording played back in the farfield may.

Quote:
They for the most part look pretty bad, even with a good EQ "room correction" program on board, esp between seats...that's if you have the measurement equipment to measure accurately at high freqs....

What if you gate it to about 13 to 15ms? Does it still look "pretty bad" in a properly setup, sufficiently wide/deep reverberant room? I suspect "no". It might have some abberations from floor/ceiling reflections but it won't look anything like a 500ms gated measurement which is moonspeak to me at HF. And the assumption returns to "how do we hear" and there just isn't enough research out there to tell me how much our brain will filter out that vertical sound as a natural, desirable artifact. Coaxials like the ones I mentioned do have decent attenuation of vertical response, which reduces intensity of the comb filtering.


The major shift seat to seat between superior speakers will only happen in the top octaves.


Ultimately I think most instruments have distinct timbre, and it varies slightly enough in different listening spaces. Likewise, it is acceptable for our listeniing spaces to slightly change timbre, but not for our playback tools (speakers) to noticably change timbre.
 

·
Registered
Klipsch Heritage, Danley subs, Emotiva, Epson 5050
Joined
·
5,145 Posts
Removing myself from the equation, I pay close attention to the "trained listener" in my room, Dharma, my German Shepherd. She has an uncolored opinion at least. She spent years in the theater never paying attention to what was happening. With the last speaker change, her ears are twitching, her head is popping up, she gets up and barks at things.


Listening habits and volume are the same as before. So what explains the difference? I would think it has fooled her to thinking things are "real". Is she an unbiased measureing device?
 

·
Registered
Joined
·
1,396 Posts

Quote:
Originally Posted by tony123 /forum/post/20828743


Removing myself from the equation, I pay close attention to the "trained listener" in my room, Dharma, my German Shepherd. She has an uncolored opinion at least. She spent years in the theater never paying attention to what was happening. With the last speaker change, her ears are twitching, her head is popping up, she gets up and barks at things.


Listening habits and volume are the same as before. So what explains the difference? I would think it has fooled her to thinking things are "real". Is she an unbiased measureing device?

You probably have some tweeter metal breakup that you can't hear that she can ;P
 

·
Registered
Joined
·
21,345 Posts
Discussion Starter · #14 ·
"Long story short trained and untrained listeners prefer all the same stuff - sound without coloration. Just that untrained listeners are not consistent in threir ratings - they are rating the same speakers differently every time they listen. It is same in all fields, where human 'testing' is involved. It takes a large sample of people to eliminate this and other nuisances from the collected data set."


thanks for the correction. i went back to olive 2003 and that is what is reported.


as for the collapsing variation in trained vs. untrained people in anything, i'm not sure that i agree completely with that proposition. you could have an interesting statistical artifact that messes up the data. if the experts tend to use the ends of the the range and the novices tend to use more of the middle of the range and there is something that is subjectively preferred during the test, then the apparent dispersion among the experts _could_ be higher than among the novices.


that said, collapsing variation as an indicator of skill within a single person's scores is absolutely a measure of good training and it might apply if there is a "right answer" that everybody agrees on after the fact, but won't work if there is some measure of personal preference at work. for example, one of my friends has more sensitive ears than i do. he prefers the volume set about 5 db lower than i do. this hasn't changed in over 20 years. we can both reliably and consistently identify the point where sound becomes "too loud", but it is a different point for each of us. so spl is off the table as a universal. maybe frequency response is on the table, but i'm not so sure, which is why i stoked the coals and made the post. if anybody has seen any recent papers on the topic, it would be cool to get them linked up. there just doesn't seem to be too many people studying this kind of stuff anymore.

 

Olive_2003.pdf 270.64453125k . file
 

Attachments

·
Registered
Joined
·
2,405 Posts
OK, I'll bite on a few of these points....by the way, I am only posting opinion based on my experience only here, which agrees with a lot of what I have read...

Quote:
Originally Posted by Eternal Velocity
I'm of the opinion that for most speakers, 4th order crossovers are simply not enough. I think that unless you've got coaxial drivers or high output is not a goal, then you need use sharper slopes to cut out the vertical lobing. At least 110db/octave if not higher. The effect on vertical polar response is outstanding. The Salon2 does not have idea vertical polars, only horizontals.


A few speakers that should have outstanding vertical polars, I would expect to be the Genelec 8260A, Pioneer S-1EX, and KEF Reference 207/2.

Regardless if you had perfect vertical polars, floor bounce and ceiling bounce make a freq response at near any point near the "perfect radiator" look like the himalayas, esp down low, where directivity is not good...


Off axis response aside, as far as comb filtering is concerned, there isn't really enough psychoacoustic research done to tell you whether comb filtering is a negative.

Agreed....


You can pad down the room completely, but all that does is make us perceive everything as being quieter, and thus we need to turn the volume up in order to get the same level of clarity that we're accustomed to in reverberant spaces, and that will only serve to stress the system.

Not in my opinion. I listen at even lower levels now that my room is mainly "dead". 10dB below reference now seems quite loud....I cannot explain why...


I know Toole's research showed that at the same SPL, people prefered reflections over no reflections.

Maybe a case of "preference" over "reference"? Again, I do not know...I can see some reflections being helpful to give a feeling of spaciousness or the like, but I have much more enjoyed the pinpoint imaging that I get with sidewall absorption. Good multichannel material will have the surrounds fill in some reverberant cues...again, only my opinion....


The thing to note is that recordings recorded in reverberant fields are generally done nearfield. You still record late recorded reflections, but you don't record all of the reflections one would experience from being there. Some multichannel setups do record those "seating position" reflections but what most of us want is the sense of being there on most recordings, and most recordings are invariably stereo if not even mono. Being there means having reflections around you filling in much of the information of what you are hearing. It might be reflections off of people sitting beside you, off of chairs, off the ground. The microphone placement of 99% of recordings is not going to capture the timbre the way we would hear it.


I don't think reflections or lack thereof should be cascaded, but setups using absorption rarely sound "right" to my ears for this exact reason. Some reflections in the recording might be preserved, but "what a listener would hear" might not be, because recordings (sensibly) aren't done from a seating position.


But the recording might lack some of the information you're supposed to hear, with the assumption that reflections will "fill it in"


But who says they're not designed to perform in a reflective space?

Agreed...


You pointed out that the Salons don't measure well in a real room, but the question isn't whether they measure well in a real room, but whether they subjectively sound better in a damped room. DO they?

Not sure, but a good question....would hearing a freq response closer to the anechoic response result in greater listener 'satisfaction'??? Again, a question I can only postulate that the answer could be yes, but since I don't have flat freq response in my room due to floor, ceiling, seating, front non-AT screen interactions causing freq response from flat, I don't know....


As far as frequency response, it's important to search for general trends rather than higher Q peaks and valley.

Agreed for the most part, but it does depend on the program material. With more films using sweeps as sound effects in 'gravitas' moments, even high Q valleys are quite pronounced, esp below 150Hz or so. Again, most would never notice, and if they knew no better, they would think it was mixed with the high Q hole on purpose....as they may have not listened to it with it filled in...


As far as polar response, we want tight spacing. Anything off axis should reasonably resemble the on axis to an extent. Preferably up to 8khz where possible.

Agreed...


Given the above, what we hear in reflective spaces, especially at the distances many of us listen, isn't going to be the direct response but the power response, so speakers are of course designed with a focus on even (probably tapering) power response.

Not sure, depends on the room, but in typical non-treated rooms, agreed...


This creates an infinite baffle, which creates phantom imaging sources. A speaker that has an absolutely minimalist baffle is prefered.

Not sure what you mean by phantom imaging sources here if we simply recess a speaker to be flush to the wall, in order to avoid 1/4 wavelength suckouts at omnidirectional freqs, and avoidance of baffle step compensation in the crossover schematic, a recessed speaker is a seemingly win-win, and seen in many control rooms, mixing stages, and every THX cinema install.....so you are telling me the THX baffle wall implementation suffers from multiple phantom imaging sources???? Maybe we just don't understand eachother...I am basically saying that in-wall speakers have potentially less problems (assuming vertical directivity is handled so that the major forward lobe is aimed at the seats)....


CBTs are nice. What do their horizontal polars look like, with the tweeter and midwoofers so far apart? Given the shape of our ears, we're definitely more sensitive to horizontal reflections.

With 4" midwoofers and dome tweets, C-C distance is less than 75mm, allowing for nulls at over 75 degrees off axis with a crossover freq of 2000Hz....pretty good off-axis performance (you are limited by woofer and tweeter directivity more than by crossover freq nulls), and this assumes omnidirectional radiation....power response in the non-longitudinal direction (horizontal for a vertical CBT) will never be as good as that in line with the CBT's longitudinal axis (usually vertical, except for center channel, non-AT screen usage).....it's the limitation of the design....which sucks. But getting things right on one axis and the ability to eliminate 2 of the room's boundaries when implemented in a groundplane configuration, it has many pluses....


We use anechoic results because their reliably repeatable. However they are not representative of how we hear because none of us has lived our lives in anechoic chambers hearing dominant direct sounds.


Again, you have to look right back to how we hear. It's like measuring your voice. You can measure your voice in an anechoic chamber to get the most pleasing visual results, but unless you measure its polar response (like Harman does or do they call it the linear spacial response?) you're not capturing all the information at any given microphone point in the measurement.


So returning to the concept of reverberant recording + reverberant room, you have to remember one major factor:


We don't listen in truly reverberant spaces. We listen in small rooms, which don't have "reverb". Treatments should reduce things like flutter echo etc, but we still need immediate (but slightly delayed) reflections to fill in what we "would" immediately hear. Then the reverberation of the recording adds ambience without being cascaded.


My only point earlier was regarding the concept of recording something in a room and playing it back in the same room. I was merely pointing out that the recording methodology must be very carefully analyzed, because a farfield recording played back in the farfield might not sound the same as the original sound in the farfield, but a nearfield recording played back in the farfield may.

Agreed. The "circle of confusion" is both great and terrible at the same time...


What if you gate it to about 13 to 15ms? Does it still look "pretty bad" in a properly setup, sufficiently wide/deep reverberant room? I suspect "no". It might have some abberations from floor/ceiling reflections but it won't look anything like a 500ms gated measurement which is moonspeak to me at HF. And the assumption returns to "how do we hear" and there just isn't enough research out there to tell me how much our brain will filter out that vertical sound as a natural, desirable artifact. Coaxials like the ones I mentioned do have decent attenuation of vertical response, which reduces intensity of the comb filtering.

If you gate it back, you lose resolution, and the lower octaves. It's a tradeoff.....again, back to how we hear, and how much listener preference adds to the equation...agreed on the directivity issue, and the coaxials... Throw the sound at the people listening....


The major shift seat to seat between superior speakers will only happen in the top octaves.

I always thought that below the Schroeder frequency the majority of seat-to seat discrepancies took place....above Schroeder, not as much, as modal density is very high....and comb filtering makes any measurement look odd without smoothing...but it depends on listener position and speaker polars, and........


Ultimately I think most instruments have distinct timbre, and it varies slightly enough in different listening spaces. Likewise, it is acceptable for our listeniing spaces to slightly change timbre, but not for our playback tools (speakers) to noticably change timbre.

Do our playback tools change timbre if we move them in the room? Or add room treatments? If timbre can be defined as frequency response, or of a harmonic signature, the answer could be maybe....Actually, that could make a pretty good investigation....would the speaker sound any different away from the front wall vs close to it, or would the difference only be in the lower freqs? And would the essentially same HF response make it sound more similar than different??? All things I can only postulate on...
Good discussion....


JSS
 

·
Registered
Joined
·
21,345 Posts
Discussion Starter · #16 ·
the problem of controlled directivity vs. omni speakers just won't go away. many folks do prefer the omnis in rooms with lots of reflections and it provides a sense of openess and space to the music, where controlled directivity speaks horns or bipoles sound more like headphones. the latter is better for dialog clarity, but i can appreciate folks' opinion on the former.
 

·
Registered
Joined
·
1,396 Posts
Quote:
If you gate it back, you lose resolution, and the lower octaves. It's a tradeoff
The lower octaves, yes, but as far as resolution is concerned, most research i've seen indicates somewhere from 10 to 20ms is where our ears determine clarity, and everything after is effectively delinieated as a late arriving reflection to our mind.

Quote:
Maybe a case of "preference" over "reference"? Again, I do not know...I can see some reflections being helpful to give a feeling of spaciousness or the like, but I have much more enjoyed the pinpoint imaging that I get with sidewall absorption.
For mastering of movies, there is perhaps a reference. But in general, there IS no reference. Here's studios of some very relevant people for example:







which of the above looks "reference" *padded to you?


As far as imaging, I find speakers can image so clearly it actually sounds unrealistic. Sometimes vocals will sound more defined in space than I've ever heard a real life vocal. I'll take depth and a diffuse, but accurate stereo image, over an exaggeratedly sharp stereo image, based on my subjective experiences.


You're right in that there's a level of subjectivity that can't be defeated, but at some point the term "reproduction" is necessarily based on the real life experience.

Quote:
Not sure what you mean by phantom imaging sources here if we simply recess a speaker to be flush to the wall
If the tweeter and midrange, sans baffle, are effectively radiating into 2pi space this makes sense. I'm sure you as well I know the requirements to get uniquivocally forward directivity. Most drivers are small enough at some frequency that they would, in open air, radiate into 4 pi space where it's in a wall or baffle, that 4pi space radiation returns to being 2 pi space, but the re-reflected sounds off the baffle will still be distinct from the original radiation to some extent or another. Thusly, a phantom imaging source.


It stands to reason that where possible, the best baffle would be an extremely minimal teardrop shaped baffle far away from any boundaries, not a flat, infinite baffle which itself is a boundary.


Of course this brings us back to the issue of baffle step compensation and the resulting tilt in power response.


I think there`s a tradeoff no matter what you do, output, extension, intelligibility, practicality. Maybe those guys on diyaudio have the right of it, with their baffle-less free air drivers hanging off tight cords. The polar, power, and frequency responses look amazing. I doubt they can handle any real power though.

Quote:
I always thought that below the Schroeder frequency the majority of seat-to seat discrepancies took place....above Schroeder, not as much, as modal density is very high....and comb filtering makes any measurement look odd without smoothing...but it depends on listener position and speaker polars, and........
Sorry, I was mainly pointing out how most tweeters beam the top octaves. In the bass, all guesses are off. Especially from around 120hz to 300+hz, you really do need bass traps. Though dipoles and cardioids tend to excite room modes differently from the convention speaker, of course.
 

·
Registered
Joined
·
5,897 Posts
As far as timbre is concerned, I'm convinced that power response is hugely influential. You can take two very dynamic low compression low distortion speakers with similarly flat on axis response but radically different power response, one somewhat close to flat and the other nowhere close, and walk out of the room to an adjoining space and instantly tell which is which. And more importantly, instantly tell which is a reproduction and then be amazed at how lifelike the other really is.


Which confuses the issue of reverberant large spaces, reflective vs dead spaces, hearing the recording vs hearing the room, toole's recent speech intelligibility and lateral reflection preferences work, etc for me. As in, whether lateral reflections (and others) are 'good' or not is confounded by my belief that using absorption to control them, thus improving power response, thus improving the accuracy of timbre is at least as important to me as any impact on speech intelligibility etc. You would think that would be reflected in toole's findings; I cannot easily explain why it does not appear to be though I do have a strong hunch.


As for baffles, flush mount ib is technically/theoretically superior in most every way, except that it robs me of one the best IMO ways to control power response.
. Though you can always use horns to work on directivity in an IB, which has of course been famously done in the octagon.


As for the original question/op, my reference is always live music. As a musician that may be more practical for than for others. I would hope that trained listeners have a similar reference, and that averaging of the untrained population would converge towards the same preferences and priorities, as was alluded to earlier.


As for how to organize an 'objective' speaker audition, that's tough for simply yourself. Near impossible for a group. Just have fun. Draw solid conclusions where differences are obvious and agreement is near unanimous, place much less importance on impressions that are subtler and more varied from one listener to the next as this is the area where psychology loves to wreak havoc.
 
1 - 18 of 18 Posts
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top