Etienne88 said:
I completely agree here, but I would not want to remove them even if I could. In my book on HT (posted at my web site) I show how to damp these modes very effectiviely.
You cannot have a quality reproduction and leave the side early refelctions. Highly directional loudspeakers are very effective at removing these very detrimental early reflections and no room modifications are required. But usually the floor and ceiling are too close and something like you suggest must still be done.
Ah, you are forgetting one key thing - I use bandpass enclousre designs. In this design the Fs of the driver in the box is at the middle of the passband NOT at the lower edge. What you say is only true for a highpass closed box. This is a major advantage of the bandpass designs and is not well understood.
In a bandpass design the driver is almost irrelavent as the box itself dominates the tuning. In other words I can put just about any 12" in one of my 12" designs and it works virtually the same. Couple that to the fact that the acoustic output is LP filtered for reduced nonlinear effects (hence the driver linearity is almost irrelavent) and you can see that the driver doesn't matter much at all.
Bandpass designs work very well with about a 1 octave bandwidth and up to 2 octaves bandwidth is reasonable. But beyond that a bandpass design can't get to higher bandwidths without a serious loss of efficiency. The technique of staggering the sub tuning allows for the subs to cover about 25 Hz through 150 - 200 Hz without any problem. The larger sub operates only over a very narrow bandwidth of about 25 - 50 Hz. Its panel resonances are not a big concern because there is no driving force at higher frequencies and there are no cabinet resonances below about 100 Hz.
In general I use no EQ at all. There have been times (very seldom) when EQ made some improvements, but I have never used more than one band of EQ. With three subs, each with gain, phase and LP filter paramters that can be set, there is always something that works. Monopoles all sum nicely and create a very controllable situation.
Ok, it seems to me that both solutions work according to your experiences! I think I will try both approaches since Earl solution does not put that much requirement on the type of driver. When it will be done is another question, so don't hassle me to much asking about progress! 
John, the acoustic resistance box add a constant delay to the rear wave, I get that!
I have a question about the type of acoustic resistor you recommend. I was thinking about this kind of UAR foam, the one below. It is an open cell foam. The density is 100kg/m3 (which should be around 6200 pound/foot3). I used the 30mm (1,2 inch) version to build bass traps which worked pretty well.
I can also imagine using wool felt, many layers if I cannot get a thick piece.
Earl, I considered bandpass before and left it aside for two reasons: the group delay is quite high (higher than closed boxes if I remember well...) and it is limited to 2 octaves as you mentioned. Your previous posts make me reconsider the bandpass construction!
As Pan said earlier in this thread, SR (Swedish Radio) recommends keeping the group delay below 0.8*frequency, which is an interesting recommendation even it was not that much background behind it... I did a try with WinISD with a Peerless SLS 12 in a 4th order BP in the proposed configuration, the GD was below 0.8*f even at 10Hz. What is your experience with other drivers?
You used 4th or 6th order bandpass?
I check your website in search of some readings about construction tips. Unfortunately only the 2 first chapters are available... So I will ask you directly! You mentioned in this thread CLD materials that you use for walls. Do you have any experience about using a sheet of roof paper between 2 plaster boards? Of course the choice of glue is of the highest importance, Liquid Nails is not easily available here so I was thinking about a neoprene glue that will not be hard and will older smoothly.
Do you recommend CLD for floor material as well? I read somewhere that it has good damping properties for taping noises.
Regards,
Etienne
John, the acoustic resistance box add a constant delay to the rear wave, I get that!
I have a question about the type of acoustic resistor you recommend. I was thinking about this kind of UAR foam, the one below. It is an open cell foam. The density is 100kg/m3 (which should be around 6200 pound/foot3). I used the 30mm (1,2 inch) version to build bass traps which worked pretty well.
I can also imagine using wool felt, many layers if I cannot get a thick piece.
Earl, I considered bandpass before and left it aside for two reasons: the group delay is quite high (higher than closed boxes if I remember well...) and it is limited to 2 octaves as you mentioned. Your previous posts make me reconsider the bandpass construction!
As Pan said earlier in this thread, SR (Swedish Radio) recommends keeping the group delay below 0.8*frequency, which is an interesting recommendation even it was not that much background behind it... I did a try with WinISD with a Peerless SLS 12 in a 4th order BP in the proposed configuration, the GD was below 0.8*f even at 10Hz. What is your experience with other drivers?
You used 4th or 6th order bandpass?
I check your website in search of some readings about construction tips. Unfortunately only the 2 first chapters are available... So I will ask you directly!
You mentioned in this thread CLD materials that you use for walls. Do you have any experience about using a sheet of roof paper between 2 plaster boards? Of course the choice of glue is of the highest importance, Liquid Nails is not easily available here so I was thinking about a neoprene glue that will not be hard and will older smoothly.Do you recommend CLD for floor material as well? I read somewhere that it has good damping properties for taping noises.
Regards,
Etienne
Etienne88 said:
I don't worry about group delay at these low frequencies. I do worry about some forms of group delay above about 1 kHz where it is know that audibility increases with frequency (untill about 5 kHz) and level. But its really the non-minimum phase group delay that I worry about. The minimum phase group dealy appears to not be nearly as audible.
I guess that my bandpass would be considered 6th order non-symmetric since I generally use the LP filter in the sub amps. But this LP in the amps need not coincide with the acoustic LP in the bandpass. They could be stagered. As I say these details are all room dependent.
CLD should work well for floors. It has worked well everywhere that I have used it, but it is a pain to do, so I usually limit it to the most important and effective situations. A floor between to living spaces would be a very effective application.
gedlee said:
May I ask how you came to the conclusion that group delay is not important in the bass?
Regarding sensitivity to GD increasing from 1k to 5k, do you have a link or something to any studies on this?
How come you worry about non-minimum phase GD and still don't worry about a bandpass sub to augment the bass of a main speaker?
/Peter
Etienne88 said:
I'm not sure SR recommends that or if this studies was made for SR. What I wanted to say was that the person that did these studies also has done work for SR (and other organisations).
Also it should be 0.8/F which gives the maximum (with some margin) tolerable GD in seconds.
If someone is interested I can try to dig for some more detailed info.
/Peter
Back to basics
There is no issue with LR4 or even LR8 XO > 1kHz. But of course there is with LR4 XO @ around 100 Hz!
A look at the definition of group delays and the audibility curves helps
@Earl: Interesting point with the audibility of group delays as a function of SPL. Do you have any references/links to that?
HiFiNutNut said:
gedlee said:
There is no issue with LR4 or even LR8 XO > 1kHz. But of course there is with LR4 XO @ around 100 Hz!
A look at the definition of group delays and the audibility curves helps
@Earl: Interesting point with the audibility of group delays as a function of SPL. Do you have any references/links to that?
Re: Back to basics
Any discussion of the audibility of group delay has to be based on at least two references 1) Brian Moores AES paper on the subject, and Lidia and my paper on the same subject (see www.gedlee.com).
Moore shows the basics, but concludes that group delay in his experiment was not audible for loudspeakers in real rooms. The problem is that his experiment was pretty specific and I hypothesized that there exists common conditions under which GD would be audible. GD is audible above about 1 kHz in real rooms IF the loudspeakers have a clean impulse response (rare) and the room does not have early reflections (just as rare). If these later two conditions are not met then the existing GD will mask the test GD.
The audibility goes up with absolute SPL as we showed.
This SPL factor is important because virtually none of the studies on GD contol this factor or even quote it. Moore does and his levels were pretty low so his results are no surprise. Unfortunately a lot of people point to this study as proof that GD is not audible - which is incorrect.
Fosti said:
A look at the definition of group delays and the audibility curves helps
@Earl: Interesting point with the audibility of group delays as a function of SPL. Do you have any references/links to that?
Any discussion of the audibility of group delay has to be based on at least two references 1) Brian Moores AES paper on the subject, and Lidia and my paper on the same subject (see www.gedlee.com).
Moore shows the basics, but concludes that group delay in his experiment was not audible for loudspeakers in real rooms. The problem is that his experiment was pretty specific and I hypothesized that there exists common conditions under which GD would be audible. GD is audible above about 1 kHz in real rooms IF the loudspeakers have a clean impulse response (rare) and the room does not have early reflections (just as rare). If these later two conditions are not met then the existing GD will mask the test GD.
The audibility goes up with absolute SPL as we showed.
This SPL factor is important because virtually none of the studies on GD contol this factor or even quote it. Moore does and his levels were pretty low so his results are no surprise. Unfortunately a lot of people point to this study as proof that GD is not audible - which is incorrect.
Skiming thru the AES paper about linear distortion at www.gedlee.com it seems like you have investigated the audibility of non linear frequency response and combfiltering from gross delays ie. reflections.
Do I need to read more carefully or do I understand the way the study was done?
/Peter
Do I need to read more carefully or do I understand the way the study was done?
/Peter
Re: Re: Re: Back to basics
Yes, this was the very earliest work. I don't recal the details, but the gist, if I recall correctly, was that it was not all that audible. They tested for thresholds, not practical levels found in real speakers in real rooms. Thats what makes the Moore paper so important and our paper an important addition. Mostly GD is NOT audible, but there ARE things that can happen that are audible. My position is that in most cases its not an audible factor of importance, but for very good loudspeakers setup in very well designed rooms the limiting factor of sound quality does appear to be GD "types" of effects. But for the most part this is not an issue because few loudspeakers are good enough and even fewer rooms are either.
al2002 said:
Yes, this was the very earliest work. I don't recal the details, but the gist, if I recall correctly, was that it was not all that audible. They tested for thresholds, not practical levels found in real speakers in real rooms. Thats what makes the Moore paper so important and our paper an important addition. Mostly GD is NOT audible, but there ARE things that can happen that are audible. My position is that in most cases its not an audible factor of importance, but for very good loudspeakers setup in very well designed rooms the limiting factor of sound quality does appear to be GD "types" of effects. But for the most part this is not an issue because few loudspeakers are good enough and even fewer rooms are either.
Re: Re: Re: Back to basics
You have to understand that we don't do "threshold" testing. I find that this tends to be misleading and leads to overdesign in most cases. We test for audibility in terms of better/worse which allows for a kind of scaling of the problem. And we tested for a mixture of GD signals into the main signal path (ala diffraction)and not for GD in the entire transmision path. These are quite different things.
But for what we did the audiblity would drop down to insignificant below about 80 dB (at the ear) for 2 ms.
Fosti said:
You have to understand that we don't do "threshold" testing. I find that this tends to be misleading and leads to overdesign in most cases. We test for audibility in terms of better/worse which allows for a kind of scaling of the problem. And we tested for a mixture of GD signals into the main signal path (ala diffraction)and not for GD in the entire transmision path. These are quite different things.
But for what we did the audiblity would drop down to insignificant below about 80 dB (at the ear) for 2 ms.
I usually prefer reading rather than commenting on these posts but there are a few things that have caused me to comment.
1. A better reference from BCJ Moore's group (re: group delay) would be the JAES article he had about 2 years ago (when you search, Moore was not the 1st author). The short summary is that Group Delay is detectable. Although the threshold is a function of a number of things including individual differences and room reverberation (where it is more difficult to detect).
2. Using music as a test signal is a poor choice if you are interested in discovering "when and where" group delay might be detectable.
3. There is little to recommend for using a preference rating. Measuring thresholds actually provides much stronger evidence. When this is done with trial-by-trial feedback it forces the subject to focus on the cue, learn the differences, provides more stable responses and can minimize the differences between listeners.
Unfortunately your technique does not offer these advantages and is also subject various response biases and contextual effects. I can not elaborate on these in a 50 words or less, but they are well documented phenomena in scaling research, measurement and experimental psychology. I think some of these problems also spill over to your published work on the audibility of harmonic distortion.
I am not trying to pick a fight. Rather I am trying add some perspective. Measurement issues on human performance are complex and are frequently short-changed. But there are some real problems in scaling and detection that interact with memory, expectation, context, alertness and attention. I would love to discuss this with you sometime; however a forum is probably not the best place. Part of the problem is that a diverse audience will have different backgrounds and use technical terms differently.
I guess the reason I am bringing this stuff up is because many of the human measurement & performance issues are being ignored or misrepresented. Since that is the field I work in I tend to see it as being critical.
-Tom
1. A better reference from BCJ Moore's group (re: group delay) would be the JAES article he had about 2 years ago (when you search, Moore was not the 1st author). The short summary is that Group Delay is detectable. Although the threshold is a function of a number of things including individual differences and room reverberation (where it is more difficult to detect).
2. Using music as a test signal is a poor choice if you are interested in discovering "when and where" group delay might be detectable.
3. There is little to recommend for using a preference rating. Measuring thresholds actually provides much stronger evidence. When this is done with trial-by-trial feedback it forces the subject to focus on the cue, learn the differences, provides more stable responses and can minimize the differences between listeners.
Unfortunately your technique does not offer these advantages and is also subject various response biases and contextual effects. I can not elaborate on these in a 50 words or less, but they are well documented phenomena in scaling research, measurement and experimental psychology. I think some of these problems also spill over to your published work on the audibility of harmonic distortion.
I am not trying to pick a fight. Rather I am trying add some perspective. Measurement issues on human performance are complex and are frequently short-changed. But there are some real problems in scaling and detection that interact with memory, expectation, context, alertness and attention. I would love to discuss this with you sometime; however a forum is probably not the best place. Part of the problem is that a diverse audience will have different backgrounds and use technical terms differently.
I guess the reason I am bringing this stuff up is because many of the human measurement & performance issues are being ignored or misrepresented. Since that is the field I work in I tend to see it as being critical.
-Tom
Etienne88 said:
I think you meant 1.602 #/ft^3.
I don't know how it would work. It is really a matter of building, testing, measuring, tuning.I've been looking at the in romm behavior of different sources. In summary:
Dipole:
Consequences of position (independent of orientation): A dipole will not excite a mode when placed at a position where that mode has a velocity node.
Additional consequences due to directionality (orientation): A dipole will not excite room modes orthogonal to the dipole axis.
Monopole:
Consequences of position: A monopole will not excite a mode when placed at a position where that mode has a pressure node.
Additional consequences due to directionality: Being omni directional, a monopole has no consequences arising from directivity.
Cardioid:
Consequences of position: In general a cardioid is not limited by position alone with regards to excitation of modes.
Additional consequences due to directionality: A cardioid will not excite a mode when placed at a position where that mode has a pressure node IF it is also oriented such that the mode is orthogonal to the axis of the cardioid.
These results are based on consideration of a room with rigid walls. Wall with finite admittance will change this a little, but obviously any change is gradual depending on how difference the admittance is from zero, frequency dependence, etc.
These are just consideration of which modes may (or should we say may not) be excited. How each mode contributes to the in room SPL at a given point is a separate issue. But it can't contribute if is isn't excited.
The dipole and cardioid are at the oppsite ends of hte spectrum. The cardioid is least sensitve to placement, with regard to how many modes may be excited, because it must be both positioned and oriented in a specific manner to NOT excite a mode.
WithTarragon said:
I appreciate your post and your comments - they are issues that I am familiar with.
First, I'm not real interested in a discussion off-line, because the point of my being here is to educate people so that audio won't continue on in its "dark ages". But I certainly understand your unwilingness to do so.
To our methods. There is a lot of discussion of intrinsic versus extrinsic testing. One being ideal signals (like imulses and sine wave) and situations that lead to very stable and "clean" results and statistically valid "thresholds". But there is a downside to this kind of testing - its not readily apparent how it applies in a real world situation. It intrinsicly valid but extrinsically (how it applies elsewhere) is not as well defined. We use extrinsical methods that test directly within the area that we are interested in. They have far less reliability and resolution, but since they used real world signals (music) and real people (non-trained students) they were valid within that area at least.
I am first and formost a loudspeaker designer not a psychoacoustician. I am very willing to do the "dirty" test if it "guides" my designs in the right direction. Hence I do not strive for "accuracy" as much as "fast answers" to pertinent questions. Accuracy will work itself out in the long run through design iterations etc. But one has to know the direction to go in even if it is only a rough idea.
We did not present any our our subjective work for publication (except one) because we knew that it was prone to criticism by those who wanted "accuracy" over "quick answers". So what you say is neither surprising or new to us.
The "better" Moore paper that you refer to is the same one that I was refering too. The authors did conclude that GD was probably not a factor on typical loudspeakers in real rooms with music. They did show that it was perceivable in certain cases however. So we are not in any disgreement on this paper at all and I am unclear why you imply that there was.
To your point (2) - ""when and where" group delay might be detectable" is most certainly NOT my interest. Detectability of aberations in loudspeakers is not very useful to a designer. He wants to know what of the various aberations that he has to contemplate is going to yield the worst sound quality in a given set of tradeoffs. For the current state-of-the-art in loudspeaker design, detectability is virtually irrelavent to this task.
To your point (3) we did not use a "preference rating". We asked the subjects to scale how audible a certain aberation was when compared to a reference - NOT which is prefered. Now there is no guarantee that the subjects did what was asked of them, but the instuctions were very clear on this point.
The whole goal of our research is to investiagte the priorities of problems in loudspeakers so that an optimum design can be achieved within a certain set of constraints. For your interests our work is very coarse and unrevealing, but for my intentions it has been quite profound.
You have to admit that our methods are an order of magnitude better than a single subject listening to his own experiment, unblind and then claiming reliable results and "proof" because they "heard something". Yet this later technique is where almost 100% of audio dogma has come from. This is what I was trying to combat and NOT to add to the body of knowledge about sound perception as a field of study.
Your comments are fine as long as everyone knows that they are geared more towards a "research paper" than an "engineering report".
Your correct that our goals are probably different. As such your test methods would meet your degree of sufficiency.
My reaction was based on terminology that gets loosely thrown around about audibility and detectability etc. It was also driven by my experience that some of the sources of variance that I mentioned do account for findings that are narrow in their applicability or fail to be replicated.
Certainly we are in agreement that a single person who twiddles some knobs and "decides" whether something is audible or not is a poor measure. But that is also a fairly low standard.
Some of the the issues I mentioned about threshold measurement (and "preference scale" is a generic term that includes your tasks) was meant to be helpful and could actually improve your measurement tasks and make them more generalizable and interpretable. If you are interested in pursuing this (and you may not be so inclined) there are handbook articles and textbooks on the subject. It would help you out on your data collection, design and analysis.
Regarding training, individual differences and test signals, we will probably need to agree to disagree on that set of issues. I simply see things very differently.
Anyhow, my comments were not meant to be a simple criticism but rather they were meant to be helpful.
-Tom
My reaction was based on terminology that gets loosely thrown around about audibility and detectability etc. It was also driven by my experience that some of the sources of variance that I mentioned do account for findings that are narrow in their applicability or fail to be replicated.
Certainly we are in agreement that a single person who twiddles some knobs and "decides" whether something is audible or not is a poor measure. But that is also a fairly low standard.
Some of the the issues I mentioned about threshold measurement (and "preference scale" is a generic term that includes your tasks) was meant to be helpful and could actually improve your measurement tasks and make them more generalizable and interpretable. If you are interested in pursuing this (and you may not be so inclined) there are handbook articles and textbooks on the subject. It would help you out on your data collection, design and analysis.
Regarding training, individual differences and test signals, we will probably need to agree to disagree on that set of issues. I simply see things very differently.
Anyhow, my comments were not meant to be a simple criticism but rather they were meant to be helpful.
-Tom
Here is a link about Time, Frequency, Phase and Delay reprinted from Speaker Builder Magazine that might be helpful for some readers. It didn't really help me... 
so I kept searching and I found a very good document from Jean-Michel Le Cléac'h that you can download here if you speak French.
Some interesting points out of JMLC's document:
The group delay can be calculated from the impulse response. Or it can easily be seen on a spectrogram (which is a waterfall plot seen from the top) of the same impulse response.
The origins of the group delay:
- preamps and amps (level negligible in comparison with the following)
- filters
- the way you load your drivers (horns, BR, BP, closed box, dipole, etc...)
- the room (JMLC adds that the reponse of a room is NOT minimal phase)
Below 100Hz the speaker is the main source of phase distortion.
Blauert and Laws results have very little significance in hifi. JMLC also says that old studies conclude that phase distortion was not audible. But more recent studies says the opposite. It seems nevertheless that the differences are quite subtle and that not everybody can hear phase distortion. A synthesis of all the works done in this field has been done by Daisuke Koya as a thesis in the university of Miami. The document is called AURAL PHASE DISTORTION DETECTION. (link)
In his document, JMLC presents how to build the filter of a 3 ways speaker with minimal phase distortion.
All what is written above comes from JMLC, I do no stand for it (even if I might agree with it) and I would be unable to counter argue. I just hope this will help somebody to understand better group delay / phase distortion as it did for me.
Thank you John for your summary. I like the cardioid principle for it easiness of positioning!
As usual I have some questions!
Do OB or acoustic resistance box induce GD? I have the feeling that OB will have no to very little GD and that acoustic resistance box will have level comparable to closed box...
Regards,
Etienne
so I kept searching and I found a very good document from Jean-Michel Le Cléac'h that you can download here if you speak French.Some interesting points out of JMLC's document:
The group delay can be calculated from the impulse response. Or it can easily be seen on a spectrogram (which is a waterfall plot seen from the top) of the same impulse response.
The origins of the group delay:
- preamps and amps (level negligible in comparison with the following)
- filters
- the way you load your drivers (horns, BR, BP, closed box, dipole, etc...)
- the room (JMLC adds that the reponse of a room is NOT minimal phase)
Below 100Hz the speaker is the main source of phase distortion.
Blauert and Laws results have very little significance in hifi. JMLC also says that old studies conclude that phase distortion was not audible. But more recent studies says the opposite. It seems nevertheless that the differences are quite subtle and that not everybody can hear phase distortion. A synthesis of all the works done in this field has been done by Daisuke Koya as a thesis in the university of Miami. The document is called AURAL PHASE DISTORTION DETECTION. (link)
In his document, JMLC presents how to build the filter of a 3 ways speaker with minimal phase distortion.
All what is written above comes from JMLC, I do no stand for it (even if I might agree with it) and I would be unable to counter argue. I just hope this will help somebody to understand better group delay / phase distortion as it did for me.
Thank you John for your summary. I like the cardioid principle for it easiness of positioning!
As usual I have some questions!
Do OB or acoustic resistance box induce GD? I have the feeling that OB will have no to very little GD and that acoustic resistance box will have level comparable to closed box...
Regards,
Etienne
Etienne88 said:As usual I have some questions!
Do OB or acoustic resistance box induce GD? I have the feeling that OB will have no to very little GD and that acoustic resistance box will have level comparable to closed box...
Regards,
Etienne
The following is my understanding;
A loudspeakerdriver on its own is a minimum phase device. So the GD would be dependent of the frequency response. A high order slope at the highpass would make GD higher than would a shallow slope. The least GD and phase distortion would be from a wide bandwith device with shallow roll off on both ends.
For example, a three way boxed speaker with a peaky reflex tuning for the woofer and a lowpass (crossover to the mid) close to the box Fs for the same driver would cause "high GD". Use a closed box (acoustic suspension) with a shallower roll of but the same Fs and move the crossover (the lowpass for the woofer) up a bit and make it shallower and things would improve.
In a OB dipole or in a cardioid it's the same way. The bandwith of the driver(s) and the shape of the HP and LP roll off determines the GD.
A speaker driver and its filters is a minimum phase device. Several such units coupled together in a system (a multiway speaker) may or may not be a minimum phase system.. all depending on the type of filters.
/Peter
Regarding the reply by Earl Geddes
You raise so many issues. Where to begin....?
Re: working in audio: You are correct and I do not work in consumer electronics-audio. I have had the privilege of being able to work in auditory science since 1982. Most of it has been basic research in human hearing capabilities and processes. But I have also been involved in applied research regarding human systems interfaces etc.
Re: level of science: What you are saying is correct for some forms of commercial audio, but not all areas. Certainly the development of audio coding & storage (codecs etc), auditory prosthetics, concert hall acoustics, and some loudspeaker work have all used the level of measurement and analysis that I am referring to. Whether this will also be usurped in the future for a more "convenient" level of work, I am not sure. I can not predict the future
I understand that in some forms of commercial audio there are pressures that not easily compatible with a more through strategy on testing etc.
Re: distortion: That is a funny and frequently misused term (as you know). I agree that most folks only conceive of a non-linearity as being steady state harmonic distortion (heck, it easy to measure). I also agree that, at some point, further lowering the level of simple harmonic distortion will not yield further gains in perceptual satisfaction. There has been enough said about that. But I think we are probably both in agreement that there are many, many other forms of nonlinearities and when those forms are not steady state they are less easily measured and probably not very well understood (physically or perceptually).
I also realize you have a practical motivation about whether a physical measure (which many would not understand anyway) has a commercial consequence. You are running a business and there are some very concrete realities that go into your work.
However different systems do sound "different" although they may measure similarly. The nature of that difference, is ultimately quantifiable. Sometimes it is easy and sometimes it is not so easy.
So I think there is still an important relation to be understood between these physical anomalies (distortion in its many forms, dispersion, response etc, etc) and the consequent perception (audibility, detectability, accuracy etc etc). I suspect we are not in a terrible disagreement on this point.
I am going to stop here. Spring has come to New England and my lawn mower needs some attention.
-Tom
You raise so many issues. Where to begin....?
Re: working in audio: You are correct and I do not work in consumer electronics-audio. I have had the privilege of being able to work in auditory science since 1982. Most of it has been basic research in human hearing capabilities and processes. But I have also been involved in applied research regarding human systems interfaces etc.
Re: level of science: What you are saying is correct for some forms of commercial audio, but not all areas. Certainly the development of audio coding & storage (codecs etc), auditory prosthetics, concert hall acoustics, and some loudspeaker work have all used the level of measurement and analysis that I am referring to. Whether this will also be usurped in the future for a more "convenient" level of work, I am not sure. I can not predict the future
I understand that in some forms of commercial audio there are pressures that not easily compatible with a more through strategy on testing etc.
Re: distortion: That is a funny and frequently misused term (as you know). I agree that most folks only conceive of a non-linearity as being steady state harmonic distortion (heck, it easy to measure). I also agree that, at some point, further lowering the level of simple harmonic distortion will not yield further gains in perceptual satisfaction. There has been enough said about that. But I think we are probably both in agreement that there are many, many other forms of nonlinearities and when those forms are not steady state they are less easily measured and probably not very well understood (physically or perceptually).
I also realize you have a practical motivation about whether a physical measure (which many would not understand anyway) has a commercial consequence. You are running a business and there are some very concrete realities that go into your work.
However different systems do sound "different" although they may measure similarly. The nature of that difference, is ultimately quantifiable. Sometimes it is easy and sometimes it is not so easy.
So I think there is still an important relation to be understood between these physical anomalies (distortion in its many forms, dispersion, response etc, etc) and the consequent perception (audibility, detectability, accuracy etc etc). I suspect we are not in a terrible disagreement on this point.
I am going to stop here. Spring has come to New England and my lawn mower needs some attention.
-Tom
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.