Home |
Search |
Today's Posts |
#1
|
|||
|
|||
DSP for loudspeaker distortion
I know DSP's are available for loudspeaker frequency response
compensation, typically combined with a non-time-coherent "room correction" function. Is anything similar available for treating loudspeaker distortion? (actually whole-of-system distortion, but measured as what comes out of the speaker and hopefully dominated by the speaker's distortion contribution). I don't mean a research tool for speaker designers. I mean something that can be used in the home or studio to reduce distortion from existing systems. Stand-alone or PC-based. I imagine something that fits in the audio chain between the sound source and the amplifier. It might generate its own test signals and analyse the speaker output through a microphone. It could treat harmonic distortion by mapping a number of harmonics of measured distortion against signal amplitude against signal frequency. It could then generate a compensating map of out-of-phase harmonics. Then when playing music through the system, it could apply the compensating map real-time to the incoming music signal. Of course it would be sensible to combine such a function in the same DSP as is used for frequency response compensation. Am I dreaming, or describing a real product? Grant |
#3
|
|||
|
|||
DSP for loudspeaker distortion
In article ,
Ethan Winer ethan at ethanwiner dot com wrote: Grant, Great question, and really great answers from Tony. You are absolutely correct that loudspeaker distortion is very important (and rarely specified!) - it's much more important than many folks realize. Here's my take: First, the notion that DSP can compensate for room acoustics problems is a myth. EQ and other electronics can help a little, in a few special cases. But the biggest room problem is a series of many peaks and dips in the low frequency response. These peaks and dips change depending on where the speakers are placed, and where you are in the room. This simply cannot be compensated for by electronics. For a similar reason - all the stuff Tony said - DSP can correct only some of the causes of speaker distortion. Not so. It is certainly possible, though not necessarily easy, to deconvolve the room response for a selected listening region. This HAS been done successfully though, as I said, it is neither simple and easy to derive the deconvolution function, nor is it simple to implement, often requiring quite of bit of horsepower. One way to reduce loudspeaker distortion is to include the loudspeaker's voice coil in the power amp's negative feedback loop. Mackie does this in their HR series speakers, as do some other manufacturers. This allows the amplifier to compensate for incorrect cone movement directly, with much more control than is possible with an external DSP unit. No, it can ONLY reduce those non-linearities that are accurately reflected in the non-linear behavior of the Dl product of the driver. It completely ignores mechanical non-linearities that do no manifest themselves in the electrical domain. -- | Dick Pierce | | Professional Audio Development | | 1-781/826-4953 Voice and FAX | | | |
#4
|
|||
|
|||
DSP for loudspeaker distortion
On Fri, 5 Sep 2003 11:01:36 -0400, "Ethan Winer" ethan at ethanwiner
dot com wrote: First, the notion that DSP can compensate for room acoustics problems is a myth. Tell that to Behringer, TACT and Meridian.................. EQ and other electronics can help a little, in a few special cases. But the biggest room problem is a series of many peaks and dips in the low frequency response. These peaks and dips change depending on where the speakers are placed, and where you are in the room. This simply cannot be compensated for by electronics. Of course it can - you just need a microphone placed at the listening position. Heck, B&O (of all people!) just launched a speaker which contains its own microphone, and adjusts its bass response depending where it's placed in the room. -- Stewart Pinkerton | Music is Art - Audio is Engineering |
#5
|
|||
|
|||
DSP for loudspeaker distortion
Dick,
It is certainly possible, though not necessarily easy, to deconvolve the room response for a selected listening region. It really isn't possible. The problem is the many peaks and dips caused by reflections off the room boundaries. Those dips can be very deep, and they are position dependent. So even if you have enough amplifier and speaker capability to overcome a 15 dB dip at 75 Hz, as soon as you apply that correction you'll have a terrible peak a foot or two away. Maybe you could make that work if you mix while sitting in a dentist chair with your head clamped into position... it can ONLY reduce those non-linearities that are accurately reflected in the non-linear behavior of the Dl product of the driver. I don't know what "D1" is, but you are correct that including the voice coil in the feedback loop cannot overcome every possible source of distortion. But if you can hit the main sources and reduce the distortion from 5% to less than 1%, (I'm just guessing at those figures) I'd say that's a worthwhile goal. --Ethan |
#6
|
|||
|
|||
DSP for loudspeaker distortion
In article ,
Ethan Winer ethan at ethanwiner dot com wrote: Dick, It is certainly possible, though not necessarily easy, to deconvolve the room response for a selected listening region. It really isn't possible. The problem is the many peaks and dips caused by reflections off the room boundaries. Those dips can be very deep, and they are position dependent. So even if you have enough amplifier and speaker capability to overcome a 15 dB dip at 75 Hz, as soon as you apply that correction you'll have a terrible peak a foot or two away. False, what you are claiming is a physical impossibility. If you have a dip at 75 Hz, the CLOSEST a peak could be is 1/4 wavelength away. You have to go through 90 degrees of phase rotation for this to occur. Further, the only way that a "15 dB dip could have a correspondong "terrible peak" is if the absorbtion coefficient of the room boundaries is damned near 0 (it isn't) and if ALL the energy is concentrated into the primary nodes (it isn't). Indeed, is HAS been done QUITE successfully in just the range of frequencies you're claiming that it's impossible in. It ain't cheap, and it's computationally quite expensive, tobe sure, but it has been done. Maybe you could make that work if you mix while sitting in a dentist chair with your head clamped into position... Maybe you could actually study the physics of the situation a bit more and actually SEE what REALLY happens in real rooms. Even in the worst case scenario you describe, the area over which the correction is valid is quite a bit larger than you claim, encompassing a volume at least 15 times your claim. it can ONLY reduce those non-linearities that are accurately reflected in the non-linear behavior of the Dl product of the driver. I don't know what "D1" is, but you are correct that including the voice coil in the feedback loop cannot overcome every possible source of distortion. Sorry, type, it's Bl (as in Bee-el), the electromagnetic transduction parameter of the motor system. But if you can hit the main sources and reduce the distortion from 5% to less than 1%, (I'm just guessing at those figures) I'd say that's a worthwhile goal. Well, yes, you are guessing at the figures, and the numbers that result give a woefully inadequate picture of the situation. For the most part, the distortion produced by the non-linear behavior of the electrical and mechanical system is excurion dependent. Right away that tells us what's important and what's not. Below the system resonance (usually corresponding to system cutoff), since the system is stiffness controlled, the suspension non-linearities and the magnetic non-linearites together play the dominat role in determining the production of spurious components due to non-linearity. THat may not seem important, but consider that much of the resulting distortion components wind up above cutoff. Above resonance, in the mass-controlled region, the magnetic system non-linearities dominate, BUT, since the excursion goes as the inverse square of the frequency, it becomes a fairly irrelevant once you get up into the region where much of the real energy is being produced. There, you have non-linearities in the diaphragm itself, and those simply are NOT reflected at all in the electrical domain and are thus non-correctable by your scheme. -- | Dick Pierce | | Professional Audio Development | | 1-781/826-4953 Voice and FAX | | | |
#7
|
|||
|
|||
DSP for loudspeaker distortion
Richard,
If you have a dip at 75 Hz, the CLOSEST a peak could be is 1/4 wavelength away. Not so at all, though maybe I could have chosen my words better. By "peak" I meant the enormous boost caused by the applied EQ, not a natural peak due to the wave in the room. An acoustic null in a room is very sharp - both in frequency and the location where the level is at a minimum. Moving just a few inches away from the null spot yields a huge difference in level. So boosting a low frequency by 15 dB will cause a big increase in that frequency nearby in the room where the naturally occurring null is not so deep. If you've never experimented with low frequency sine waves in a room - and it sounds like you haven't - I suggest you play 80 Hz (or any low frequency your speakers can reproduce cleanly) and walk toward and away from the rear wall. At a point 1/4 wavelength from every boundary there is a deep null. (Since the floor and ceiling are also boundaries and they all combine, you may have to stand up tall or crouch down to find the place where the null is most complete.) Even with a decent amount of bass trapping installed you can easily hear and measure these nulls. But especially in a room without bass traps the effect is very obvious. Indeed, is HAS been done QUITE successfully No need to shout. Maybe if you define "quite successfully" as "it seemed to help a little." Seriously, besides the response/position problem, and the inability to eliminate echoes and ringing, yet another room problem you cannot solve with DSP is excessive reverb at low frequencies. This is one of the biggest problems for mixing engineers in small untreated rooms because it makes bass instruments sound muddy with the notes ill-defined. There's no way can you reduce reverb time with DSP. those simply are NOT reflected at all in the electrical domain and are thus non-correctable by your scheme. Hey, it's not my scheme! I'm just reporting what an EE friend said Mackie does. You gave a lot of explanations, but it would have been even better had you included real world figures of speaker distortion and how much it can be improved by various methods. Without any supporting data it sounds like you're guessing as much as I was. :-) you are hypothesizing more and more pathologically absurd scenarios. No need for name calling either. Please let's keep the discussion professional and courteous. How many listening rooms suffer from flutter echoes and ringing of a sufficient Q to cause the problems you describe? Careful, I HAVE measured a number of rooms. Have you? Yes, acoustic treatment is my business and I have measured lots of rooms. As soon as you get even one "boing" caused by echoes between parallel surfaces, you have exceeded what can be controlled electronically. This is especially a problem with the now-common (and inept) practice of treating home theaters by leaving the upper walls completely bare. So as soon as a gun goes off in an action movie, how could DSP leave the sound of the bullet intact but prevent the sound from bouncing off the walls? The only practical solution for all of these problems - peaks and dips caused by acoustic interference, excessive reverb, flutter echo and ringing - is broadband acoustic treatment. Even if it is possible to at least reduce room effects a little, by your own admission (earlier) it requires huge resources and thus huge expense. So why even bother to pursue that angle, when acoustic treatment is so much simpler, less expensive, and does a much better job? --Ethan |
#8
|
|||
|
|||
DSP for loudspeaker distortion
On Fri, 5 Sep 2003 14:37:57 -0400, "Ethan Winer" ethan at ethanwiner
dot com wrote: Stewart, Tell that to Behringer, TACT and Meridian ... you just need a microphone placed at the listening position. Heck, B&O (of all people!) just launched a speaker which contains its own microphone See my answer to Dick. It's not possible unless you don't mind being strapped into a chair with your head clamped in place so it can't move even one inch. If that's acceptable then okay, I concede. :-) Dick already answered that point, which simply is not true. Meridian and TACT have offered excellent room-correction for several years, and works very effectively over movements of several feet, as should be obvious from simple consideration of the wavelengths involved. Two other room problems that cannot be compensated for with DSP are flutter echoes and ringing. Yes, you can make a really deep and really narrow notch at the ring frequencies. But the tones are still there and still sustain after the sound source stops. And any legitimate musical content at that frequency will be removed. And there's really no way to get rid of echoes using electronics. Agreed, except that *serious* flutter echoes and ringing are very rare in domestic rooms (as opposed to commercial concrete shells), and in any event, it's not hard to deal with those by choosing suitable furnishings. You can drop the Q of such echoes to negilgible values by the simple hanging of a couple of wall rugs, and by carpeting the floor. -- Stewart Pinkerton | Music is Art - Audio is Engineering |
#9
|
|||
|
|||
DSP for loudspeaker distortion
Stewart,
Meridian and TACT have offered excellent room-correction for several years How do you define excellent? More to the point, how can any electronic device reduce reverb time or eliminate echoes? I had heard of the Meridian system and I just looked up their PDF article again. Since they offer several graphs as proof the system works, should I be suspicious that none of the graph legends are readable? I had never heard of TACT so I looked them up too, but found nothing meaningful in the way of explanations. Maybe I'm biased against an electronic solution because I manufacture and sell high performance acoustic treatment. But the idea of using DSP to control the three common and fundamental room problems - low frequency peaks and dips, echo and reverb, and ringing - makes no sense. How can you possibly get rid of even a little echo with DSP? Okay, I suppose you could generate an equal and opposite echo, and time it to arrive at the listening position. But again, as soon as you move your head an inch or two the sound is much worse. And what a kludge that is compared to the standard proven solutions. Agreed, except that *serious* flutter echoes and ringing are very rare in domestic rooms I deal with smallish recording studios all the time, and I assure you that after low frequency peaks and dips, echoes and ringing are the second most common complaint. You can drop the Q of such echoes to negilgible values by the simple hanging of a couple of wall rugs, and by carpeting the floor. Which is my whole point. Why bother with the complexity of DSP when you can do a much better job with conventional treatment? I didn't notice prices on either of those other sites, but I assume the cost for those DSP boxes is much higher than conventional treatment, yes? --Ethan |
#10
|
|||
|
|||
DSP for loudspeaker distortion
On Sat, 6 Sep 2003 06:51:01 -0400, "Ethan Winer" ethan at ethanwiner
dot com wrote: Richard, If you have a dip at 75 Hz, the CLOSEST a peak could be is 1/4 wavelength away. Not so at all, though maybe I could have chosen my words better. By "peak" I meant the enormous boost caused by the applied EQ, not a natural peak due to the wave in the room. An acoustic null in a room is very sharp - both in frequency and the location where the level is at a minimum. Actually no, it isn't in any halfway reasonable domestic environment. Moving just a few inches away from the null spot yields a huge difference in level. So boosting a low frequency by 15 dB will cause a big increase in that frequency nearby in the room where the naturally occurring null is not so deep. Could we have that again, in English please? If you've never experimented with low frequency sine waves in a room - and it sounds like you haven't - I suggest you play 80 Hz (or any low frequency your speakers can reproduce cleanly) and walk toward and away from the rear wall. At a point 1/4 wavelength from every boundary there is a deep null. (Since the floor and ceiling are also boundaries and they all combine, you may have to stand up tall or crouch down to find the place where the null is most complete.) Even with a decent amount of bass trapping installed you can easily hear and measure these nulls. But especially in a room without bass traps the effect is very obvious. Firstly, I would *strongly* suggest that you avoid intimating that Dick Pierce has not 'been there and done that' regarding loudspeaker or room measurements. Secondly, what you are claiming simply does not happen except in the most pathologically horrendous rooms (think cube). -- Stewart Pinkerton | Music is Art - Audio is Engineering |
#11
|
|||
|
|||
DSP for loudspeaker distortion
On Sat, 6 Sep 2003 13:45:38 -0400, "Ethan Winer" ethan at ethanwiner
dot com wrote: Stewart, Meridian and TACT have offered excellent room-correction for several years How do you define excellent? Able to correct for room resonance effects and loudspeaker response errors. More to the point, how can any electronic device reduce reverb time or eliminate echoes? It can't, but these are easily corrected by other means. I had heard of the Meridian system and I just looked up their PDF article again. Since they offer several graphs as proof the system works, should I be suspicious that none of the graph legends are readable? No, since the system is effective in reality. I had never heard of TACT so I looked them up too, but found nothing meaningful in the way of explanations. Maybe I'm biased against an electronic solution because I manufacture and sell high performance acoustic treatment. Oh, could there perhaps be a conflict of interest here? :-) But the idea of using DSP to control the three common and fundamental room problems - low frequency peaks and dips, echo and reverb, and ringing - makes no sense. Actually, it makes perfect sense, at least for the first one. The other two (which are actually the same thing, if you understand acoustics) are very easily controlled without resorting to overpriced 'audiophile' solutions. How can you possibly get rid of even a little echo with DSP? You can't, but you can with a $20 rug hung on the wall or thrown on the floor.......... Okay, I suppose you could generate an equal and opposite echo, and time it to arrive at the listening position. But again, as soon as you move your head an inch or two the sound is much worse. And what a kludge that is compared to the standard proven solutions. That's absolute garbage. DSP solutions work over several feet, as should be obvious if you consider the basic physics of the situation. Agreed, except that *serious* flutter echoes and ringing are very rare in domestic rooms I deal with smallish recording studios all the time, and I assure you that after low frequency peaks and dips, echoes and ringing are the second most common complaint. And this relates to domestic situations, exactly how? You can drop the Q of such echoes to negilgible values by the simple hanging of a couple of wall rugs, and by carpeting the floor. Which is my whole point. Why bother with the complexity of DSP when you can do a much better job with conventional treatment? I didn't notice prices on either of those other sites, but I assume the cost for those DSP boxes is much higher than conventional treatment, yes? Certainly, but they don't attempt to deal with slap echo, which can be controlled for a few tens of dollars by the intelligent use of floor rugs and wall drapes, to the taste of the owner. Now, what do you charge for your 'audiophile' room treatments, and what do they look like? Ethan, you're a troll attempting to push your room treatments on people who don't know any better. This is probably not the best place to make your sales pitch. -- Stewart Pinkerton | Music is Art - Audio is Engineering |
#12
|
|||
|
|||
DSP for loudspeaker distortion
Both Ethan and Dick Pierce know exactly what they are talking about. Ethan
is a legend in the acoustic treatment community. Dick is a legend in the Loudspeaker and Bass Cabinet reviewing, educating community. They are both looking at this question their very different perspectives and are arguing because their views of acoustic reproduction are very different. I have seen excellent uses of DSP that blew me away. Dick has too. In fact he has tested and reviewed them. I have seen excellent uses of bass traps, Helm-Holtz resonators, Acoustic Foam and Acoustic diffusors that blew me away. Ethan has tested these systems and designs very competent acoustic solutions. He writes excellent articles for DIY'ers and frugal professionals. I think there are things that both solutions can provide that the other cannot. DSP cannot correct EVERY negative aspect of an acoustic space, nor can acoustic treatment. Let's just have some peace and return to discussing this and stop picking fights. - FLINT |
#13
|
|||
|
|||
DSP for loudspeaker distortion
In article ,
Stewart Pinkerton wrote: How can you possibly get rid of even a little echo with DSP? You can't, but you can with a $20 rug hung on the wall or thrown on the floor.......... Except at the low frequencies (below a few hundred hertz) which are most problematic due to the near omni-polar response of most speakers resulting in a lower ratio of direct sound to the reverberant field. This is made worse by the relatively close spacing of the equal loudness curves. -- a href="http://www.poohsticks.org/drew/"Home Page/a The Congress shall assemble at least once in every Year, and such Meeting shall be on the first Monday in December, unless they shall by Law appoint a different Day. |
#14
|
|||
|
|||
DSP for loudspeaker distortion
Ethan Winer wrote:
Stewart, Tell that to Behringer, TACT and Meridian ... you just need a microphone placed at the listening position. Heck, B&O (of all people!) just launched a speaker which contains its own microphone See my answer to Dick. It's not possible unless you don't mind being strapped into a chair with your head clamped in place so it can't move even one inch. If that's acceptable then okay, I concede. :-) Two other room problems that cannot be compensated for with DSP are flutter echoes and ringing. Yes, you can make a really deep and really narrow notch at the ring frequencies. But the tones are still there and still sustain after the sound source stops. And any legitimate musical content at that frequency will be removed. And there's really no way to get rid of echoes using electronics. Echoes, flutters, and ringing are all effects that can be modeled perfectly as a linear system. Within certain constraints (minimum phase, well-conditioned), they can be removed. The concept is called deconvolution or inverse filtering. One of those constraints is the listener position, as you have stated. Of course that constraint is dependent on the frequencies at which the correction is being applied, so if you're talking about low-frequency EQ, the "head lock" problem is not present. The other serious constraint I didn't hear you mention is that the equalization is only valid for a single listener. -- % Randy Yates % "...the answer lies within your soul %% Fuquay-Varina, NC % 'cause no one knows which side %%% 919-577-9882 % the coin will fall." %%%% % 'Big Wheels', *Out of the Blue*, ELO http://home.earthlink.net/~yatescr |
#16
|
|||
|
|||
DSP for loudspeaker distortion
On Sun, 07 Sep 2003 04:28:45 GMT, Randy Yates wrote:
Ethan Winer wrote: Stewart, Tell that to Behringer, TACT and Meridian ... you just need a microphone placed at the listening position. Heck, B&O (of all people!) just launched a speaker which contains its own microphone See my answer to Dick. It's not possible unless you don't mind being strapped into a chair with your head clamped in place so it can't move even one inch. If that's acceptable then okay, I concede. :-) Two other room problems that cannot be compensated for with DSP are flutter echoes and ringing. Yes, you can make a really deep and really narrow notch at the ring frequencies. But the tones are still there and still sustain after the sound source stops. And any legitimate musical content at that frequency will be removed. And there's really no way to get rid of echoes using electronics. Echoes, flutters, and ringing are all effects that can be modeled perfectly as a linear system. Within certain constraints (minimum phase, well-conditioned), they can be removed. The concept is called deconvolution or inverse filtering. Interesting, I didn't know that. I would still tend to use acoustic treatment to remedy reflection effects, combined with DSP room correction to sort out eigentones. One of those constraints is the listener position, as you have stated. Of course that constraint is dependent on the frequencies at which the correction is being applied, so if you're talking about low-frequency EQ, the "head lock" problem is not present. The other serious constraint I didn't hear you mention is that the equalization is only valid for a single listener. Ah the loneliness of the true audiophile! :-) -- Stewart Pinkerton | Music is Art - Audio is Engineering |
#17
|
|||
|
|||
DSP for loudspeaker distortion
"flint" wrote in message ... Both Ethan and Dick Pierce know exactly what they are talking about. Ethan is a legend in the acoustic treatment community. Dick is a legend in the Loudspeaker and Bass Cabinet reviewing, educating community. They are both looking at this question their very different perspectives and are arguing because their views of acoustic reproduction are very different. I have seen excellent uses of DSP that blew me away. Dick has too. In fact he has tested and reviewed them. I have seen excellent uses of bass traps, Helm-Holtz resonators, Acoustic Foam and Acoustic diffusors that blew me away. Ethan has tested these systems and designs very competent acoustic solutions. He writes excellent articles for DIY'ers and frugal professionals. I think there are things that both solutions can provide that the other cannot. DSP cannot correct EVERY negative aspect of an acoustic space, nor can acoustic treatment. FWIW I'll second this! I admire both guys for the good stuff they do in their respective areas of audio. Left to my own choices, I follow what Ethan says about room acoustics first, and when you have the room with the acoustics of your dreams or merely the best you can do because of other constraints, then start thinking about the DSP-based room enhancements. |
#18
|
|||
|
|||
DSP for loudspeaker distortion
Dick,
Your grasp of acoustical physics is, well, underwhelming. I freely admit that I am not a physicist nor a mathematician. But I am technically minded, and I have plenty of practical hands-on experience with this stuff. Please, for the gathered crowd, show how high the Q of such a resonance must be to make your assertion true. You are confusing the Q of the frequency response at a given location with the size of that location. If you find a spot in the room where the response is down 15 dB at 100 Hz, a high Q will make the dip less severe at 101 Hz. But even with a low Q the depth of the null can be very deep - theoretically infinite - and the size of the "zone" where the null is that deep can be very small. If you move away just a little from where the null is deepest the level will come back up quite a bit. EW: There's no way can you reduce reverb time with DSP. RP: Wanna bet? Again, do you understand the principal of deconvolution? You are correct that deconvolution involves complex math I do not understand. But please explain in plain English the mechanism by which DSP can remove echoes and reverb. I can see how DSP could create equal and opposite echoes, and maybe even equal and opposite reverb (if you have a Cray). But how could this be applied in a typical room and not suffer from the "move your head one inch" restriction? This is the crux of the matter: How is a complex solution having severe positional restrictions sensible, when the problem can be solved better and much less expensively using bass traps and fiberglass panels? And you STILL did not answer the question: how many listening rooms in domestic situations suffer from the problems you describe? All of them! If the response in the room is linear, and it contains finite energy, it is correctable. Yeah, okay, theoretically for a tiny zone. But not only is the zone tiny, for everyone outside the zone the correction will make things many times worse. --Ethan |
#19
|
|||
|
|||
DSP for loudspeaker distortion
Stewart,
EW: An acoustic null in a room is very sharp - both in frequency and the location where the level is at a minimum. SP: Actually no, it isn't in any halfway reasonable domestic environment. Yes, it really is. And it's so easy to prove I don't know why we're arguing about it. Take a sine wave oscillator or test tone CD and play it through your system. Or a friend's system. Or a system in ANY room without adequate acoustic treatment. Now walk around the room very slowly listening for drops in level. You will immediately see that at any reasonable frequency (say, 80 Hz to 4 KHz) there are many locations that have severe drops in level, and those locations are extremely restricted in size. Moving your head even one inch at mid/high frequencies is enough to go from full level to a nearly complete cancellation. At low frequencies the size of the null zone is larger, but still very small - if you move to the adjacent seat on the couch the level can change many dB. EW: So boosting a low frequency by 15 dB will cause a big increase in that frequency nearby in the room where the naturally occurring null is not so deep. SP: Could we have that again, in English please? Let's say at the listening position you measure a 15 dB dip at 80 Hz. So you set the room EQ to boost 80 Hz by 15 dB. Now in that precise spot the level at 80 Hz is back to normal. But now in the adjacent seat the level is 8 dB higher than normal because the dip was only 7 dB there initially, but that location is also receiving the 15 dB of EQ boost. I would *strongly* suggest that you avoid intimating that Dick Pierce has not 'been there and done that' regarding loudspeaker or room measurements. I had never heard of Richard Pierce before meeting him here, and I have no idea what he's done or what he knows. I gladly discuss these issues and assess all points on their merit, without regard to who said them. what you are claiming simply does not happen except in the most pathologically horrendous rooms (think cube). You are very wrong. The peaks and dips I describe happen in every home-sized room with exactly the severity I described. And they happen at ALL frequencies, not just those related to the room dimensions. It's a surprisingly common misconception, even among acoustics people, that peaks and nulls occur only at modal frequencies. Please do as I suggest and play some sine waves and walk around. Also see this excellent article by studio designer Wes Lachot: www.recording.org/users/acoustics/waves_wl.html --Ethan |
#20
|
|||
|
|||
DSP for loudspeaker distortion
On Sun, 7 Sep 2003 11:38:28 -0400, "Ethan Winer" ethan at ethanwiner
dot com wrote: Dick, Your grasp of acoustical physics is, well, underwhelming. I freely admit that I am not a physicist nor a mathematician. But I am technically minded, and I have plenty of practical hands-on experience with this stuff. Please, for the gathered crowd, show how high the Q of such a resonance must be to make your assertion true. You are confusing the Q of the frequency response at a given location with the size of that location. If you find a spot in the room where the response is down 15 dB at 100 Hz, a high Q will make the dip less severe at 101 Hz. But even with a low Q the depth of the null can be very deep - theoretically infinite - and the size of the "zone" where the null is that deep can be very small. If you move away just a little from where the null is deepest the level will come back up quite a bit. You are confusing standing waves with bulk resonances. The change in level along a standing waves - in a and out of nulls - is governed by a Cosine law, as is normal with the amplitude of any two co-frequency signals sliding over each other. So the change in level between two points is not governed by an idea of a Q factor. The one aspect in which the room makes a difference is in the depth of the null. For an infinite null you need perfect reflectivity. For a deep null, you need good reflectivity, and a parallel wall area big enough to reflect essentially plane waves. It is in the Helmholtz mode that a room can have a Q factor, and that does not behave as you describe here. EW: There's no way can you reduce reverb time with DSP. RP: Wanna bet? Again, do you understand the principal of deconvolution? You are correct that deconvolution involves complex math I do not understand. But please explain in plain English the mechanism by which DSP can remove echoes and reverb. I can see how DSP could create equal and opposite echoes, and maybe even equal and opposite reverb (if you have a Cray). But how could this be applied in a typical room and not suffer from the "move your head one inch" restriction? This is the crux of the matter: How is a complex solution having severe positional restrictions sensible, when the problem can be solved better and much less expensively using bass traps and fiberglass panels? The room is simply a complex transmission line, with an amplitude and delay response. If it can be measured, then its conjugate can be synthesised and placed in series with the signal. It seems counter-intuitive, but this does indeed result in the elimination of all the unflatnesses and echoes that have been measured. How big a volume of the room you can treat in this way depends on how high a frequency you need to go to, and how bad the room is to begin with. Certainly up to a couple of hundred Hz shouldn't be a problem, and if you can't sort the room out otherwise above that frequency, you are really in trouble. And you STILL did not answer the question: how many listening rooms in domestic situations suffer from the problems you describe? All of them! If the response in the room is linear, and it contains finite energy, it is correctable. Yeah, okay, theoretically for a tiny zone. But not only is the zone tiny, for everyone outside the zone the correction will make things many times worse. --Ethan d _____________________________ http://www.pearce.uk.com |
#21
|
|||
|
|||
DSP for loudspeaker distortion
In article ,
Ethan Winer ethan at ethanwiner dot com wrote: Dick, Your grasp of acoustical physics is, well, underwhelming. I freely admit that I am not a physicist nor a mathematician. But I am technically minded, and I have plenty of practical hands-on experience with this stuff. Please, for the gathered crowd, show how high the Q of such a resonance must be to make your assertion true. You are confusing the Q of the frequency response at a given location with the size of that location. Sir, there is n such thing as "the Q of the frequency response." The question I asked is perfectly valid If you find a spot in the room where the response is down 15 dB at 100 Hz, a high Q will make the dip less severe at 101 Hz. But even with a low Q the depth of the null can be very deep - theoretically infinite - and the size of the "zone" where the null is that deep can be very small. If you move away just a little from where the null is deepest the level will come back up quite a bit. Sorry, sir, your statment is simply nonsensical mathematically., EW: There's no way can you reduce reverb time with DSP. RP: Wanna bet? Again, do you understand the principal of deconvolution? You are correct that deconvolution involves complex math I do not understand. But please explain in plain English the mechanism by which DSP can remove echoes and reverb. I can see how DSP could create equal and opposite echoes, and maybe even equal and opposite reverb (if you have a Cray). Actually, modern DSP chips are far more effcieicnt at this sort of thing than a general purpose Cray. Beyond that, you have just stated that what you first claimed impossible is in fact quite possible. But how could this be applied in a typical room and not suffer from the "move your head one inch" restriction? BECAUSE YOU SIMPLY CANNOT HAVE A THE LEVEL OF SUCH CHNAGE OVER THE SPACE OF AN INCH IF THE WAVELENGTH IS MANY FEET LONG. You have already admitted you do not understand "complex math." Are you admitting that you do not understand this as well? How is a complex solution having severe positional restrictions sensible, when the problem can be solved better and much less expensively using bass traps and fiberglass panels? That is NOT the crux of the issue: you have made a number of incorrect and nonsensical technical assertions, and simply failed to support them with anything other than reassertions If the response in the room is linear, and it contains finite energy, it is correctable. Yeah, okay, theoretically for a tiny zone. But not only is the zone tiny, for everyone outside the zone the correction will make things many times worse. Sir, you have now made this wrong statement at least a half dozen times, and if you make it another half dozen, it will STILL not get any less wrong. -- | Dick Pierce | | Professional Audio Development | | 1-781/826-4953 Voice and FAX | | | |
#22
|
|||
|
|||
DSP for loudspeaker distortion
In article ,
Ethan Winer ethan at ethanwiner dot com wrote: about it. Take a sine wave oscillator or test tone CD and play it through your system. Or a friend's system. Or a system in ANY room without adequate acoustic treatment. Now walk around the room very slowly listening for drops in level. You will immediately see that at any reasonable frequency (say, 80 Hz to 4 KHz) there are many locations that have severe drops in level, and those locations are extremely restricted in size. Moving your head even one inch at mid/high frequencies is enough to go from full level to a nearly complete cancellation. Mr. Winer, this is MOST dishonest of you. You first stated in your assertion that these nulles were extremely sharp at frequencies of 70 Hz. Now, all of a suddenm you have TOTALLY changed the scenario by stating that these phenomenon occur at "mid/high frequencies." I am sorry, but you've now gone from vagueness and imprecise claims to simpl,y changing your assertions in midstram. You have already admitted you do not understand the physics and methematics that are crucial to the topic, what more is there for you to admit? I had never heard of Richard Pierce before meeting him here, and I have no idea what he's done or what he knows. I gladly discuss these issues and assess all points on their merit, without regard to who said them. They are not my ideas, kinf sir, they are basic and fundamental concepts of acoustics going all the way back to the wave equation. YOu are arguing against that, and there's a couple of centuries of pretty well-founded science and evdience that suggests your assertions are way off the amrk. -- | Dick Pierce | | Professional Audio Development | | 1-781/826-4953 Voice and FAX | | | |
#23
|
|||
|
|||
DSP for loudspeaker distortion
Don,
the change in level between two points is not governed by an idea of a Q factor. But that's exactly what *I* said! That Q is a frequency bandwidth concept, versus the physical size of the null zone and how far away you have to move for the null to not be as deep. this does indeed result in the elimination of all the unflatnesses and echoes that have been measured. But the echo level and arrival time is different for every location in the room. Move one foot away and the arrival time changes by 1 millisecond. So how many milliseconds would you have the DSP use to remove echoes for all three seats on my living room couch? And how would the EQ boost that's applied for my seat affect the seats on either side? --Ethan |
#24
|
|||
|
|||
DSP for loudspeaker distortion
Thanks for the educational discussion on room correction. Much
appreciated. Returning to my original question about treating loudspeaker distortion with DSP, nobody mentioned whether there was a real product available, even if it only tackled one tyupe of distortion (e.g. harmonic distortion) and not all of the varied speaker distortions mentioned. Any comment? I would be quite satisfied to only treat one listening space in the room. The problems of catering for multiple listeners do not especially concern me. Although I noticed that pro EQ tools like SIA Smaart sometimes can average corrections over a number of listening positions, which may be useful for EQ work. But it may be a moot point. If distortions emitted from the speaker (irrespective of their origin in the audio reproduction chain) were being reduced, I thought this improvement would not be dependent on listening position. Is that right? Grant My original post: ================================================== ================== I know DSP's are available for loudspeaker frequency response compensation, typically combined with a non-time-coherent "room correction" function. Is anything similar available for treating loudspeaker distortion? (actually whole-of-system distortion, but measured as what comes out of the speaker and hopefully dominated by the speaker's distortion contribution). I don't mean a research tool for speaker designers. I mean something that can be used in the home or studio to reduce distortion from existing systems. Stand-alone or PC-based. I imagine something that fits in the audio chain between the sound source and the amplifier. It might generate its own test signals and analyse the speaker output through a microphone. It could treat harmonic distortion by mapping a number of harmonics of measured distortion against signal amplitude against signal frequency. It could then generate a compensating map of out-of-phase harmonics. Then when playing music through the system, it could apply the compensating map real-time to the incoming music signal. Of course it would be sensible to combine such a function in the same DSP as is used for frequency response compensation. Am I dreaming, or describing a real product? ================================================== ================== |
#25
|
|||
|
|||
DSP for loudspeaker distortion
On Sun, 7 Sep 2003 11:27:06 -0400, "Ethan Winer" ethan at ethanwiner
dot com wrote: Stewart, what do you charge for your 'audiophile' room treatments, and what do they look like? Ethan, you're a troll attempting to push your room treatments on people who don't know any better. This is probably not the best place to make your sales pitch. That's not true, and it's unfair to accuse me of that. What's your affiliation, BTW? I don't think that I was the one who said that, and I have no commercial interest in audio. You OTOH certainly do, so perhaps you are indeed just trying to 'knock' DSP room-correction. My take is pretty much the same as Arny's - find a good room, give it some basic acoustic treatment, and leave the LF stuff to DSP. I've been championing the importance of acoustic treatment since the late 1970s. I've been selling bass traps and broadband absorbers for less than a year. I jumped into this thread only because it irks me to see half-baked solutions touted as workable. The original poster accepted the claim that DSP is a reasonable way to cure the peaks and dips caused by acoustic interference, and that's what I addressed. Unfortunately for your argument, DSP in fact *is* a reasonable way to cure the peaks and dips caused by acoustic interference. -- Stewart Pinkerton | Music is Art - Audio is Engineering |
#26
|
|||
|
|||
DSP for loudspeaker distortion
On Sun, 7 Sep 2003 11:43:49 -0400, "Ethan Winer" ethan at ethanwiner
dot com wrote: Stewart, EW: An acoustic null in a room is very sharp - both in frequency and the location where the level is at a minimum. SP: Actually no, it isn't in any halfway reasonable domestic environment. Yes, it really is. And it's so easy to prove I don't know why we're arguing about it. Take a sine wave oscillator or test tone CD and play it through your system. Or a friend's system. Or a system in ANY room without adequate acoustic treatment. Now walk around the room very slowly listening for drops in level. You will immediately see that at any reasonable frequency (say, 80 Hz to 4 KHz) there are many locations that have severe drops in level, and those locations are extremely restricted in size. Moving your head even one inch at mid/high frequencies is enough to go from full level to a nearly complete cancellation. At low frequencies the size of the null zone is larger, but still very small - if you move to the adjacent seat on the couch the level can change many dB. Thgis is utter rubbish. Firstly, you will *never* exceed a 10dB difference in any normal domestic living room, secondly, moving your head an inch at 4kHz will certainly *not* achieve anything approaching a complete cancellation in any real-word room, and thirdly, at low frequencies (which I'll take to be less than 100Hz) you would have to move about three feet to go from reinforcement to partial cancellation - and even then there won't be 'many dB' of difference. As you say, this is trivially easy to prove with an oscillator and a microphone. EW: So boosting a low frequency by 15 dB will cause a big increase in that frequency nearby in the room where the naturally occurring null is not so deep. SP: Could we have that again, in English please? Let's say at the listening position you measure a 15 dB dip at 80 Hz. So you set the room EQ to boost 80 Hz by 15 dB. Now in that precise spot the level at 80 Hz is back to normal. But now in the adjacent seat the level is 8 dB higher than normal because the dip was only 7 dB there initially, but that location is also receiving the 15 dB of EQ boost. However, you will *never* find a 15dB null in any normal domestic living room. I would *strongly* suggest that you avoid intimating that Dick Pierce has not 'been there and done that' regarding loudspeaker or room measurements. I had never heard of Richard Pierce before meeting him here, and I have no idea what he's done or what he knows. I gladly discuss these issues and assess all points on their merit, without regard to who said them. Actually, you are doing no such thing in this thread, as Dick has pointed out on several occasions. You are ducking, diving and spraying wild assertions all over the place, none of which bear any relation to very well known acoustic principles. what you are claiming simply does not happen except in the most pathologically horrendous rooms (think cube). You are very wrong. The peaks and dips I describe happen in every home-sized room with exactly the severity I described. No, that is simply *not* the case. And they happen at ALL frequencies, not just those related to the room dimensions. Absolute garbage. These effects may be *noticeable* at a few tens of Hertz, but they rapidly avearage out by the lower midrange, except in the most pathologically unrealistic rooms, such as a concrete and glass commercial property. It's a surprisingly common misconception, even among acoustics people, that peaks and nulls occur only at modal frequencies. Please do as I suggest and play some sine waves and walk around. I've been there and done that on several occasions. You are talking rubbish. -- Stewart Pinkerton | Music is Art - Audio is Engineering |
#27
|
|||
|
|||
DSP for loudspeaker distortion
Stewart,
you will *never* find a 15dB null in any normal domestic living room. Please see my other post. No, that is simply *not* the case. Absolute garbage. You are talking rubbish. C'mon Stewart, tell us how you REALLY feel! :-) --Ethan |
#28
|
|||
|
|||
DSP for loudspeaker distortion
Grant,
nobody mentioned whether there was a real product available, even if it only tackled one tyupe of distortion The closest thing I'm aware of is the pre-distortion scheme for analog tape I already mentioned. Other than something like that, or what Mackie appears to do putting the speaker voice coil in the power amp's feedback loop, I'm not aware of any such commercial devices. Not that there isn't or couldn't be such a device. SIA Smaart sometimes can average corrections over a number of listening positions, which may be useful for EQ work. Averaging is really not as useful as you might think. This was discussed in detail recently in John Sayers' acoustics forum (www.johnlsayers.com/phpBB2/index.php, Studio Design section) where a similar "discussion" erupted over the validity of pink noise versus sine wave testing. The problem with pink noise is that it averages all frequencies within a third octave (usually) band. But within a single band there could be a deep null and also a big peak. So the averaged measurement shows a reassuringly flat response that *completely hides* response variations as large as 20 dB! --Ethan |
#29
|
|||
|
|||
DSP for loudspeaker distortion
"Stewart Pinkerton" wrote in message ... On Sun, 7 Sep 2003 11:43:49 -0400, "Ethan Winer" ethan at ethanwiner dot com wrote: Yes, it really is. And it's so easy to prove I don't know why we're arguing about it. Take a sine wave oscillator or test tone CD and play it through your system. Or a friend's system. Or a system in ANY room without adequate acoustic treatment. Now walk around the room very slowly listening for drops in level. You will immediately see that at any reasonable frequency (say, 80 Hz to 4 KHz) there are many locations that have severe drops in level, and those locations are extremely restricted in size. Moving your head even one inch at mid/high frequencies is enough to go from full level to a nearly complete cancellation. At low frequencies the size of the null zone is larger, but still very small - if you move to the adjacent seat on the couch the level can change many dB. Been there, done that, it works. I'm not sure I've seen this work well as high as 4 KHz, but I've definitely seen strong effects at 400 Hz. This is utter rubbish. Firstly, you will *never* exceed a 10dB difference in any normal domestic living room, secondly, moving your head an inch at 4kHz will certainly *not* achieve anything approaching a complete cancellation in any real-word room, and thirdly, at low frequencies (which I'll take to be less than 100Hz) you would have to move about three feet to go from reinforcement to partial cancellation - and even then there won't be 'many dB' of difference. Been there, done that and seen and heard the nulls. Not complete nulls, but 20-30 dB nulls, easy. It's a simple experiment since all you need is a pure sine wave source. As you say, this is trivially easy to prove with an oscillator and a microphone. Yes, and Ethan describes a relevant experiment in another post and I've done similar things. EW: So boosting a low frequency by 15 dB will cause a big increase in that frequency nearby in the room where the naturally occurring null is not so deep. SP: Could we have that again, in English please? Let's say at the listening position you measure a 15 dB dip at 80 Hz. So you set the room EQ to boost 80 Hz by 15 dB. Now in that precise spot the level at 80 Hz is back to normal. But now in the adjacent seat the level is 8 dB higher than normal because the dip was only 7 dB there initially, but that location is also receiving the 15 dB of EQ boost. However, you will *never* find a 15dB null in any normal domestic living room. I wish. If you measure with pure tones, you will find nulls deeper than that. However, the relevance of a single null is questionable. This is a major reason why swept tones, warble tones, and noise and other means are popularly used for measuring the frequency response of loudspeakers in reverberant rooms. I would *strongly* suggest that you avoid intimating that Dick Pierce has not 'been there and done that' regarding loudspeaker or room measurements. I can only presume that people are agreeing loudly or missing the meaning of words and sentences. I had never heard of Richard Pierce before meeting him here, and I have no idea what he's done or what he knows. I gladly discuss these issues and assess all points on their merit, without regard to who said them. IME Dick Pierce is a trustworthy source of audio knowledge. But, so is Ethan Winer. I've learned from life that it's not always correct to presume that when two people appear to disagree, that one of them is absolutely wrong. FWIW George Middius, Ed Shain and David Weil among others used to play this childish game on RAO with JJ and I, and as wrong as it was then, its at least that wrong here, now. Let's stick to the facts, eh? |
#30
|
|||
|
|||
DSP for loudspeaker distortion
Arny,
the relevance of a single null is questionable. This is a major reason why swept tones, warble tones, and noise and other means are popularly used See my other post about the suitability of sine waves for testing. In fact, I have lately come to the conclusion that sine waves are perhaps the ONLY worthwhile test if you really want the brutal truth about a room. The problem with pink noise and swept sine waves, unless they are very narrow and swept very slowly, is too many different frequencies are all lumped together in one measured band. As you know, just a couple of Hertz is enough to make a large difference in the response. I can measure 15 dB nulls in my room all day long, yet a pink noise test shows my room as being flat within just a dB or two across most of the range. I love showing the pink noise graph to my friends, but I know it's pure fiction! :-) Let's stick to the facts, eh? Yes. Thanks for keeping that in the forefront. --Ethan |
#31
|
|||
|
|||
DSP for loudspeaker distortion
Arny,
AFAIK current "revealed truth" among those who *know* is that 1/12 octave is warranted. I doubt even 1/12th octave is narrow enough. Here's a figure from my Acoustics FAQ showing what my partner and I measured in his 10x16 foot control room using sine waves: www.ethanwiner.com/response.gif Look at the region starting at 125 Hz in the top half. If you measure first at 160 Hz and then 1/12th octave above that (169.5 Hz) you'll have gone from a deep null to its adjacent peak. So at those frequencies in that room you'll get the results expected. But in a room where the peaks and dips start just a few Hertz lower or higher, readings at those same frequencies will measure flat because one reading is on the upside and the other is on the downside. And that completely hides the 10 dB difference that's actually present. --Ethan |
#32
|
|||
|
|||
DSP for loudspeaker distortion
"Ethan Winer" ethan at ethanwiner dot com wrote in message ... Arny, Please get back when you do some sine-wave measurements of your own. You might be surprised! Yes, and I'm still hoping some of the other naysayers will try it too. It sure did get quiet around here, all of a sudden... ;-) |
#33
|
|||
|
|||
DSP for loudspeaker distortion
Ethan Winer wrote:
[...] And again, even if the frequency response errors can be corrected with EQ, that does nothing for the lack of clarity on notes played by bass instruments dues to excessive LF reverb. And again, the process implemented using DSP is not EQ in the sense you are thinking. But you are certainly free to believe whatever you wish. -- % Randy Yates % "...the answer lies within your soul %% Fuquay-Varina, NC % 'cause no one knows which side %%% 919-577-9882 % the coin will fall." %%%% % 'Big Wheels', *Out of the Blue*, ELO http://home.earthlink.net/~yatescr |
#34
|
|||
|
|||
DSP for loudspeaker distortion
Randy,
the process implemented using DSP is not EQ in the sense you are thinking. But you are certainly free to believe whatever you wish. Then talk to me! :-) If I'm missing something important, explain so I'll learn. Don't just hit and run. And explain How and What room correcting DSP can do, not just say "It fixes the problems" because that explains nothing. If you were to ask me how bass traps and absorbers solve room problems, I could easily explain it in great detail! Room nulls are very deep and they change depth and frequency over very small distances. So I can't see how electronics can compensate unless you put your head in a vise, which no sane person would accept. Last week the group argued whether nulls are really that deep and change 15-20 dB over a few inches. The premise being that DSP doesn't need to correct for such serious anomalies. I showed that all rooms do in fact vary that much over tiny distances, so now the burden is back on the DSP promoters to explain how such severe nulls can be corrected for all listeners on the couch. Severe nulls are just the start. Much harder to solve are echoes and reverb. How can DSP remove echoes and reverb that are floating around a room for more than a pinpoint sized area? This is the crux of it. Explain that sufficiently and you'll be well on your way toward convincing me. Heck, you don't have to explain anything at all - just tell me where I can audition one of these things in a lousy sounding room so I can switch the DSP on and off and hear the change for myself. My ears are very attuned to poor rooms and how they sound after adding bass traps and other treatment. If a DSP solution can really do the same thing, I'll change my opinion on a dime the moment I hear it. --Ethan |
#35
|
|||
|
|||
DSP for loudspeaker distortion
Ethan Winer wrote:
Randy, the process implemented using DSP is not EQ in the sense you are thinking. But you are certainly free to believe whatever you wish. Then talk to me! :-) If I'm missing something important, explain so I'll learn. Don't just hit and run. And explain How and What room correcting DSP can do, not just say "It fixes the problems" because that explains nothing. If you were to ask me how bass traps and absorbers solve room problems, I could easily explain it in great detail! Is it not possible that a topic may be complex enough that one or two usenet news postings won't even come close to describing it in enough depth that one really understands it? This type of topic is usually covered in a one-semester engineering course called "Linear System Theory" or "Fourier Transforms" or some such. The relevent concepts involved are the definition of a linear, time-invariant system, the description of such systems by their impulse response, the operation of such systems through the operation of convolution (in the time domain), and various properties such systems (e.g., causality, stability, invertibility, etc.), and various properties and elementary results of Fourier transforms (both discrete-time and continuous-time). If you really want to understand at this level, I suggest you read the relevent parts of the book "Signals and Systems" by Oppenheim et al. There is a more recent equivalent by this author, "Discrete-Time Signal Processing," but I much prefer the older book. Also, I have stated in no uncertain terms that such a system is going to be listener-position dependent. Why do you insist on asking me to describe a solution which is not? Ethan, I don't mean to be rude or excessively inaccessible, but really - why must everything be elementary? It ain't. Besides, I DID attempt to lead you to the path of understanding by throwing out a few of the concepts (e.g., convolution), but it (understandably) didn't stick. -- % Randy Yates % "...the answer lies within your soul %% Fuquay-Varina, NC % 'cause no one knows which side %%% 919-577-9882 % the coin will fall." %%%% % 'Big Wheels', *Out of the Blue*, ELO http://home.earthlink.net/~yatescr |
#36
|
|||
|
|||
DSP for loudspeaker distortion
Y'know, several days ago, someone made a comment about how this
thread has suddenly gone silent, as if it was somehow a sign that the other side had somehow lost. In article , Rusty Boudreaux wrote: "Ethan Winer" ethan at ethanwiner dot com wrote in message ... If I'm missing something important, explain so I'll learn. Don't just hit and run. And explain How and What room correcting DSP can do, Like Dick Pierce and others have said, deconvolution. Like Randy Yates said, and like many others have said, if Mr. Winer would just take the time to explore beyond his narrow viewpoint. Measure the impulse response of a room, invert it, and convolve with the incoming audio. If done properly with enough horsepower you get perfect results at ANY point in the room. To "deconvolve" the acoustics of a room: y(n)=hroom(z)*x(n) http://www.isip.msstate.edu/publicat...e_4773/lecture s/current/lecture_19/ Here's some recorded examples of correcting audio from a very echoy lounge. The program was written by an undergrad EE student for a class...so this is far from perfect...but it will show how reverb and echos are reduced. http://www.cs.caltech.edu/~ashman/ee...oject107c.html More explanations: http://www.clsp.jhu.edu/seminars/abs...997/soong.html http://www.cs.tut.fi/~temmi/MSc_Thesis/Abstract.html http://www.iop.org/EJ/abstract/0957-0233/12/1/311 http://www.google.com/search?q=audio+deconvolution So I can't see how electronics can compensate unless you put you head in a vise, which no sane person would accept. Yes, it has become quite apparent that you can't see this, and that's where your problem is. YOu have, again no insult intended, a VERY limited view of this topic, and you can't see beyond so thus you assume there is nothing beyond it to see. distances, so now the burden is back on the DSP promoters to explain how such severe nulls can be corrected for all listeners on the couch. No. The burden is on you to research and understand deconvolution. Indeed, the "DSP promoters" HAVE already done that, and there is extensive description of this in the literature should you care to engage in the effort. People like myself are NOT going to do it in this forum because it is a complex topic that also requires some willingness on the other participant to do the work. Severe nulls are just the start. Much harder to solve are echoes and reverb. Nope, just deconvolve the room. Done. How can DSP remove echoes and reverb that are floating around a room for more than a pinpoint sized area? Deconvolution. Mr. Winer, does the word "deconvolution" sound at all familiar? Now, to see if you HAVE done the work necessary for you to understand the process, see if you can answer this question: deconvolution techniques can be made to work quite well for small rooms, but a heretofore undiscussed property makes the solution for very large acoustic spaces (such as cathedrals, auditoria, stadia) much more difficult. And, as a hint, it's not, per se, a CORRECTION problem by a measurement problem that is based on the assumption of stationary signals. If you can understand even the question, you're halfway there. Now, go do the homework. -- | Dick Pierce | | Professional Audio Development | | 1-781/826-4953 Voice and FAX | | | |
#37
|
|||
|
|||
DSP for loudspeaker distortion
Randy,
Is it not possible that a topic may be complex enough that one or two usenet news postings won't even come close to describing it in enough depth that one really understands it? All I'm asking for is an explanation of the concepts, not for you to hold my hand through a bunch of higher math. When someone truly understands an advanced process, they should be able to explain the basics in plain English. For example, if the DSP you promote works be creating equal and opposite echoes, then say so. Or say whatever it does do. Over the years I've been guided by people a lot smarter than me, and the one common trait all the real experts shared was being able to explain complex matters in simple terms. I now strive to do that myself, to help others who know less than me. See my mini-book "The Hardware Tutor" where I explain the basics of analog and digital electronics in mechanical terms with no math. It's about 1/3 of the way down the page on my Articles page: www.ethanwiner.com/articles.html If you really want to understand at this level No, I really don't. All I want is a plain-English explanation. Or just tell me where I can hear such a system at work. Do stores demo these things? I have stated in no uncertain terms that such a system is going to be listener-position dependent. Why do you insist on asking me to describe a solution which is not? Because if this system is dependant on the listenener remaining in one very small place, then it is not viable! It may work in theory, but it's destined to remain more of a technical curiosity than a practical solution. I have asked repeatedly how wide an area can be realistically included in the correction zone, and all I've gotten back is "it depends." That would be an acceptable answer if it also included a range. Like, "We've been able to perfectly correct an area 3x8 feet in a room that's at least 20x30 feet." Or whatever. Remember, this all started when I objected to the notion that DSP is a PRACTICAL solution to room problems. Then y'all insulted me, and claimed it's physically impossible for a null to vary by 15 dB across four inches. I hope you're not still disputing that! Don't get me wrong, I love science and math and theory. And I'm here to learn too. But we all must be able to disinguish a cool lab experiment from a practical solution. --Ethan |
#38
|
|||
|
|||
DSP for loudspeaker distortion
Richard D Pierce wrote: Y'know, several days ago, someone made a comment about how this thread has suddenly gone silent, as if it was somehow a sign that the other side had somehow lost. In article , Rusty Boudreaux wrote: "Ethan Winer" ethan at ethanwiner dot com wrote in message ... If I'm missing something important, explain so I'll learn. Don't just hit and run. And explain How and What room correcting DSP can do, Like Dick Pierce and others have said, deconvolution. Like Randy Yates said, and like many others have said, if Mr. Winer would just take the time to explore beyond his narrow viewpoint. Measure the impulse response of a room, invert it, and convolve with the incoming audio. If done properly with enough horsepower you get perfect results at ANY point in the room. To "deconvolve" the acoustics of a room: y(n)=hroom(z)*x(n) http://www.isip.msstate.edu/publicat...e_4773/lecture s/current/lecture_19/ Here's some recorded examples of correcting audio from a very echoy lounge. The program was written by an undergrad EE student for a class...so this is far from perfect...but it will show how reverb and echos are reduced. http://www.cs.caltech.edu/~ashman/ee...oject107c.html More explanations: http://www.clsp.jhu.edu/seminars/abs...997/soong.html http://www.cs.tut.fi/~temmi/MSc_Thesis/Abstract.html http://www.iop.org/EJ/abstract/0957-0233/12/1/311 http://www.google.com/search?q=audio+deconvolution So I can't see how electronics can compensate unless you put you head in a vise, which no sane person would accept. Yes, it has become quite apparent that you can't see this, and that's where your problem is. YOu have, again no insult intended, a VERY limited view of this topic, and you can't see beyond so thus you assume there is nothing beyond it to see. distances, so now the burden is back on the DSP promoters to explain how such severe nulls can be corrected for all listeners on the couch. No. The burden is on you to research and understand deconvolution. Indeed, the "DSP promoters" HAVE already done that, and there is extensive description of this in the literature should you care to engage in the effort. People like myself are NOT going to do it in this forum because it is a complex topic that also requires some willingness on the other participant to do the work. Severe nulls are just the start. Much harder to solve are echoes and reverb. Nope, just deconvolve the room. Done. This sounds very interesting. How is this done in practice? Are there products available that would allow an audiophile to do this and about how much would it cost? George Deliz |
#39
|
|||
|
|||
DSP for loudspeaker distortion
In article ,
Ethan Winer ethan at ethanwiner dot com wrote: Richard, as if it was somehow a sign that the other side had somehow lost. My goal is not to argue and "take sides." And I've said repeatedly that I'm here to learn. But are you still maintaining that peaks and nulls cannot vary by 15 dB across a span of four inches? Now, Mr. Winer, I did not say that, did I? What I said was in direct refutation of YOUR claim of such variations at 70 Hz over a span of one inch. -- | Dick Pierce | | Professional Audio Development | | 1-781/826-4953 Voice and FAX | | | |
#40
|
|||
|
|||
DSP for loudspeaker distortion
In article ,
Ethan Winer ethan at ethanwiner dot com wrote: Thanks. The math is way over my head, but I did glean this nugget from one of those referenced pages: "However, in the above single- and multi-channel approaches, the requirement that the room impulse response needs to be identified first renders the solution not readily applicable to a real situation." What that means is that there is no such thing as a one-size- fits-all solution. Since the impulse response contains all the information needed to fully characterize the linear response of ANY system, and since each system (i.e., room) has its own unique impulse response, you need to measure the impulse response for that room. -- | Dick Pierce | | Professional Audio Development | | 1-781/826-4953 Voice and FAX | | | |
Reply |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Forum | |||
Digital high frequency distortion | High End Audio | |||
FS: HEWLETT-PACKARD 334A Distortion Analyzer - Late Model | Pro Audio | |||
FS: HEWLETT-PACKARD 334A Distortion Analyzer with Warrenty | Pro Audio | |||
Pioneer Clipping and Distortion was:DEH-P840MP, infinity kappa 693.5i and kappa 50.5cs component. | Car Audio | |||
FS: HEWLETT-PACKARD 334A Distortion Analyzer | Pro Audio |