jfischer@sco.COM (Jonathan A. Fischer) (03/23/91)
Something that's really caught my interest lately (I suppose it was after reading a review of the Meridian D6000 "digital speaker") is the possibility of the following scenario: You buy a good speaker with no glaring flaws. Its frequency response is pretty good, varying +/- a couple of dB over yer basic 40-ish to 20K Hz range. Its phase accuracy varies +/- <n> degrees over the spectrum (whatever's typical). So you buy a programmable DSP "package," containing the DSP unit (which also performs as a frequency generator), and a mike or Sound Pressure Level meter. You set up the SPL meter in your listening spot, press the "setup" button on the DSP unit, and it commences to send frequency sweeps through your sound system, reads the levels and the phase response. Finally, using these variables, it sets up a digital equalization + phase doctoring DSP program which will transform your sound system, no matter what your room's or your speaker's acoustical properties, into one with a completely flat frequency response curve, and with zero phase shift across the entire spectrum. Is this a pipe dream or is it feasible? -- Jonathan A. Fischer SCO Canada, Inc. jfischer@scocan.sco.COM Toronto, Ontario, Canada Usenet's first law of flamodynamics: For every opinion, there is an equal and opposite counter-opinion.
jroth@allvax.enet.dec.com (Jim Roth) (03/23/91)
In article <1991Mar22.171203.8665@sco.COM>, jfischer@sco.COM (Jonathan A. Fischer) writes... > > So you buy a programmable DSP "package," containing the DSP >unit (which also performs as a frequency generator), and a mike or >Sound Pressure Level meter. You set up the SPL meter in your [ ... and optimize your speaker response with it ... ] > Is this a pipe dream or is it feasible? These experiments have been done from time to time and have been reported in the JAES; more recent work has equalized the reverberant sound field as well as the direct arrival sound (a weighted optimization of the two...) A recent paper is from someone at KEF, make what you will of that. It's certainly well within the capability of off the shelf DSP technology, and could probably make otherwise good but not identical sounding speakers sound virtually the same if a good LEDE listening room was used. I'd try an experiment myself if I had an audio workstation handy... - Jim
mcmahan@netcom.COM (Dave Mc Mahan) (03/24/91)
In a previous article, jfischer@sco.COM (Jonathan A. Fischer) writes: > > Something that's really caught my interest lately (I suppose >it was after reading a review of the Meridian D6000 "digital speaker") >is the possibility of the following scenario: > > You buy a good speaker with no glaring flaws. Its frequency >response is pretty good, varying +/- a couple of dB over yer basic >40-ish to 20K Hz range. Its phase accuracy varies +/- <n> degrees over >the spectrum (whatever's typical). > > So you buy a programmable DSP "package," containing the DSP >unit (which also performs as a frequency generator), and a mike or >Sound Pressure Level meter. You set up the SPL meter in your >listening spot, press the "setup" button on the DSP unit, and it >commences to send frequency sweeps through your sound system, reads >the levels and the phase response. Finally, using these variables, it >sets up a digital equalization + phase doctoring DSP program which >will transform your sound system, no matter what your room's or your >speaker's acoustical properties, into one with a completely flat >frequency response curve, and with zero phase shift across the entire >spectrum. > > Is this a pipe dream or is it feasible? Sure, it's feasible. How much do you want to spend and how much correction do you think you will need? The only other thing that you have to remember is that your system of amplifiers, etc. must be capable of time invariant response. That means that if you turn up the volume a bit or the amplifier temperature drifts a bit due to a warmer room, your parameters won't drift. This system wouldn't work very well if every time you adjust the volume you get a totally different response curve. The other thing you could never do is move plants or furniture within the room and not re-tune the system. You won't be able to ever get down to a sub-woofer response without having a speaker that is capable of such low bass as well. You could compensate a current system, but don't look at getting concert hall performance with mail-order speaker systems and amplifiers. >Jonathan A. Fischer SCO Canada, Inc. >jfischer@scocan.sco.COM Toronto, Ontario, Canada -dave -- Dave McMahan mcmahan@netcom.com {apple,amdahl,claris}!netcom!mcmahan
gt0869a@prism.gatech.EDU (WATERS,CLYDE GORDON) (03/24/91)
In article <1991Mar22.171203.8665@sco.COM> jfischer@sco.COM (Jonathan A. Fischer) writes: some deleted... >sets up a digital equalization + phase doctoring DSP program which >will transform your sound system, no matter what your room's or your >speaker's acoustical properties, into one with a completely flat >frequency response curve, and with zero phase shift across the entire >spectrum. > > Is this a pipe dream or is it feasible? >-- If anyone has any relevant information on this, _please_email me a copy of it too. I am interested in digitally altering not just the whole system response, but the individual parts. Suppose, by chance, you have a tweeter that roll off too soon on the low end (maybe you're trying to use 6db slopes) However, the resonsnce behavior severely complicates attempts to alter the response. My guess would be that some system of higher order(like differenrial equations-analogy) would be required to solve this problem. However, a digital system, done properly could "numerically solve" the problem... I am interested in not only sloutions, but will entertain any theories anyone has on this subject. I know that there are devices similar to the ones Mr. Fische mentioned in development, but have no relevant data. This type of design seems to offer a chance of "sidestepping" some of the shortfalls of current audio technology.Until the "better speaker" comes along, maybe this could"bridge the gap"I know this has little hope in solving transient distortion problems, but it seems that by removing a couple of variables from the solution equations the job would be a lot easier(ie, if the frequency and phase response were correct)... Neat subject to think about, isn't it? Gordon. -- WATERS,CLYDE GORDON Georgia Institute of Technology, Atlanta Georgia, 30332 uucp: ...!{decvax,hplabs,ncar,purdue,rutgers}!gatech!prism!gt0869a Internet: gt0869a@prism.gatech.edu
cwpjr@cbnewse.att.com (clyde.w.jr.phillips) (03/25/91)
In article <1991Mar22.171203.8665@sco.COM>, jfischer@sco.COM (Jonathan A. Fischer) writes: > > Something that's really caught my interest lately (I suppose > it was after reading a review of the Meridian D6000 "digital speaker") > is the possibility of the following scenario: > > You buy a good speaker with no glaring flaws. Its frequency > response is pretty good, varying +/- a couple of dB over yer basic > 40-ish to 20K Hz range. Its phase accuracy varies +/- <n> degrees over > the spectrum (whatever's typical). > > So you buy a programmable DSP "package," containing the DSP > unit (which also performs as a frequency generator), and a mike or > Sound Pressure Level meter. You set up the SPL meter in your > listening spot, press the "setup" button on the DSP unit, and it > commences to send frequency sweeps through your sound system, reads > the levels and the phase response. Finally, using these variables, it > sets up a digital equalization + phase doctoring DSP program which > will transform your sound system, no matter what your room's or your > speaker's acoustical properties, into one with a completely flat > frequency response curve, and with zero phase shift across the entire > spectrum. > > Is this a pipe dream or is it feasible? > -- > Jonathan A. Fischer SCO Canada, Inc. > jfischer@scocan.sco.COM Toronto, Ontario, Canada > Usenet's first law of flamodynamics: > For every opinion, there is an equal and opposite counter-opinion. Jon, I had the same dream about 2-3 years ago, when I first got interested in DSP's. Yeah I think this is possible. A good high end audio product. Another variation I thought of is to incorporate a listener profile and do the same sort of thing to get the speaker responce flat to a individuals ears... Amazing how dreams can come true, isn't it? Clyde
eric@cinnet.com (Eric Bardes) (03/27/91)
It is a nifty idea. Run enough DSP to counteract defects in the speaker and room, BUT ... What about the response curve of the microphone? I think a much more likely idea is some serious acoustic computer modeling of the microphone, can't do it for real because of real world problems, so the DSP knows those limitations too. I give it three to seven years depending on consumer demand. Eric Bardes
exspes@gdr.bath.ac.uk (P E Smee) (03/27/91)
In article <1991Mar22.171203.8665@sco.COM> jfischer@sco.COM (Jonathan A. Fischer) writes: > So you buy a programmable DSP "package," containing the DSP >unit (which also performs as a frequency generator), and a mike or >Sound Pressure Level meter. You set up the SPL meter in your >listening spot, press the "setup" button on the DSP unit, and it >commences to send frequency sweeps through your sound system, reads >the levels and the phase response. Finally, using these variables, it >sets up a digital equalization + phase doctoring DSP program which >will transform your sound system, no matter what your room's or your >speaker's acoustical properties, into one with a completely flat >frequency response curve, and with zero phase shift across the entire >spectrum. Marantz are said to be working on such a box for the home market. They've even demonstrated a prototype, which will 'flatten' naked cone drivers in a normal room. Of course (to answer someone else's comments) it will only flatten them over the frequency range that the laws of physics, and the basic limitations of the components, allow. Knowing audio companies, 'prototype' in this context probably means that they've got the user interface of the box completed, but with a fairly hunky computer simulating the internal workings. Still, with the miracles of modern chippery, should get there eventually. Would surprise me if other companies aren't researching this as well. We'd note also that since it uses digital processing, there will be lots of folk who won't like it, no matter how good it is. -- Paul Smee, Computing Service, University of Bristol, Bristol BS8 1UD, UK P.Smee@bristol.ac.uk - ..!uunet!ukc!bsmail!p.smee - Tel +44 272 303132
DCROWE@GTRI01.GATECH.EDU (03/27/91)
This is certainly possible, within the performance limits of the DSP (and, of
course, the software would have to protect the hardware from being "corrected"
beyond it's physical capabilities).
The really intersting thing is to go beyond the suggestion to correct amplitude
and phase response, and to model and predict the form of harmonic and
intermodulation distortion, so that a cancelling signal can be added as a
function of the current program material. Perhaps the CD source could be read
ahead by a second laser to give the DSP time to calculate the correction.
Devon Crowe / "There are more linear functions in
dcrowe@gtri01.gatech.edu / Physics than in Nature"
/ _Norbert Weiner.wilf@sce.carleton.ca (Wilf Leblanc) (03/28/91)
eric@cinnet.com (Eric Bardes) writes: >It is a nifty idea. Run enough DSP to counteract defects in the speaker >and room, BUT ... What about the response curve of the microphone? >[stuff deleted] One point that I haven't seen mentioned: Sure, the response might be great at the microphone, but horrible a couple of wavelengths away. Realistically, the DSP could counteract the (minor) defects in the speaker, but it may be difficult to counteract the room acoustics. Besides, do you really want to counteract the room acoustics ?? Don't we want to transform the room into a concert hall (or some such desirable place) ?? If we do all the DSP in the world to counteract the non-flat frequency response of the speaker and the room acoustics, it may end up sounding rather poor. -- Wilf LeBlanc, Carleton University, Systems & Comp. Eng. Ottawa, Canada, K1S 5B6 Internet: wilf@sce.carleton.ca UUCP: ...!uunet!mitel!cunews!sce!wilf Oh, cruel fate! Why do you mock me so! (H. Simpson)
hillman@newsserver.sfu.ca (Steve Hillman) (03/28/91)
Rockford Fosgate has already brought out just such a project. For a brief description of it, look in the "New Products" section of April's Car Audio & Electronics (note that this product is for car stereos, but would probably work equally well in a home.) But it ain't cheap! -- Steve "Skillman" Hillman "Everyone generalizes" hillman@whistler.sfu.ca skillman@tz.wimsey.bc.ca
lstowell@pyrnova.pyramid.com (Lon Stowell) (03/28/91)
In article <1991Mar27.042821.14392@cinnet.com> eric@cinnet.com (Eric Bardes) writes: > >It is a nifty idea. Run enough DSP to counteract defects in the speaker >and room, BUT ... What about the response curve of the microphone? > >I think a much more likely idea is some serious acoustic computer modeling of >the microphone, can't do it for real because of real world problems, so the >DSP knows those limitations too. > >I give it three to seven years depending on consumer demand. > I'll have to agree.... All that would be needed is an "open" interface into the Yamaha DSP processors....your specific room environment is fed as a correction signal into their existing DSP processing. You would need a lot of (computer or otherwise) knobs, etc. to allow tuning for personal tastes to override any automation....most people don't really much care for a "flat" response generated by the automated equalizers of today...and I doubt if any computer would be able to satisfy everyone's ears. You can play around with this today if you have DSP hardware and a fast enough PC...all that is really needed is to take the algorithms and move it to VLSI to get the parts count down and speed up. Anyone got a few megabucks?
dpm@msc.edu (David P. Mottaz) (03/29/91)
What fun to watch this thing go berserk when a cat runs through the room upsetting the "waves", or if the phone rings. It will need a "I Have A Cold And My Head Is Stuffed Up" mode. For a little extra you get a great feature, the Air Conditioner Fan Noise Compensation Chip. The acoustical properties are much different in the summer, when you have a glass bottle of beer in your hand(that's right, High Frequency Deflection) than in the winter, with you wearing a thick sound-absorbing sweater. -Dave/dpm@msc.edu/Minnesota Supercomputer Center :-) 8*> I-](RoboCop smile)
mitchemt@silver.ucs.indiana.edu (Terry Mitchem) (03/29/91)
In article <3783@uc.msc.umn.edu> dpm@msc.edu (David P. Mottaz) writes: >What fun to watch this thing go berserk when a cat runs through the >room upsetting the "waves", or if the phone rings. It will need a "I >Have A Cold And My Head Is Stuffed Up" mode. For a little extra you get >a great feature, the Air Conditioner Fan Noise Compensation Chip. The >acoustical properties are much different in the summer, when you have a >glass bottle of beer in your hand(that's right, High Frequency Deflection) >than in the winter, with you wearing a thick sound-absorbing sweater. > Fine. So we hook it up to a Cray III in order to process the signal in real time :-) In all seriousness, you would need some serious compute power to accomplish this. It might nit be worth the money. Later, Terry
wilf@sce.carleton.ca (Wilf Leblanc) (03/29/91)
dpm@msc.edu (David P. Mottaz) writes: >What fun to watch this thing go berserk when a cat runs through the >room upsetting the "waves", or if the phone rings. It will need a "I >Have A Cold And My Head Is Stuffed Up" mode. For a little extra you get >a great feature, the Air Conditioner Fan Noise Compensation Chip. The >acoustical properties are much different in the summer, when you have a >glass bottle of beer in your hand(that's right, High Frequency Deflection) >than in the winter, with you wearing a thick sound-absorbing sweater. Exactly, although I think you need a ;-), in there somewhere. However, there are two issues here: 1. Improving the sound due to poor speaker characteristics; 2. Improving the sound due to poor room acoustics. Item 1 is easy using DSP, and need not be adaptive. Item 2 is very difficult, because of the issues you mentioned. I would hate to see what happens when you are using a system which compensates (adaptively) for poor room acoustics in a very small room and someone opens the door ;-). However, think of the $ people are going to make selling these advanced (?) features. If people are going to spend money, there just has to be many modes such as you mention: 1. Cat mode; 2. I have a cold mode; 3. Air conditioner on mode; 4. I'm drinkin' a beer mode; 5. Canada mode, (i.e. I got my sweater on mode). However, you forget many modes: 1. I've got big ears mode; 2. I'm old, so could you boost the high frequencies a bit mode; 3. I've got a thick forehead mode (also known as great bone conduction mode); 4. And of course, quit adapting, you're driving me crazy mode. >-Dave/dpm@msc.edu/Minnesota Supercomputer Center :-) 8*> I-](RoboCop smile) -- Wilf LeBlanc, Carleton University, Systems & Comp. Eng. Ottawa, Canada, K1S 5B6 Internet: wilf@sce.carleton.ca UUCP: ...!uunet!mitel!cunews!sce!wilf Oh, cruel fate! Why do you mock me so! (H. Simpson)
chrisc@gold.gvg.tek.com (Chris Christensen) (03/30/91)
In article <1991Mar27.042821.14392@cinnet.com> eric@cinnet.com (Eric Bardes) writes: > >It is a nifty idea. Run enough DSP to counteract defects in the speaker >and room, BUT ... What about the response curve of the microphone? > Do us (recording engineers) a favor and don't try and correct microphone response curves! I guess the real question is how would you compensate for the Microphones response? Even the minimalists may use two or three different types of microphones for Classical recordings. Just my 2 cents worth. Asbestos suit on.! Chris Christensen
ulfl@kuling.UUCP (Ulf Lagerstedt) (03/31/91)
In article <1991Mar22.171203.8665@sco.COM> jfischer@sco.COM (Jonathan A. Fischer) writes: > So you buy a programmable DSP "package," containing the DSP >unit (which also performs as a frequency generator), and a mike or >Sound Pressure Level meter. You set up the SPL meter in your >listening spot, press the "setup" button on the DSP unit, and it >commences to send frequency sweeps through your sound system, reads >the levels and the phase response. Finally, using these variables, it >sets up a digital equalization + phase doctoring DSP program which >will transform your sound system, no matter what your room's or your >speaker's acoustical properties, into one with a completely flat >frequency response curve, and with zero phase shift across the entire >spectrum. It seems to me that this procedure would not be enough. The imperfectness of your speaker/system might not correspond simply to single frequencies, but instead, say, a loud bass pulse X following a midrange pulse Y. I suppose you would have to analyse each musical piece individually. Furthermore, since your speaker is less than perfect, the corrections you apply will need counter-corrections and counter-counter corrections. It is not obvious that subsequent results will be closer to the original, or even that a given speaker of good quality is theoretically capable of emitting a certain signal given *any* possible input. I recall the motional feedback (MFB) speakers made by Philips in the early 1970's, which had a built-in amp and a piezo crystal placed on the moving bass cone. The crystal would sense the acceleration, and information of the motion of the bass cone would be compared to the input signal and corrected by the amp. The speakers had unusually good low bass response, but that was about it. I don't think the speakers were commercially successful, since they were a bit expensive. Besides, most people already had paid for their own power amps. As a side note, the model I examined had a rather dangerous mains connection for the built-in amp. One speaker was connected to the wall socket, and the other speaker to the first one with a male-to-male cable... -- "Television - a medium. So called because Ulf Lagerstedt it is neither rare nor well done" ZYX Sweden AB ulf@zyx.se
gt0869a@prism.gatech.EDU (WATERS,CLYDE GORDON) (04/05/91)
In article <3783@uc.msc.umn.edu> dpm@msc.edu (David P. Mottaz) writes: >What fun to watch this thing go berserk when a cat runs through the >room upsetting the "waves", or if the phone rings. It will need a "I > >-Dave/dpm@msc.edu/Minnesota Supercomputer Center :-) 8*> I-](RoboCop smile) What I meant when I posted earlier on speaker compensation was NOT to try to compensate for room effects, just for the speaker itself (measure once, preferably anechoic, near field) and leave it alone. I do not personally believe room compensation should be tried in any electronic form (except maybe for Nelson Pass' (I think) noise cancellation/ standing wave cancellation scheme) Passive room comp (deadening, etc) can have good effect, but adaptive room eq is too big a can of worms for me :-) Any further comments welcomed on speaker compensation. Thanks Gordon. -- WATERS,CLYDE GORDON Georgia Institute of Technology, Atlanta Georgia, 30332 uucp: ...!{decvax,hplabs,ncar,purdue,rutgers}!gatech!prism!gt0869a Internet: gt0869a@prism.gatech.edu