The trick is not to go backward, sacrificing spectral flatness and first-arrival integrity on the altar of directivity control. This, I feel, is the problem with so-called "bipole" loudspeakers, add-on rear tweeters, or other types of multi-directional speakers.
Comparing the spectra of direct-arrival versus the sum of all reflections (using the FFT method described above) keeps the designer honest when playing around with the dispersion and frequency response of additional drivers.
The reason I mentioned going to the trouble of clipping off the direct-arrival from the long-duration 100 mSec FFT measurement is the 100 mSec measurement is dominated by the energy of the direct-arrival wave if it is left in the measurement, which then masks the spectra of the summed reflections. (This is also true of RTA measurements; in domestic environments, the direct-arrival wave dominates the measurement.) Once you move the starting gate by 3 to 5 mSec, removing as much as possible of the direct-arrival wave, the spectral shape of the room reflections becomes much more obvious.
If you want to match the spectra of the summed reflections with the direct-arrival, you have to look at each one by itself, and then adjust the radiation pattern and spectra of the added drivers so the spectra of the two are a closer match to each other. Measurements showing direct-arrival spectra, overlaid with the spectra of summed reflections, are not easy to come by ... but you can make your own easily enough.
In addition, I also strongly feel that the stray radiation from the added drivers, no matter how small, should be time-matched to the forward-facing driver with a tolerance of 100 microseconds or less, so first-arrival transient integrity is not impaired. Time-alignment is usually ignored in the majority of multi-directional loudspeakers.
The absence of spectra-matching, along with time misalignment, are the two dominant problems with conventional multi-directional loudspeakers. By lucky accident, well-designed (and accurately equalized) dipoles have good spectral matching and time performance, which gives them the "coherent" sound they are known for. I feel that multi-directional speakers, if designed with care, can offer the same spatial qualities as dipoles.
Comparing the spectra of direct-arrival versus the sum of all reflections (using the FFT method described above) keeps the designer honest when playing around with the dispersion and frequency response of additional drivers.
The reason I mentioned going to the trouble of clipping off the direct-arrival from the long-duration 100 mSec FFT measurement is the 100 mSec measurement is dominated by the energy of the direct-arrival wave if it is left in the measurement, which then masks the spectra of the summed reflections. (This is also true of RTA measurements; in domestic environments, the direct-arrival wave dominates the measurement.) Once you move the starting gate by 3 to 5 mSec, removing as much as possible of the direct-arrival wave, the spectral shape of the room reflections becomes much more obvious.
If you want to match the spectra of the summed reflections with the direct-arrival, you have to look at each one by itself, and then adjust the radiation pattern and spectra of the added drivers so the spectra of the two are a closer match to each other. Measurements showing direct-arrival spectra, overlaid with the spectra of summed reflections, are not easy to come by ... but you can make your own easily enough.
In addition, I also strongly feel that the stray radiation from the added drivers, no matter how small, should be time-matched to the forward-facing driver with a tolerance of 100 microseconds or less, so first-arrival transient integrity is not impaired. Time-alignment is usually ignored in the majority of multi-directional loudspeakers.
The absence of spectra-matching, along with time misalignment, are the two dominant problems with conventional multi-directional loudspeakers. By lucky accident, well-designed (and accurately equalized) dipoles have good spectral matching and time performance, which gives them the "coherent" sound they are known for. I feel that multi-directional speakers, if designed with care, can offer the same spatial qualities as dipoles.
Last edited:
wish i had access to the equipment to see first hand what you describe in the entirety in 4110482.
what do you mean exactly by spectral matching? i assume coloration, timbre, etc.
No. Not subjective matching. This is using a conventional FFT measurement system, adjusting the time window to look at the frequency response of the direct-arrival, re-adjusting the FFT time window so you're only looking at the summed room reflections, and then comparing the two sets of frequency response measurements by overlaying them on the same graph (after adjusting the magnitudes).
Here's a measurement I made on the Ariel about 20 years ago with the MLSSA system, comparing the difference between leaving the first-arrival in the measurement versus removing it. The Ariel has very broad dispersion compared to other speakers of the day, so the two curves are fairly similar.
The "Beyond" project has narrower dispersion at HF than the Ariel, so I'm considering additional up-firing drivers. Since the purpose of the additional drivers is to get the room spectra to match the direct spectra, their response will need to be shaped in the crossover.
My usual method of loudspeaker design is getting the curves where I want them, then do small (1 dB or smaller) tweaks for subjectively flat response. If I can't reconcile measurements with subjective impressions of flatness, I've made a mistake somewhere, and will try and find where that might be.
Attachments
Last edited:
Change stuff in the crossover, whether it's passive, active, or DSP. Since the crossovers for the forward-facing drivers and the additional drivers are independent, they can have different response-shaping.
That's why I'm advocating independent measurements for the first-arrival sound and all subsequent room reflections; it makes it easier to adjust the two sets of crossovers. If the measurements are commingled, it's hard to tell what's going on with the room spectra.
Granted, listening rooms are different, but the substantial change in directivity index of most loudspeakers dominates what the room is going to do. The only way to escape the shift in DI is listen to the speaker outdoors ... which is where controlled-directivity sound reinforcement speakers come in.
I met an exhibitor at the RMAF who thought that all FR measurements were pointless and misleading, so he never measured his speakers. I don't agree, but then, I worked at Tektronix for nine years, so I'm interested in measurement technology. I'm a big believer in looking at the time, frequency, and spatial domains so it can be determined what the drivers (and enclosure) are doing in the physical world, and then correlating these measurements with perception.
Last edited:
Thank you very much, Lynn. I'm honored that an adaptation of something we did might find its way into the Great Beyond.
Just to give a bit of background for readers, Jim Romeyn invented the Late Ceiling Splash configuration, as he calls it. Basically it's an (imo rather ingenious) evolution of the controlled-pattern offset bipole configuration that I've been using. The rear array gets its path-length-induced time delay by taking the long route bounce off the ceiling.
"If anything, [the sound images] were more firmly located in space...." This improvement in "image density" (for lack of a better term) is not something I would have predicted. I'm usually more concerned with timbre, and much less interested in imaging per se, but it looks like Jim's configuration delivers wothwhile benefits there as well.
I've participated in discussions with people who believe that all the reverberant energy you should hear is already on the recording, and anything the room contributes by way of reflections is by definition "coloration" and will superimpose the room's signature on the original recording (both timbrally and spatially). I wish I could do that little a/b demonstration for these people, as it's arguably a demonstration of more room sound vs less room sound (with the additional room sound imo done pretty well in this case). And somewhat counter-intuitively, the imaging was improved rather than degraded.
The main idea is to get additional spectrally-correct energy into the reverberant field, but have it arrive after a fairly significant time delay. Our target is ten milliseconds (I got that from Earl Geddes).
The upfiring drivers cover the full range. The back of the cabinet acts as a "shadow" to block sidelobes from arriving too early at the listening position. Obviously it's imperfect in this respect, and long wavelengths are going to wrap around anyway, but apparently we're in the ballpark. Also, the upfiring HF array we use has some directional control of its own going on.
The reason for not placing the upfiring drivers on top of the enclosure are a) you would get unwanted early-arrival sidelobe energy and b) the path-length-induced delay for the ceiling reflection is reduced below our ten millisecond target.
The frequency response of the upfiring array is shaped to suit its role (there are right ways and wrong ways to add extra reverberant energy). Sorry, the specifics of that response shaping are proprietary. But here's a hint as to where the goalposts are: Lynn noted in a post quoted above that "the apparent spectral balance didn't change much" when the upfiring arrays were switched in and out. Take into account the fact that its local environment is significantly different from that of the front-firing array.
Last edited:
Thanks for chiming in, Duke. I'm glad that you and Jim Romeyn have gone ahead and actually built, fine-tuned, and put into production what was just an idle thought of mine. You and Jim did the heavy lifting and got there first, and I congratulate you on the obviously successful results.
Having plowed my way through Floyd Toole's book (which is a must-read for aspiring speaker designers), I have to disagree with Floyd's disparagement of two-channel stereo as a serious high-fidelity medium. The brutal fact is that more than 99% of the musically significant recordings made in the last fifty years are only available in 2-channel distribution, either on LP or CD.
Another odd thing about the book is the typical listening environment is depicted as a 13-seat home theater, with 3 seats in the front row, 5 in the middle row, and 5 in the back row. Is Floyd seriously proposing that people listen to music discs in a (very) high-end dedicated home theater room with a blank screen? Does anyone do this?
Anyone with the disposable income to build a dedicated home theater room, complete with 13 seats and the latest 1080P or 4K projector, isn't going to listen to music in the dark. The high-end audio market has split into two parts: audio for movies, and audio for music lovers. The best environment for movies ... a dimly lit, air-conditioned, windowless room with a large screen ... is not the best environment for music, where you want light and space, and a pleasant environment to socialize, share drinks and refreshments, and chat while you listen to music. It can be an intimate couple, a group of friends, a small party, or hifi nerds showing off the system, it's still a social experience.
Given that some people still enjoy listening to music in a social environment ... we haven't all been forced onto headphones yet ... and the dominant source of music is 2-channel, there are implications for loudspeaker design. To the best of the ability of the designer, the resulting loudspeaker must be optimized for 2-channel playback, if it's intended for music playback. The criteria for home theater may be different, and that's OK, since the requirements are not really the same.
As I see it, audiophiles and music-lovers love music for its own sake, and have large, sometimes very large, 2-channel music collections. Serious music lovers have many formats they can play: mono LP's, stereo LP's, CD's, high-res downloads on computers, and sometimes even reel-to-reel tape recorders. Yes, there are a handful of multichannel recordings in SACD, DVD-A, and download formats ... and they sound wonderful on a fully symmetric 5-channel system. I started out in audio as the inventor of the Shadow Vector quadraphonic decoder, and have a soft spot for really well-done ambient sound.
But I have to acknowledge the reality of the 2-channel catalogue, and honor the older mono catalogue as well. This is the finest music of all time, and it is the task of the playback system to do the best job possible with it. This is where paying attention to the ambient impression can yield a more satisfying audio system.
Having plowed my way through Floyd Toole's book (which is a must-read for aspiring speaker designers), I have to disagree with Floyd's disparagement of two-channel stereo as a serious high-fidelity medium. The brutal fact is that more than 99% of the musically significant recordings made in the last fifty years are only available in 2-channel distribution, either on LP or CD.
Another odd thing about the book is the typical listening environment is depicted as a 13-seat home theater, with 3 seats in the front row, 5 in the middle row, and 5 in the back row. Is Floyd seriously proposing that people listen to music discs in a (very) high-end dedicated home theater room with a blank screen? Does anyone do this?
Anyone with the disposable income to build a dedicated home theater room, complete with 13 seats and the latest 1080P or 4K projector, isn't going to listen to music in the dark. The high-end audio market has split into two parts: audio for movies, and audio for music lovers. The best environment for movies ... a dimly lit, air-conditioned, windowless room with a large screen ... is not the best environment for music, where you want light and space, and a pleasant environment to socialize, share drinks and refreshments, and chat while you listen to music. It can be an intimate couple, a group of friends, a small party, or hifi nerds showing off the system, it's still a social experience.
Given that some people still enjoy listening to music in a social environment ... we haven't all been forced onto headphones yet ... and the dominant source of music is 2-channel, there are implications for loudspeaker design. To the best of the ability of the designer, the resulting loudspeaker must be optimized for 2-channel playback, if it's intended for music playback. The criteria for home theater may be different, and that's OK, since the requirements are not really the same.
As I see it, audiophiles and music-lovers love music for its own sake, and have large, sometimes very large, 2-channel music collections. Serious music lovers have many formats they can play: mono LP's, stereo LP's, CD's, high-res downloads on computers, and sometimes even reel-to-reel tape recorders. Yes, there are a handful of multichannel recordings in SACD, DVD-A, and download formats ... and they sound wonderful on a fully symmetric 5-channel system. I started out in audio as the inventor of the Shadow Vector quadraphonic decoder, and have a soft spot for really well-done ambient sound.
But I have to acknowledge the reality of the 2-channel catalogue, and honor the older mono catalogue as well. This is the finest music of all time, and it is the task of the playback system to do the best job possible with it. This is where paying attention to the ambient impression can yield a more satisfying audio system.
Last edited:
Between what Duke and I have posted, I think readers with a little imagination can figure out the rest. Yes, you need measurement gear, and yes, careful subjective assessments are just as important.
As a minor digression, some of my experience with ambience retrieval comes from listening to the Shadow Vector decoder with conventionally mixed stereophonic LP's as the source. To my surprise, records with decent-quality mixing (good-quality stereo playback) sounded even better on the Shadow Vector, with much more clarity, since the reverb could now clearly be separated from the instruments. It didn't always work, since spaced-microphone recordings have unintentional random phase relationships that could knock the decoder for a loop. But most recordings sounded better; not just more spacious (which you would expect), but also a lot clearer and more natural sounding.
The reverb (which is pretty much always there on a classical recording) wasn't crammed into a tunnel between the speakers, which is what you usually hear with 2-speaker playback. The reverb evenly filled the room (as it should), with no build-up or "detenting" towards the loudspeakers. In fact, that became one of the fine-tuning criteria for the decoder: an even spread of reverb everywhere, with crisp positioning of the actual instruments. Unfortunately, modern decoders built into HT receivers tend to sound "dryer" and more closed-in than the Shadow Vector prototype, so I still kind of miss the old thing.
But it is possible to make a speaker do some of the same thing; project the reverb across the room, while leaving the instruments crisp and clear. I've heard it for myself when I was experimenting with low diffraction and multi-directional drivers back at Audionics, so I know it's possible. The Ariels fill about 1/2 to 2/3'rds of the room with ambient impression, depending on the recording, so even better results must be possible.
Last edited:
Would this type of design work with fullrange drivers. IOW, just one fullrange driver on the front and one in the back?
Thanks,
Deon
The rear (or upwards-facing) driver will need EQ that is different than the front-facing driver, and some technique to make sure the sound from the rear driver doesn't diffract around the enclosure. If you do it right, when you flip the A/B switch, there will be no tonal change in the overall sound, no increase in blur, just a substantial increase in the size of the ambient impression.
To clarify: "Ambient Impression" is a technical term that refers to the subjective sense of performance room size when you play a recording. Some recordings don't have any sense of room size, since the vocalist was recorded in a tiny booth and the digital reverb doesn't have a realistic-sounding algorithm. This can be distracting on a movie soundtrack, since the visual space on the screen will be at odds with the acoustically closed-in booth used to re-record the actor's dialog.
Some recordings are even weirder, with different instruments recorded in different studios (for the convenience of the musicians). If the loudspeaker provides a realistic sense of ambient impression, a recording like this will have multiple overlapping images, overlaid on each other. A multi-miked classical recording with lots of action on the faders can sound very odd indeed, with spotlighted instruments zooming forward and backwards with every touch of the fader.
Soffit-mounted studio monitors typically create very little ambient impression, so the recording and/or mixdown engineers may be flying blind when they twiddle the knobs on the panpots and digital reverb section. On the other hand, the loudspeakers used at the final mastering stage are typically freestanding monitors well out in the room, so there's one last opportunity to adjust the spatial impression in a closer approximation to a domestic setting.
To clarify: "Ambient Impression" is a technical term that refers to the subjective sense of performance room size when you play a recording. Some recordings don't have any sense of room size, since the vocalist was recorded in a tiny booth and the digital reverb doesn't have a realistic-sounding algorithm. This can be distracting on a movie soundtrack, since the visual space on the screen will be at odds with the acoustically closed-in booth used to re-record the actor's dialog.
Some recordings are even weirder, with different instruments recorded in different studios (for the convenience of the musicians). If the loudspeaker provides a realistic sense of ambient impression, a recording like this will have multiple overlapping images, overlaid on each other. A multi-miked classical recording with lots of action on the faders can sound very odd indeed, with spotlighted instruments zooming forward and backwards with every touch of the fader.
Soffit-mounted studio monitors typically create very little ambient impression, so the recording and/or mixdown engineers may be flying blind when they twiddle the knobs on the panpots and digital reverb section. On the other hand, the loudspeakers used at the final mastering stage are typically freestanding monitors well out in the room, so there's one last opportunity to adjust the spatial impression in a closer approximation to a domestic setting.
Last edited:
Would an open-baffle driver, like the one shown in the Briggs speaker (see pic below) then not be ideal? An OB driver automatically cancels out side-lobes of energy and show a figure-of-eight (8) pattern in radiation. The only addition I would make is a pipe, preferably about 1.5 times the diameter of the driver, mounted below it, stuffed lightly with wool. The length of the pipe would have to be equal to of greater than the 1/2-wavelength of the lowest frequency that the driver is asked to reproduce. The idea is to absorb as much of the rear (downward) wave of the driver so that it does not simply bounce off the floor and mess up the forward (upward) radiation of the driver (see illustration).
Briggs speaker:
Illustration:
An externally hosted image should be here but it was not working when we last tested it.
Another option would be a tweeter like the Fountek NeoCd3.5H Horn Tweeter. It has very consistent diffusion off-axis in both the horizontal and vertical dimensions. Being a horn, it might show a lot of roll-off outside the coverage of the horn. See pics below for diffusion patterns of the Fountek NeoCd3.5H Horn Tweeter.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
Deon, the pipe would act as a narrow bandstop filter if used as such. Also cannot see using a pipe that could easily be 16' long.
The other day I was pondering why my EDT and T30 room curves were so good. I haven't applied any acoustical treatments as of yet. Then it hit me, I have more than just a few sonotubes of various diameters placed around my living room for build materials. Stored upright work quite well as unintentional bass traps.
The other day I was pondering why my EDT and T30 room curves were so good. I haven't applied any acoustical treatments as of yet. Then it hit me, I have more than just a few sonotubes of various diameters placed around my living room for build materials. Stored upright work quite well as unintentional bass traps.
Interesting where this discussion has led. As mentioned earlier, I have heard "spatial recovery" methods sound quite convincing on the right recordings. While I do have a couple hundred classical recordings, plus some other live recordings, the majority of my listening is still with studio recorded albums. I got into this hobby as a marriage with my love of music, not searching for some holy grail, and thus I don't let my hobby dictate my music selection most of which would not be typical of the audiophile demographic.
I wholly agree with you Lynn about the 2 channel reproduction goals versus the dedicated home theater for music listening. I believe Dr. Geddes is more in the Toole camp judging by a brief conversation I had with him at his place. I have two systems (mostly tube based) as a result of this, upstairs is my two channel system where I listen to most of my music often times while cooking, hosting small parties, reading or working on something. Downstairs is my HT system with typical receiver, plasma, surround in a garden level fairly dark room. I have some really great concert DVDs that are very enjoyable to watch here, but still the majority of listening to music is upstairs. The upstairs system is more of a lifestyle system to borrow a Bose term. Its how music becomes part of my everyday life even in social situations.
I wholly agree with you Lynn about the 2 channel reproduction goals versus the dedicated home theater for music listening. I believe Dr. Geddes is more in the Toole camp judging by a brief conversation I had with him at his place. I have two systems (mostly tube based) as a result of this, upstairs is my two channel system where I listen to most of my music often times while cooking, hosting small parties, reading or working on something. Downstairs is my HT system with typical receiver, plasma, surround in a garden level fairly dark room. I have some really great concert DVDs that are very enjoyable to watch here, but still the majority of listening to music is upstairs. The upstairs system is more of a lifestyle system to borrow a Bose term. Its how music becomes part of my everyday life even in social situations.
Actually, since the driver will mainly be used for high frequencies, I doubt it would have to be longer than 7". And I forgot to mention that the bottom of the pipe must be closed, otherwise it would not do what I suggested for it to do. OTOH, I know too little about acoustics, so you might probably be correct that it would just be a bandstop filter.
Last edited:
I think that I need to chime in here on several points.
JoshK - I strongly believe that a good listening room for music is the same as it is for movies. I don't see where there would be differences. But I am strongly opposed to Toole's position on stereo as a "flawed" medium. To me two channel is perfect - it is the medium, it cannot be anything less than perfect for what it is. It's just like oil painting versus pen and ink drawings, neither is a "perfect" medium for creating realism - that's not what they are for. But each is perfect as the medium that it is to the artist. Two channel may be incapable of perfectly recreating the impression of being in another venue, but it is perfect at creating a work of art for two channel. And lets face it, two channel is not going away soon. Two channel is the medium and art works aimed at this medium must be judged as such, not as a surrogate to going to a concert hall.
Lynn - I use my HT for serious two channel listening, mostly in the dark (visual is a distraction) several times a week. It is one of my greatest pleasures in life. I do not agree that music rooms are for social events. In fact when we have a social party, the HT is usually empty and people mill about elsewhere. I find that serious listening is a solitary event - do you have a conversation with your wife when you go to the symphony? I hope not! That annoying background noise would mask the conversation anyways. To me social listening and serious listening are two entirely different things, favor entirely different forms of music and most likely entirely different sound system setups.
JoshK - I strongly believe that a good listening room for music is the same as it is for movies. I don't see where there would be differences. But I am strongly opposed to Toole's position on stereo as a "flawed" medium. To me two channel is perfect - it is the medium, it cannot be anything less than perfect for what it is. It's just like oil painting versus pen and ink drawings, neither is a "perfect" medium for creating realism - that's not what they are for. But each is perfect as the medium that it is to the artist. Two channel may be incapable of perfectly recreating the impression of being in another venue, but it is perfect at creating a work of art for two channel. And lets face it, two channel is not going away soon. Two channel is the medium and art works aimed at this medium must be judged as such, not as a surrogate to going to a concert hall.
Lynn - I use my HT for serious two channel listening, mostly in the dark (visual is a distraction) several times a week. It is one of my greatest pleasures in life. I do not agree that music rooms are for social events. In fact when we have a social party, the HT is usually empty and people mill about elsewhere. I find that serious listening is a solitary event - do you have a conversation with your wife when you go to the symphony? I hope not! That annoying background noise would mask the conversation anyways. To me social listening and serious listening are two entirely different things, favor entirely different forms of music and most likely entirely different sound system setups.
Here I actually have to agree. I also like listening in the dark. If I have a lot of guests, the music will be in the background and things that are important to me in a good system would not really matter then. I do however want a system that 2 or 3 people can enjoy to the fullest at the same time, because for me one of the greatest pleasures in life is to share music with other people. But not with a crowd, just one or two. Just my $0.02.
Nothing like solo piano music for telling one whether the system is in the zone: wind up the volume to full, realistic levels, and wander around the house, talk to people, etc. This should be convincing whether you're at the far end of the house, or standing immediately adjacent to the speakers. And 2 channel has zero problems conveying the visceral impact and intensity of that instrument if all is in order, no surround or darkness should be necessary, at all ...