Dave
The definition of DI does not require the definition of a single axis as you incorrectly state. Look at the definition in Beranek. It is defined for any angle as the ratio of the field response along that axis to the power response.
The reverberant response is only correlated with DI if you define a single axis as you incorrectly do. It is always correlated with the power response. Hence if the DI changes with direction and the power does not then clearly the reverberant field does not track with the DI.
We all know you have your opinion and preferences and I have mine.
The definition of DI does not require the definition of a single axis as you incorrectly state. Look at the definition in Beranek. It is defined for any angle as the ratio of the field response along that axis to the power response.
The reverberant response is only correlated with DI if you define a single axis as you incorrectly do. It is always correlated with the power response. Hence if the DI changes with direction and the power does not then clearly the reverberant field does not track with the DI.
We all know you have your opinion and preferences and I have mine.
I've never seen it used other than as being with respect to some preferred and hotter axis. (as is Q, that d.i. Is derived from) If it is defined as you say, then it falls in nicely with what I was explaining: that a directional speaker (high d.i.) is effectively less directional when you sit off the main beam. That is, the gain of direct field over reverberant field is diminished.
Not following you here. I was using a more flexible definition of d.i. as being defined for whatever axis you are sitting on. Now you are telling me this is the true definition. Great.
Either way we don't have to get hung up on d.i. as it is just a convenience for calculating the direct field and then the reverberent field. Calculate them however you would like. The point you have missed is that we can define a desired axial direct field level and then our reverberant level will be set by the relative strength of the speaker's power response (plus distance and room acoustics). When you stipulate a directional speaker aimed away in a lively room (your preference) I am saying that I can match it completely with (for example) a wide dispersion speaker in a deader acoustic.
Actually, I'm trying to be open minded in my preferences as I believe that several solutions can give exactly the same thing.
Regards,
David
Earl, this idea still needs to be proven and I have evidence of the opposite being true. My aim, as yours, in developing loudspeakers, is to have constant directivity. However, you strive for cd within a certain opening angle. My focus is on a cd as much as possible 360 degrees around. Even in a hard room, the early reflections don't at all interfere with imaging. Physiologically, I blame the Haas effect for this. There is one technical condition I found that needs to be fulfilled in order for this to happen; there should not be any sharp discontinuities in the FR at different listening angles.
Well I don't believe that - they will be different in significant ways. We do not seem to be convincing each other of much of anything. Nothing new there.
Vacuphile
You should read and maybe talk to Prof. Blauert because he, and I, are very much of the opinion that there is a tradeoff between spaciousness and imaging as regards very early reflections. They do affect imaging and spaciousness in opposite ways.
Earl,
Thanks for having me open up again Blauerts book 'Spatial Hearing', which has sat on my bookshelf gathering dust for quite some while now. Couple of things (after having refreshed my memory):
1) ability to localize a sound source is linked to the size of that sound source, a point source being most precise. A sound source that has constant directivity in within a limited window by necessity has to be larger than a point source. My speakers are virtual point sources. As a consequence, they radiate in 4 pi. Almost, that is.
2) early reflections get surpressed, as in, they are inaudible. They do not lead to more spaciousness, but add to loudness (of the leading sound source). This is not an argument for cd over a limited window. The early reflections of a omniradiating sound source do not lead to smearing because of the precedence effect.
At any rate, puddings are proved by earing. I am preparing a series of loudspeakers -virtual point sources- for demonstration purposes, and I can assure you that they can place voices and instruments with pin point position in 3D space, even in highly reverberant rooms.
In the meanwhile, the book I am referring to was first published in 1974 and I have the English translation with new material from 1997 by MIT. If Blauert has published since on this issue in ways that might shift my thinking, I would be much obliged if you could let me know.
Thanks for having me open up again Blauerts book 'Spatial Hearing', which has sat on my bookshelf gathering dust for quite some while now. Couple of things (after having refreshed my memory):
1) ability to localize a sound source is linked to the size of that sound source, a point source being most precise. A sound source that has constant directivity in within a limited window by necessity has to be larger than a point source. My speakers are virtual point sources. As a consequence, they radiate in 4 pi. Almost, that is.
2) early reflections get surpressed, as in, they are inaudible. They do not lead to more spaciousness, but add to loudness (of the leading sound source). This is not an argument for cd over a limited window. The early reflections of a omniradiating sound source do not lead to smearing because of the precedence effect.
At any rate, puddings are proved by earing. I am preparing a series of loudspeakers -virtual point sources- for demonstration purposes, and I can assure you that they can place voices and instruments with pin point position in 3D space, even in highly reverberant rooms.
In the meanwhile, the book I am referring to was first published in 1974 and I have the English translation with new material from 1997 by MIT. If Blauert has published since on this issue in ways that might shift my thinking, I would be much obliged if you could let me know.
Perhaps not, but what I am claiming is basic acoustics. With the variables of speaker directivity, listener distance, aiming and room acousics there are a wide range of combinations that will get to any desired direct to reflected ratio.
Particular reflection patterns or frontal versus lateral enrgy ratios may tie to specific situations, but if room treatments are allowed then even that can be manipulated as desired.
These are not theoretical musings. As I have mentioned, I am experimenting with a system at work that is based on a 901 with updated drivers and DSP EQ. Front driver and rear drivers have seperate EQ paths. Direct and reflected sound have independent delay, level and response curve.
The ability to adjust early and late response curve and energy balance is very enlightening. There is definitely a tradeoff between spaciousness and precision in my experiments. Strong early reflections were initially a problem, giving a slap echo effect when reverb level was raised. Adding a few carefully placed reflectors has allowed me to raise reverberent level.
Its still not clear if delay of the onset of reflected energy will give a better combination of spaciuosness and precision. It is an interesting study, and much of it could be duplicated by others with conventional speakers and judicious placement of absorbers, perhaps guided by Holm Impulse.
David
Interesting stuff, David. Sounds like a system somewhat similar to Ken Kantor's 'Magic Speaker', combined with smart room treatment.
You and Earl have had this particular discussion before and in my view you are right. It's quite obvious to me, so I'm a bit confused Earl doesn't seem to get it.
You and Earl have had this particular discussion before and in my view you are right. It's quite obvious to me, so I'm a bit confused Earl doesn't seem to get it.
Hi Keyser,
Yes, it is similar in concept to the "Magic", a very interesting system. I don't think there are any commercial prospects for it but it makes a great test bed for just these types of psychoacoustic studies. I may have to revise it to get more differentiation between direct and reverberant outputs, but it is early days in the experiments. My interest, as always, is in the perception of frequency response but the spaciousness aspect is easy to play with as well.
I think that sometimes Earl has his engineer/physicist hat on and sometimes his product marketing hat....
David
In loudspeaker design there are a few things that most of us agree with: smoothness of direct response, smoothness of power response, both should fall slightly, and this means a smooth near flat DI. What we don't agree on is what DI is best, which is very likely to be situation dependent. There is scant evidence in this regard and what there is, is weak. My experience is that as the DI falls in a given room the imaging capability falls as well, while spaciousness increases. This is consistent with my understanding of how we hear from sources like Blauert and Griesinger. We all interpret what we understand in ways that tend to support our position, that's natural. It may be years before this situation gets ironed out.
Most speaker designers (we?) agree that the direct response should not be flat but fall slightly?
As mentioned before, the desired DI should be situational dependent based on the room variables and placement options. Measuring and designing in room is certainly a challenge if not impossible. That's where good ol' voicing comes into play.......a purely subjective function that can be considered an evil necessity. No one wants to go to the darkside of audiophool tweeking, but to achieve optimal results in YOUR space! welcome to the darkside.......may the force be with you! Lol.
Again, my take is that we hear primarily on a vertical plain with localization above and below coming from reflections. Using off axis vertical radiators to smooth reflective response anomolies makes the most sense to me........unless you're one of the crowd that considers off axis high frequency response vertically as neccessary........such as standing opposed to sitting or multiple rows of seating in a theater.
Again, my take is that we hear primarily on a vertical plain with localization above and below coming from reflections. Using off axis vertical radiators to smooth reflective response anomolies makes the most sense to me........unless you're one of the crowd that considers off axis high frequency response vertically as neccessary........such as standing opposed to sitting or multiple rows of seating in a theater.
Mayhem, what are you attempting to achieve with the smaller mid-bass drivers (6 inches +- vs 10-15 inches)? Are you looking at waveguide tweeter designs that are not narrow so you can match them up with a 6 inch driver crossed lower than the point where it starts to narrow? Or are you just looking to match up the 6 inch mid-bass somewhere between 2.5-3.5 kHz so that is DI is close to that of a tweeter in a 6 inch wave guide?
I am assuming you want the DI consistent between mid-bass and tweeter at crossover point regardless of your type of DI your are going for (narrow, wide, something in between...), but I could not tell for certain from your earlier posts.
I do like what Dr Geddes has done with his designs, but given my own "wide" DI speaker project I am very curious what you, Speaker Dave, and others look to do to manage side or rear wall VER, which is going to be a challenge for me in my room.
I am assuming you want the DI consistent between mid-bass and tweeter at crossover point regardless of your type of DI your are going for (narrow, wide, something in between...), but I could not tell for certain from your earlier posts.
I do like what Dr Geddes has done with his designs, but given my own "wide" DI speaker project I am very curious what you, Speaker Dave, and others look to do to manage side or rear wall VER, which is going to be a challenge for me in my room.
Earl, I agree fully with this analysis. Somewhere in the next year I will be touring the US with speakers and would very much like to do a comparison. It may all boil down to listening environment and personal preference, but for the moment I have the impression that omni's do give unsurpassed stereo imaging, in spite of early reflections. I gather it is also the basis for the remarkeable success in listening tests of Linkwitz's Pluto and LX mini (although these flip midband from 4pi to 2pi).
I think most designers agree that the in room steady state response should shelve or slope to some degree. The direct response is more frequently flat.
What axial response, what power response and what polar curves would make the perfect speaker?
A very good question...
David
Since the steady state response is highly coupled with the power response, according to Toole, and this makes perfect sense, then in a lively room with a constant directivity speaker the direct response would need to fall as well as the power response. This is precisely what I find. A non-CD speaker can have a flat listening axis response, but a CD speaker will sound bright if the listening axis response is flat.
In my opinion the most desirable DI depends on a few factors: 1) room reverb time, the larger the reverb the more desirable a higher DI becomes. A dead room needs a low DI or it will lack spaciousness; 2) musical preference, classical music performed in large venues is all about spaciousness and there really isn't an image per-see. Other forms of music have little inherent spaciousness intent but have very precise phantom images in the mix. The former lends itself well to low DI speakers, while the later lends itself more to a higher DI.
In my opinion the most desirable DI depends on a few factors: 1) room reverb time, the larger the reverb the more desirable a higher DI becomes. A dead room needs a low DI or it will lack spaciousness; 2) musical preference, classical music performed in large venues is all about spaciousness and there really isn't an image per-see. Other forms of music have little inherent spaciousness intent but have very precise phantom images in the mix. The former lends itself well to low DI speakers, while the later lends itself more to a higher DI.
Earlier, I indicated I would rather cross outside of 1-4khz. To avoid distortion products, the options are extremely limited at and below 1khz. Directivity control requires a large guide.....something I simply don't have the space for so I'm targeting 4khz and higher. LOTS of options there but new tradeoffs arise with midrange selection. 6-6.5" midrange drivers which can cross that high without breakup are already narrowing. Those with phase plugs, curvilinear cones and rear mounted on a baffle can reach my 4khz goal...........and they can get us low enough to avoid crossing above the modal region to woofer or sub. Used in multiples with a 1st order low pass (.5 coil) get avoid the three way dilemma entirely while aiding in the control of vertical directivity. Its the HF device that's questionable at this point. I'm a BIG fan of ribbons and the like when used properly. Excellent horizontal coverage which can be further improved with mild horn loading. Again, for my space vertical is a must as there's excellent benefits to these designs if done right. Lots to still consider.
I would welcome this comparison in my listening room.
The Omni's that I heard at a audio fair last year (some German very high end brand) had loads of spaciousness, quite smooth response, but no imaging ability at all by my standards. Of course too many variables there to make any solid judgments, but this example certainly did not contradict my beliefs.
I also heard the Pluto at this same show. Probably the best speaker at the show, but that is not saying much - they ranged from unbelievably bad to OK.
Hopefully when you are done you will share the polar maps with all of us. Get those right and you are 90% of the way there. Serious issues will be pretty evident.
Depends on how you define "breakup". Damped wideband breakup as opposed to undamped narrowband breakup is not absence of breakup.
Pretty much any 6" driver - paper, poly, aluminum, will break up within an octave of that XO frequency. Exotic materials like Be might get you a bit higher, but how much higher?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.