You address some interesting points!
Obviously true.
Although I don't experience in-head localisation with my speaker, I agree about the second part of what you say above. I think that is why we do want late reflections or some amount of reverberation.
This is probably true. I think novice listeners may not like high directivity speakers as much as those with more experience. In my personal experience early reflections can help establish a pleasantly big sound, they add to ASW. Because of this easily noticeable effect a layman may be best served with conventional direct-radiators, whereas more experienced listeners may be more sensitive to coloration and defects in imaging and as such may prefer higher directivity speakers with a minimum of early reflections.
In principle flat should be the goal. What goes in should come out unchanged. But of course the difference in HRTF between a sound at 0 degrees and 30 degrees is real, and it will as such make a centre-image sound overly bright. However, if you change the response of your speakers to compensate for this effect, you're not only compensating for sounds that are at 0 degrees, but you are also applying the same compensation to sounds from any other direction between 0 and 30 degrees. Correction for the phenomenon should be done at the recording side of things.
This is an interesting point. I wonder what others think about this.
Obviously true.
Although I don't experience in-head localisation with my speaker, I agree about the second part of what you say above. I think that is why we do want late reflections or some amount of reverberation.
This is probably true. I think novice listeners may not like high directivity speakers as much as those with more experience. In my personal experience early reflections can help establish a pleasantly big sound, they add to ASW. Because of this easily noticeable effect a layman may be best served with conventional direct-radiators, whereas more experienced listeners may be more sensitive to coloration and defects in imaging and as such may prefer higher directivity speakers with a minimum of early reflections.
In principle flat should be the goal. What goes in should come out unchanged. But of course the difference in HRTF between a sound at 0 degrees and 30 degrees is real, and it will as such make a centre-image sound overly bright. However, if you change the response of your speakers to compensate for this effect, you're not only compensating for sounds that are at 0 degrees, but you are also applying the same compensation to sounds from any other direction between 0 and 30 degrees. Correction for the phenomenon should be done at the recording side of things.
This is an interesting point. I wonder what others think about this.
Is there a link?
You obviously don't read many of my posts. Yes, I did here something, there was sound coming from it
The presentation is spacious as soon as certain reflections from boundaries are loud enough but that's where the list of positive characteristics for such a system currently ends.
In theory yes, in practice I doubt whether any recordings are corrected in this way, at least not intentionally, and I don't think deliberately making speakers non flat is the answer either.
One possible approach would be to do it in decoding in the receiver - many receivers can decode stereo L+R into 3 channel L+C+R with the centre channel signal eliminated entirely from the L & R channels for reproduction when a physical centre channel is available.
If this is possible, why not take this internally decoded 3 channel signal, apply an HRTF correction to the centre channel only, then re-mix this corrected centre channel back into L+R for 2 channel output ?
Seems like it wouldn't be hard to do in a multi-channel DSP, but do any on the market receivers or signal processors attempt such a correction ? Would it make that much of a difference in practice ?
By the way, it is my opinion that for judging loudspeaker performance a processed or mixed recording is useless because it has to be mixed on other speakers. Also, it includes the preferences of the recording engineer.
I'm especially wondering whether a flat on-axis curve is correct and which directivity curve (flat or rising) would be correct. I think one might be able to test this with naturally recorded sounds (not mixed) and then playback on different stereo systems with different directivity patterns in an anechoic environment, a small room with early reflections and a larger room. Subjects would then have to judge the perceived timbral characteristics of the system.
Of course, there is one problem with this kind of test. Multiple effects, which are rather hard to separate, are combined in one judgement of timbre: the well-known comb filtering in the phantom sources in a stereo system and HRTF coloration of phantom sources.
Testing in an anechoic environment will aggravate the audibility of the phantom source comb filtering while minimizing influence of the polar response. In a room, the comb filtering would be less audible but the polar response will contribute more. Maybe testing with a center channel would already alleviate the comb filtering problem.
Also, reflections change the perception of timbre by adding 'richness' and 'body' to sound, variation in the balance of direct vs reflected sound between environments may very well bring an extra perceptual variable.
I'm especially wondering whether a flat on-axis curve is correct and which directivity curve (flat or rising) would be correct. I think one might be able to test this with naturally recorded sounds (not mixed) and then playback on different stereo systems with different directivity patterns in an anechoic environment, a small room with early reflections and a larger room. Subjects would then have to judge the perceived timbral characteristics of the system.
Of course, there is one problem with this kind of test. Multiple effects, which are rather hard to separate, are combined in one judgement of timbre: the well-known comb filtering in the phantom sources in a stereo system and HRTF coloration of phantom sources.
Testing in an anechoic environment will aggravate the audibility of the phantom source comb filtering while minimizing influence of the polar response. In a room, the comb filtering would be less audible but the polar response will contribute more. Maybe testing with a center channel would already alleviate the comb filtering problem.
Also, reflections change the perception of timbre by adding 'richness' and 'body' to sound, variation in the balance of direct vs reflected sound between environments may very well bring an extra perceptual variable.
Last edited:
I agree with keyser that ideally any form of correction should be done on the recording side. Even better would be the use of a center speaker. It would be possible to upmix a stereo signal by taking the cross-correlation of left and right, but that only works for amplitude-panned signals. It would also require quite a processor. Unfortunately I don't have the time or skills to make such a processor, although I could work out it's basic method of processing. I'd rather buy a good upmixer, but I don't really trust all the different types on the market. Most of them probably add up L and R for C, as this is much simpler and cheaper to implement.
Last edited:
The stuff we debate ad nauseum has frequently been studied by proper researchers. I ran across this by Flindel and McKenzie. Subjective evaluations of preferred loudspeaker directivity AES preprint 3076. They set up a simulation in an anechoic chamber with a pair of normally placed speakers and 5 extra pair that represented the first and second front and sidewall bounces. They could vary the amount of rolloff of all the extra speakers over a wide range, representing a continuum of speaker directivity from omni to very directional as perceived in a typical room.
They split the listeners into a group of industry professionals and a group of "naive" listeners. The professional group was about equally split between liking the most directional extreme and the omni-directional extreme (few votes for any middle directivities). The naive group had a strong preference for the omni settings.
This from John Crabbe (long term editor of HIFi News) in an Audio magazine March 1973 issue on omnidirectional speakers: "On a stereo recording of a harpsichord concerto on which a seemingly small solo instrument is contrasted nicely with a broad orchestral backcloth, the harpsichord stubbornly occupied the full space between the speakers as heard from any point in the listening area. This is not good stereo-it is hardly stereo at all-and I must beg to differ most strongly with critics who state that omnis 'do provide a good stereo image virtually anywhere in the room'."
Then: "In terms of sound reproduction, of producing an accurate acoustic equivalent of the signals passing through the stereo amplifier, omni-directional loudspeakers represent a firm step backwards."
Is it as simple as sophisticated listeners prefer directional speakers and less sophisticated listeners, omnis?
David S.
They split the listeners into a group of industry professionals and a group of "naive" listeners. The professional group was about equally split between liking the most directional extreme and the omni-directional extreme (few votes for any middle directivities). The naive group had a strong preference for the omni settings.
This from John Crabbe (long term editor of HIFi News) in an Audio magazine March 1973 issue on omnidirectional speakers: "On a stereo recording of a harpsichord concerto on which a seemingly small solo instrument is contrasted nicely with a broad orchestral backcloth, the harpsichord stubbornly occupied the full space between the speakers as heard from any point in the listening area. This is not good stereo-it is hardly stereo at all-and I must beg to differ most strongly with critics who state that omnis 'do provide a good stereo image virtually anywhere in the room'."
Then: "In terms of sound reproduction, of producing an accurate acoustic equivalent of the signals passing through the stereo amplifier, omni-directional loudspeakers represent a firm step backwards."
Is it as simple as sophisticated listeners prefer directional speakers and less sophisticated listeners, omnis?
David S.
I recommend you to consult the thread i started last year
http://www.diyaudio.com/forums/full-range/176556-dipol-08-baffle-dimension-list-crossover-parts.html
as dipol 08 does not rely on distance weighting solely but also
on power tapering of the array.
Overall placement of the drivers is not exactly according to golomb
ruler, but golomb rulers served as a starting point for optimization.
There is no "psychoacoustic principle" as such. But as you reported
you were irritated by "localizing the tweeters" in many multiway
arrangements, you may find this not being the case with this design.
The concept is breaking the rules by tolerating interference as such,
which cannot be circumvented in using fullrangers in a line array
anyhow. But frequency response varies only smoothly with
vertical listening angle and distance. Tonal balance is very
consistent with different listening positions.
For closer listening <1.5 meters you may wish to make a slightly
downsized version as when sitting close and low, you will loose some
of the top end "sparkle".
The speaker provides a compromise between directivity in the vertical
plane, but also some diffusivity in the reflections especially from the
ceiling. Floor bounce is cancelled out completely.
In dipole arrangement it is a "dry" sounding concept, allowing
larger listening distances even in rather "live" rooms.
Different to a conventional (unweighted an non tapered) line array,
phantom sources have a subjective height in soundstage, which is
stable at ear height, as the phantom sources do not move with your
head (e.g. when knee bending) while listening as they do in a
"real" fullrange line source.
You may have a look at the characteristics of the passive compensation
network to get inspiration for your own design.
Having the fullrangers not pointing at least near to the listener, will
cause rolloff in the top end, which has to be accounted for when thinking
about adapting this to "beveridge placement" ...
More than 30 degrees off axis is not recommended with drivers used.
dipol 08 is designed for conventional dipole placement, i personally prefer
to toe in for L-R crossing somehere behind the nearest listening position.
Last edited:
Well I certainly am, and commercially there is the entire music DVD industry which is a pretty big market and will have good sound quality as a format when Bluray displaces DVD.
So the technology is in the shops, the market is established. One could say there is plenty of interest but on the other hand there is a degree of elitist prejudice in favour of 2 channel.
But it is OT for this thread so I have refrained from bringing it up.
oh, I'm so with you there, Dave. But these forums are so often not used for learning but as a venue to state one's personal listening experiences and ask the world to explain why it is real, from first principles.
I would say there is a problem that the experiments are in an anechoic chamber, so *many* normally occurring reflections would still be missing and listeners would 'hunger' for the option which satisfies this most, i.e. all speakers running. One can imagine what their experience would be when only two speakers are turned on: the standard horrible anechoic chamber experience.
While that's true, it doesn't mean that everything has been figured out yet by said researchers, or that it's not worth discussing it. Much of the (admittedly limited set of) research I've seen is quite narrowly focused on specific parameters or issues, and while it's good scientific method to isolate variables individually, it doesn't really address the complex interactions of the myriad of different parameters which a recording, speaker, room, listener system chain can have.
It's an extremely multidimensional problem space, and exploring one axis at a time in isolation doesn't necessarily give all the answers, and now and then debating the same data points "ad nauseum" may actually give new insights in how all this stuff fits together. Sometimes it's the insights into the data as a whole that are more valuable than the research that led to the gathering of individual data points, as important as that may be.
Besides, its fun to discuss and if nothing else, tends to bring those of us who have gaps in their knowledge up to speed, when things that have already been figured out or researched are pointed out to us.
Ouch, that's bound to ruffle a few feathers
Could it simply be that naive listeners are not good judges of sound because they haven't experienced truly great sound before ? Typical mid-fi systems poorly set up are so bad that they have no frame of reference to judge great sound, especially if they're not regular listeners of real, un-amplified performances.
A high direct to reflected ratio would sound alien to most naive listeners. The trouble with sound quality is once you learn to recognise good sound quality, you can never un-learn it, and the flaws in what was once acceptable is all too obvious. Something that I think everyone in this thread will appreciate...
Thanks for sharing, interesting point. I wouldn't be surprised if it would simply be this way. Maybe a stereo system needs to be a compromise because of the fundamentally limited spatial qualities. I'm not sure what I prefer; an omni-directional speaker with a large, inflated sound stage (increased ASW thanks to early lateral reflections) can be nice and a little more spacious, but it always lacks the definition of a strongly directional speaker.
I think it's a shame there's such a lack of proper surround sound music recordings. I think a 5.1 setup solves most problems: it partly solves phantom source HRTF coloration and comb filtering, it can introduce better spaciousness with lateral reflections with sufficient delay which will also result in better masking of early reflections.
Last edited:
Thanks Dave very interesting review. Your conclusion certainly did not "ruffle my feathers".
Technically could, but it will never happen to be very common.
First, 7 + 1 good quality speakers will be too expensive. And where to place them in normal 25 m2 living room ?
But there are more reasons behind. I believe most normal people do not want to be surrounded by the orchestra or the band. If you go to live concert the sound will come from the stage, which in a field concert is in very narrow angle maybe 10 degrees.
And then there is the evolutionary survival theory that sound from the back are considered threathning and not pleasant, and thus they should be avoided.
The only reason for more than 2 loudspeakers are to generate the surrounding ambiance reverberant field.
(Movies are another matter but not interested.)
- Elias
Naive listeners also prefer the sound of the "smile curve" (otherwise known as boom and tizz, or loudness contour) over a flat response. Does that make it correct or more accurate ? No.....
Most naive listeners also prefer highly compressed audio (dynamic range compression, not mp3 compression) simply because it "sounds louder", even though it may fatigue them in the long term. Does that make it correct or more accurate ?
Most naive TV viewers also prefer the "torch mode" of a TV, which is the mode that TV's are put into in stores which makes the picture very bright, contrasty and bluish in tone to compete with other TV's and fluorescent lighting in the store. Most viewers don't know any better and will leave their TV in this mode at home unless the instruction manual explicitly tells them to change it, or they have it calibrated by a professional, and even then they will probably find it "too dim and yellow", even though it's now more accurate.
Asking a naive listener their opinion on what sounds best (especially when they have heard a limited range of system types before) is a bit like asking someone that knows nothing about wine tasting which of two wines is best, and why.
You won't find out which wine is better, only which wine naive wine drinkers prefer
Likewise asking a group of naive listeners which sound configuration sounds better will only tell you which configuration naive listeners exposed to typical systems prefer, not which system is "better". I wouldn't place too much faith in sound quality testing done with naive listeners, as listening in itself is a skill that has to be learnt.
Last edited:
I should add that they simulated all the reflections in order from the earliest. When they got to the next few, that were from the back wall, they found they were generally undetectable and so they stopped there. They also tried digital reverberation to give a reverberant "tail" but found it was artifical sounding and wasn't changing the results.
It wasn't a perfect experiment in several aspects but the bimodal preference result is interesting and falls in line with other experiments. Note that the half of the expert group that choose heavy filtering of the simulated reflections were voting for anechoic (2 sources only) stereo above 500Hz.
David S.
No ! Trained vs. untrained people listen and rate equally good. It only takes some more time for the untrained "to fill in the questionnaire".
I believe 7.x mid quality speakers can perform better than 2 very good ones. Sound quality isn't in the electronics (anymore).
The reason for multichannel is not to put the listener in the middle of the band. There are a lot of recordings that try to do that and I think it does a disservice to high quality music reproduction. The only reason for more speakers is ASW and LEV. LEV is easy to do in 5.1. ASW would require wide and height speakers just what Audyssey DSX does.
Any system has to work both with movies, TV and music. Otherwise there's no market for it.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.