I find it puzzling. But as the proof of the pudding is in the eating (see eg. Moulton's demonstration for Olhsson) and other Moulton observations seem right to me I think it's lack of precision. I believe that what Moulton says here means that loudspeakers can be treated as instruments under one particular aspect - auditory dereverberation - with the reservation that they had to meet certain requirements for this psychoacoustic mechanism to work
Nate - thank you very much, that is exactly the data that I was looking for and clearly points out the issue that I was discussing.
I do smooth my data using a Gammatone filter bank, but not as a sliding window. The sliding window is not precisely the same as smoothing with Gammatone filters. Admittedly, a sliding window is a lot easier. But I actually do use Gammatone smoothing (also known as ERB smoothing) to EQ my speakers.
I read some of the Moulton interview but stopped when he got his "scientific" claims wrong. He lumps all sound in 50 ms together as one thing, which is wrong. According to Blauert in his book Spatial Hearing, the first few ms are distinctly different than the later ones in that they form what is called the "summing localization" region where we form our impression of where a sound is coming from. As I said above this can easily be explained from the Gammatone impulse above. Mr. Moulton has conveniently ignored this critical aspect of the problem. I don't. It helps to read about what we know about hearing and to get it right.
Hi, I think the gammatone pictures has it backwards. ....
My understanding is that phase alignment happens when initial rise of all frequencies occurs at t=0.
Not with peaks being in sync.....
Maybe I should ask what a gamma tone is ?
Cool. Then our listening experiences point to some of the same things, even if our preferences are not fully in sync. For me, the problem is that "reflection free imaging" has always felt artifcial, in a way. It's like my brain is not buying it: "An orchestra here in my room, and there's no reflections coming from behind and the sides? No way". So I'm wiling to make a compromise: I prefer to get a consistent feeling that good acoustic recordings are detailing acoustic events that go on right where I'm at, even if this comes at a certain price - that the rooms I look into in the recordings may become less unique and distinct from each other.
Btw, a small comment: Even though one can generalize to a certain extent about "omnis", they still behøve fairly different from each other. I happen to think that 99 % of commercial omni designs are flawed. I wouldn't want to live with the MBLs or the German Physiks even if I could afford them. I briefly had the "quasi-omni" Heed Enigma 5, and they annoyed me more and more to the point that I couln't wait to get rid of them. But then again, there are a few omni designs out there that are good.
But this is no different from some other speaker designs. I'm a big fan of big horns, for example. Nevertheless, I think that 99 percent of all commercial horn designs have a coloured presentation with lots of mini-distortion (feeding them with tubes and vinyl doesn't help avoid distortion either, if I may say so). But some of the best systems I've heard in my life have been home-brewn horn systems. I therefore think it's unfair to judge the "omni concept" based on how MBL or German Physiks sound, just as it's unfair to judge "horns" based on some of the more well-known commercial offerings.
My impression is that this looks like a depiction of tone bursts for a minimum phase system (left) vs a Linear Phase system (right). Group delay would then be flat for the chart on the right. Is this thought correct?
100% agreed all the way.
Incidentally, the omnis I'm most familiar with (and upon which the general impression that I described is based) are in fact MBLs (several models, including the top-of-the-range ones).
And I also agree with your comments on horn systems (and specifically those I highlighted in bold).
I think left is linear phase....with 0 deg phase and 0 group delay.
Took me forever to get this....that initial rise time is the alignment point.
Peaks occur when they occur, ie 1/4 wavelength past initial rise
when one wants to increase delay of early reflections then side-wall positioning - "Beveridge placement" - is the best in any given rectangular room, including small rooms
anyone can check it using this handy tool: http://www.linkwitzlab.com/LX521/Dipole%20First%20Room%20Reflections%20SL%20rev4-b.xls
from Listening_room
In case of side-wall "Beveridge placement" one has just to treat the wall near the speakers, eg. with deflecting panel, then this method of speaker placement seems to be the winner as far as increasing delay of reflection is the goal
And the best way to manage both the floor reflection and the ceiling reflection is to use short up-firing speakers.
Pure geometry.
Last edited:
That's basically what I do, except my speakers have a narrow directivity avoiding the first wall reflection - the one not shown in the picture. The sound will diffract around the speaker in that drawing and radiate from the wall its on - not shown!
So your way is "the best way"? I prefer to absorb and/or diffuse the floor and ceiling reflections
So your way is "the best way"? I prefer to absorb and/or diffuse the floor and ceiling reflections
Last edited:
This is my motivation for staying no more narrow than required to avoid early reflections (at the top end) and response dips lower down. If I can hear the room clearly, even if only slightly then I can be happy.
You're right - "best" is not the best word here, my post was partly a re-post and in the original context of speaker placement methods vs early reflections it fitted
But here I agree that in this context words "simple and effective" would be better.
I do not claim that "my way" is the best way. However I do claim that it is in some aspects a viable alternative.
Hi Earl, I know we've had this discussion before and we were unable to persuade you: diffraction as a function of total power radiation is non minimum phase, but as observed as SPL at one point in space is minimum phase. Feyz demonstrated this to you analytically, I demonstrated it to you using MLSSA measures.
I do agree with concept of a phase altering audibility to an extent, though it is a contentious topic. I thought jj put it best: anything that alters the waveform envelope presented to the basilar membrane has the potential to be audible. It's why I find distortion audibility threshold tests, before declared as having ubiquitous applicability, must be explored over reasonable ranges of assumed loudspeaker group delay distortion.
Dave
I can see where minimum phase as a term might be incorrect, but something in the way diffraction modifies the signal, and it will, becomes more audible with level. That was clear in the tests.
Thinking back to Brian Moore's paper he referred to there being a higher audibility of group delay with level. It seems to me that what you are saying and what Moore was claiming may not be completely in sync. I'd have to think about it in some detail and I'm not sure that is going to happen.
And what exactly does "but as observed as SPL at one point in space is minimum phase" mean exactly? How does one define this unless we have an input and an output to compare
PS It could well be that the diffraction does not make the result non-minimum phase, but does significantly alter the group delay while remaining minimum phase. That would be possible and then its this group delay artifact that becomes more audible with level.
I can see where minimum phase as a term might be incorrect, but something in the way diffraction modifies the signal, and it will, becomes more audible with level. That was clear in the tests.
Thinking back to Brian Moore's paper he referred to there being a higher audibility of group delay with level. It seems to me that what you are saying and what Moore was claiming may not be completely in sync. I'd have to think about it in some detail and I'm not sure that is going to happen.
And what exactly does "but as observed as SPL at one point in space is minimum phase" mean exactly? How does one define this unless we have an input and an output to compare
PS It could well be that the diffraction does not make the result non-minimum phase, but does significantly alter the group delay while remaining minimum phase. That would be possible and then its this group delay artifact that becomes more audible with level.
Last edited:
Hi Earl,
Measure any loudspeaker at a point in space with a microphone, remove the excess phase, and the remaining phase response is exactly predicted by the Hilbert Transform (of course assuming no appreciable non linearities are stimulated). I've run these tests myself and satisfied myself unconditionally.
More in response to andy:
Thanks to Earl for sharing the nugget that diffraction altering phase is audible. One thing I struggle with (and where I'll disagree with Andy) is that in audio, subjective tests are conducted and then extrapolated to be ubiquitous (when clearly they aren't) merely because the industry doesn't have the money to assess any of this broadly. I appreciate Sean's tests were conducted accurately but I stand by my point that his tests can not assess the audibility of high q resonances over all possible test signals (even the aspect of delay between incident and resonance was just incidentally explored). Some songs will be better at uncovering specific effects than others. Much music will have short sections of largely tonal non complex portions, whether sustained or not. I truly understand ABX, forced choice, JND etc, I used to be very closely involved with this professionally. Audio subjective testing is extremely expensive and testing often only touches the tip of the icerberg.
The fact that we're sitting here discussing that recent test results demonstrate phase of diffraction is audible contradicts the long held orthodoxy that phase is largely inaudible, illustrating my observation.
Lets be honest, this is a "poor" industry with no money and much research jealously held close to the chest. Its one (of several) reasons I moved to fiber optics long ago. Its still fun as heck, but let's be open minded.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.