I was thinking about this and I found that I had forgotten some of my own work. If the distortion that I heard was due to diffraction in the playback system then my experience would be exactly correct. Reducing the absolute SPL reduced my ability to hear this diffraction. It was still there, but at the lower SPL it was masked.
The more i read stuff like this, the more i think syco acoustics is a total waste of time...
Yes, that's correct, a point I often make as well. It is difficult to impossible to sort out if what we hear as nonlinear distortion is actually nonlinear distortion or a nonlinear perception of a linear defect - like diffraction.
yeah, I know that You are so because of that and since You fail to follow let me ask straight - would You label Bob Olhsson or Dave Moulton as "audiophools"?
I never said that they were and I still don't see how that has anything to do with the topic. Your comments are not clearing it up any either.
I don't know anything about Bob Olhsson, but I have met Dave Moulton. I would not put Dave in the same category as Toole, but he is not a fool either.
I don't know anything about Bob Olhsson, but I have met Dave Moulton. I would not put Dave in the same category as Toole, but he is not a fool either.
Diffraction can be perceived as the generation of ghost harmonics or im products? Now that is fasinating. Any further info to share on this? ive modelled the coclea and cant imagine which physiological nonlinear effect would cause this.
Dave
It is not a nonlinear effect, but a variable sensitivity to non-minimum phase characteristics in a signal. Moore noted in his AES paper on the audibility of non-minimum phase that it is a function of level, by this I mean absolute SPL. Lidia and I later did a study where we simulated the diffraction effect as one might expect it to exist in a horn. In this study we found that absolute playback level as a significant correlate to perception, i.e. the louder the signal the more audible the effect was. But these were all linear signal modifications, nothing nonlinear.
What this means is that as the level of a playback goes up we will become ever more aware of diffraction so that what may not be audible at a lower level will be audible at some level. I have always thought that this was a significant finding, because it means that what we may interpret as nonlinear distortion may in fact be the nonlinear perception of a linear signal aberration.
It is not a nonlinear effect, but a variable sensitivity to non-minimum phase characteristics in a signal. Moore noted in his AES paper on the audibility of non-minimum phase that it is a function of level, by this I mean absolute SPL. Lidia and I later did a study where we simulated the diffraction effect as one might expect it to exist in a horn. In this study we found that absolute playback level as a significant correlate to perception, i.e. the louder the signal the more audible the effect was. But these were all linear signal modifications, nothing nonlinear.
What this means is that as the level of a playback goes up we will become ever more aware of diffraction so that what may not be audible at a lower level will be audible at some level. I have always thought that this was a significant finding, because it means that what we may interpret as nonlinear distortion may in fact be the nonlinear perception of a linear signal aberration.
I've looked at a room as presented by Dave Moulton, my question would be, what does he aim for as far as early side wall reflections go.
The left wall reflection from the left speaker would arrive about ~2.9 ms after the main signal (unless most of it is absorbed by the front wall) and the right wall reflection of the left speaker would arrive almost 9 ms behind the main signal. I'm sure he'd still call it an early reflections while others might say it is not that early at all.
Which reflections do we call early reflections? Even Toole advised to keep some early reflections, but seeing a Harmon listening room it does not quite compare to my room
.
If a reflection arrives after ~7 ms I would expect it to have much less of a detrimental effect on imaging than the ones arriving within the ~ 0 - 7 ms.
The ones arriving in those first zero to 6 to 7 ms skewed perception in my own experience. (sounding artificially wide)
While those arriving later than 6-7 ms had a much more pleasing effect.
At least that's how I experienced it.
The left wall reflection from the left speaker would arrive about ~2.9 ms after the main signal (unless most of it is absorbed by the front wall) and the right wall reflection of the left speaker would arrive almost 9 ms behind the main signal. I'm sure he'd still call it an early reflections while others might say it is not that early at all.
Which reflections do we call early reflections? Even Toole advised to keep some early reflections, but seeing a Harmon listening room it does not quite compare to my room
. If a reflection arrives after ~7 ms I would expect it to have much less of a detrimental effect on imaging than the ones arriving within the ~ 0 - 7 ms.
The ones arriving in those first zero to 6 to 7 ms skewed perception in my own experience. (sounding artificially wide)
While those arriving later than 6-7 ms had a much more pleasing effect.
At least that's how I experienced it.
Wesayso
Your opinion and mine are completely in sync. Be careful, you might get attacked for that.
In my view the earlier and greater the first few refection the more detrimental they are to our ability to perceive a solid image. But there is no clear-cut time scale for this. To me the reason can be seen very clearly if one looks at the impulse responses for a bank of Gammatone filters as are used to simulate our hearing (if I knew were a nice picture of this was I'd attach it.) The gammatone model indicates that at very high frequencies (HF), say 10 kHz, the impulse response is so short that it is over much less than a ms, maybe even .1 ms. So a reflection after this time will have no effect on how our ear/brain interprets these frequencies. But as we go lower in frequency the impulse gets longer and at 1 kHz, the time is 1 ms, so a .5 ms delay will cause a comb filtering effect at 1 kHz, but not much higher. At 100 Hz its 10 ms etc. So clearly there is some limit on both ends that the reflection will not have a major effect, but at virtually all time lags there is some effect.
PS. actual numbers are guesses.
Your opinion and mine are completely in sync. Be careful, you might get attacked for that.
In my view the earlier and greater the first few refection the more detrimental they are to our ability to perceive a solid image. But there is no clear-cut time scale for this. To me the reason can be seen very clearly if one looks at the impulse responses for a bank of Gammatone filters as are used to simulate our hearing (if I knew were a nice picture of this was I'd attach it.) The gammatone model indicates that at very high frequencies (HF), say 10 kHz, the impulse response is so short that it is over much less than a ms, maybe even .1 ms. So a reflection after this time will have no effect on how our ear/brain interprets these frequencies. But as we go lower in frequency the impulse gets longer and at 1 kHz, the time is 1 ms, so a .5 ms delay will cause a comb filtering effect at 1 kHz, but not much higher. At 100 Hz its 10 ms etc. So clearly there is some limit on both ends that the reflection will not have a major effect, but at virtually all time lags there is some effect.
PS. actual numbers are guesses.
Last edited:
I also fail to see what relevance it has to refer to authority. That said, my hunch is that contemporary "audiophools" usually try to avoid reflections, not increase them.
I'll try to sum up this debate how I see it:
1. There is a fairly strong line of research which claims that early reflections are mostly beneficial - for envelopment, spaciousness, listening pleasure, etc. Toole and Olive, but also others have come to the some conclusion. A prerequisite for this is that the off-axis dispersion of the loudspeakers is as similar as possible in frequency to the on-axis. Most proponents of this school advance half-baked box speakers with wide dispersion, but there are people who go further and advocate omnis or semi-omnis (such as D. Moulton and John Watkinson).
2. There is an experience-based claim - with less empirical research to back it up - that says the exact opposite. According to this school of thought, early reflections can significantly degrade imaging. This is claimed by dr. Geddes, but also by other manufacturers such as Kii, B&O/Geoff Martin, Dutch & Dutch, etc. This view is also very dominant among recording and studio professionals. My anecdotal impression is that this view also is rather dominant among audiophiles these days.
There is at present less psychoacoustic research to support this claim, but as dr. Geddes points out there hasn't been done that much research on the outcome of interest for people who adhere to this view - namely, imaging.
Can these two tendencies be reconciled? It is possible, of course, that people in one of the camps are 100% mistaken, and that people in the other camp are 100% right, and/or that the listening impressions of the people in the "wrong" camp have no objective validity at all.
I would be skeptical of that interpretation. My hunch would be that both of these camps are on to something objective. That's why it felt so illuminating to read dr. Geddes' comments on studio recordings vs. field recordings, and the different things one tries to recreate. I suspect that the wider the dispersion (assuming it is even with frequency and ideally point-source), the better it would be for getting the musicians into one's own room - for fusing the recording field with one's own listening space. I also suspect that reverberant recordings - typically classical music and acoustic jazz in the purist fashion etc - will benefit much from this kind of setup. It is about recreating "real" acoustic events in one's own room.
BUT if the goal is to recreate the artificial imaging of a typical studio recording, and to look into a different room alltogether - not to get the musicians to play in your very own room - then highly directional speakers will work better, and it will be beneficial to attenuate early reflections as much as one can.
Does this make sense?
I'll try to sum up this debate how I see it:
1. There is a fairly strong line of research which claims that early reflections are mostly beneficial - for envelopment, spaciousness, listening pleasure, etc. Toole and Olive, but also others have come to the some conclusion. A prerequisite for this is that the off-axis dispersion of the loudspeakers is as similar as possible in frequency to the on-axis. Most proponents of this school advance half-baked box speakers with wide dispersion, but there are people who go further and advocate omnis or semi-omnis (such as D. Moulton and John Watkinson).
2. There is an experience-based claim - with less empirical research to back it up - that says the exact opposite. According to this school of thought, early reflections can significantly degrade imaging. This is claimed by dr. Geddes, but also by other manufacturers such as Kii, B&O/Geoff Martin, Dutch & Dutch, etc. This view is also very dominant among recording and studio professionals. My anecdotal impression is that this view also is rather dominant among audiophiles these days.
There is at present less psychoacoustic research to support this claim, but as dr. Geddes points out there hasn't been done that much research on the outcome of interest for people who adhere to this view - namely, imaging.
Can these two tendencies be reconciled? It is possible, of course, that people in one of the camps are 100% mistaken, and that people in the other camp are 100% right, and/or that the listening impressions of the people in the "wrong" camp have no objective validity at all.
I would be skeptical of that interpretation. My hunch would be that both of these camps are on to something objective. That's why it felt so illuminating to read dr. Geddes' comments on studio recordings vs. field recordings, and the different things one tries to recreate. I suspect that the wider the dispersion (assuming it is even with frequency and ideally point-source), the better it would be for getting the musicians into one's own room - for fusing the recording field with one's own listening space. I also suspect that reverberant recordings - typically classical music and acoustic jazz in the purist fashion etc - will benefit much from this kind of setup. It is about recreating "real" acoustic events in one's own room.
BUT if the goal is to recreate the artificial imaging of a typical studio recording, and to look into a different room alltogether - not to get the musicians to play in your very own room - then highly directional speakers will work better, and it will be beneficial to attenuate early reflections as much as one can.
Does this make sense?
Last edited:
I've looked at a room as presented by Dave Moulton, my question would be, what does he aim for as far as early side wall reflections go.
Moulton's Takes
maybe but the two camps really are exemplified by:
therefore those are unsuitable for serious audio, perhaps for audiophools
vs
Bob Olhsson at gearslutz said:
*Full disclosure up front*: My own listening experience has placed me firmly in the second camp.
IMO, it does. In fact, this is exactly consistent with my own experience.
It's the whole "they are here" vs. "you are there" thing.
BUT, personally, I'd add that to me, even acoustic recordings sound more interesting when listened to using directional speakers, because they seem to let me hear more 'into' the recording. Omnis, instead, appear to me to tend to impart the same type of 'signature' to all recordings, which after a while just sounds contrived to me.
Marco
Except that if we do that, then the same listening room's own sound signature is always superimposed and inextricably added into the mix, which I believe is at the root of my own listening impression when listening to omnis.
Like I said before, to me, this may sound 'nice' in the very beginning (especially, say, on choral music), but after a short while it begins to sound contrived, and when that happens then this effect is exposed as somewhat of a gimmick, and the speakers that create it as a bit of a 'one trick pony'.
From the same web page:
Here's another statement (the part that I highlighted in bold font) that I strongly disagree with.
It boils down to music production (by live instruments) vs. music reproduction (by stereo loudspeakers).
IMO, the only case in which the loudspeakers' dispersion should - in theory - be made to coincide with that of the musical instrument(s) they are reproducing would be if all we listened to were anechoic recordings.
And even if that were the case (and disregarding for a moment the fact that all instruments are different in their own dispersion characteristics), by doing so we would completely forego the possibility of even attempting to recreate the impression of different acoustic spaces, and all music would sound as if played in our listening room.
Here's another statement (the part that I highlighted in bold font) that I strongly disagree with.
It boils down to music production (by live instruments) vs. music reproduction (by stereo loudspeakers).
IMO, the only case in which the loudspeakers' dispersion should - in theory - be made to coincide with that of the musical instrument(s) they are reproducing would be if all we listened to were anechoic recordings.
And even if that were the case (and disregarding for a moment the fact that all instruments are different in their own dispersion characteristics), by doing so we would completely forego the possibility of even attempting to recreate the impression of different acoustic spaces, and all music would sound as if played in our listening room.
Earl, would this nice picture suffice? Way out of my depth here but I've seen you speak of the gammatone model before and I've wondered if this might be a good basis for a freq dependant window for in room measurements. Might actually get us "close" to what we actually hear. Of course I'm not going to do speaker correction based on that. As it is there are so many different ways to display that data in REW it's ridiculous. I can easily set it up so there is a nice in room curve or, with a few clicks, make it look awful just by changing the sliding window. In this case I set it up so that it follows what I think I hear but that's certainly not definite.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.