tuga
Major Contributor
The impulse response of Reflection-Free Zone studio:
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Evidence-based Speaker Designs
- Thread starter Ilkless
- Start date
- Joined
- Jun 19, 2018
- Messages
- 6,652
- Likes
- 9,410
@maverickronin, I've been trying to find research along the lines you suggested, and in fact it's amazingly difficult to find anything remotely similar.
I'm sure (although increasingly less so) there must be studies out there, but all I could find were either:
(1) According to research by Blauert and others, in the case of a primary sound with multiple reflections, a given reflection is less likely to be audible or to cause image spread/shifting if additional reflections are present between the primary sound and this particular reflection. In the example given in his book, the echo threshold for a particular reflection at a particular intensity was extended from 10ms to 20-30ms when preceding reflections were added.
Another way to put this would be to say that a given reflection in a room is less likely to have an effect on image localisation/spread/shift if it follows preceding reflections. So e.g. a sidewall reflection is less likely to be audible if it follows a floor reflection, and a back wall reflection is less likely again to be audible if it follows a floor reflection and ceiling/sidewall reflections etc.
(2) Also according to Blauert, all else equal, there is a correlation between the "locatedness" of a primary sound and the degree of coherence of a delayed sound. Translating this to room acoustics (with some assumptions), one could infer that the more correlated a reflection is to the direct sound, the less likely it is to be audible/shift/spread the perceived image.
(3) The most relevant paper to your specific point is unfortunately not a listening study, but a theoretical hypothesis put forward by Angus, who modelled the ITD differences of reflections vs direct stereo sound for angles from -30° to +30° and frequencies below 700Hz (i.e. the range in which ITDs predominate in human sound localisation).
To explain what's happening here, you need to appreciate that, when a stereo signal is panned to some position between L and R (other than the centre position), the the ipsilateral signal at each ear sums with the contralteral signal (the crosstalk) in such a way that the resultant sum signal mimics the effects of an interaural time difference cue from a real acoustic source, as illustrated here on the right:
It's not often appreciated that pan-pot/Blumlein stereo has this characteristic, i.e. that by modifying interchannel level differences at the speakers, interaural time difference cues are created at the ears of the listener (I hope the above diagram illustrates this adequately).
A similar calculation to that shown in the picture above (right) can be performed for not only direct stereophonic sound, but also for a given reflection. Angus caclulated the degree of error that would be created in this ITD cue for a reflection from a given direction.
These data are shown below:
As you can see, while the degree of error is low for floor/ceiling reflections, the degree of error is gross for lateral reflections.
Angus further calculated that an absorption coefficient of only 0.5 would be required to bring the degree of error for floor/ceiling to a level below the degree of error measured for human listeners for a direct sound in an anechoic chamber (he says "only", but 0.5 is no mean feat at <100Hz!).
Meanwhile, an absorption coefficient of 0.99 (essentially removing the reflection entirely) would be required for lateral reflections.
However, Angus did not test actual human subjects to determine whether this error in the ITD's created by lateral reflections actually affected image localisation.
There would at least be a strong argument that, on the basis of the precedence effect (especially as formulated in the "law of the first wavefront" mentioned above), directional cues will be determined entirely from the first arriving wavefront, i.e. the direct sound, which will of course contain the correct ITD cues.
To cut a long story short, I think some experiments along the lines you've suggested would be very interesting
I'm sure (although increasingly less so) there must be studies out there, but all I could find were either:
- studies concerning localisation of a mono source with presence/intensity of reflections as a variable
- studies concerning localisation of a phantom source under either reflective or anechoic conditions (but none in which the presence/intensity of reflections was a variable).
(1) According to research by Blauert and others, in the case of a primary sound with multiple reflections, a given reflection is less likely to be audible or to cause image spread/shifting if additional reflections are present between the primary sound and this particular reflection. In the example given in his book, the echo threshold for a particular reflection at a particular intensity was extended from 10ms to 20-30ms when preceding reflections were added.
Another way to put this would be to say that a given reflection in a room is less likely to have an effect on image localisation/spread/shift if it follows preceding reflections. So e.g. a sidewall reflection is less likely to be audible if it follows a floor reflection, and a back wall reflection is less likely again to be audible if it follows a floor reflection and ceiling/sidewall reflections etc.
(2) Also according to Blauert, all else equal, there is a correlation between the "locatedness" of a primary sound and the degree of coherence of a delayed sound. Translating this to room acoustics (with some assumptions), one could infer that the more correlated a reflection is to the direct sound, the less likely it is to be audible/shift/spread the perceived image.
(3) The most relevant paper to your specific point is unfortunately not a listening study, but a theoretical hypothesis put forward by Angus, who modelled the ITD differences of reflections vs direct stereo sound for angles from -30° to +30° and frequencies below 700Hz (i.e. the range in which ITDs predominate in human sound localisation).
To explain what's happening here, you need to appreciate that, when a stereo signal is panned to some position between L and R (other than the centre position), the the ipsilateral signal at each ear sums with the contralteral signal (the crosstalk) in such a way that the resultant sum signal mimics the effects of an interaural time difference cue from a real acoustic source, as illustrated here on the right:
It's not often appreciated that pan-pot/Blumlein stereo has this characteristic, i.e. that by modifying interchannel level differences at the speakers, interaural time difference cues are created at the ears of the listener (I hope the above diagram illustrates this adequately).
A similar calculation to that shown in the picture above (right) can be performed for not only direct stereophonic sound, but also for a given reflection. Angus caclulated the degree of error that would be created in this ITD cue for a reflection from a given direction.
These data are shown below:
As you can see, while the degree of error is low for floor/ceiling reflections, the degree of error is gross for lateral reflections.
Angus further calculated that an absorption coefficient of only 0.5 would be required to bring the degree of error for floor/ceiling to a level below the degree of error measured for human listeners for a direct sound in an anechoic chamber (he says "only", but 0.5 is no mean feat at <100Hz!).
Meanwhile, an absorption coefficient of 0.99 (essentially removing the reflection entirely) would be required for lateral reflections.
However, Angus did not test actual human subjects to determine whether this error in the ITD's created by lateral reflections actually affected image localisation.
There would at least be a strong argument that, on the basis of the precedence effect (especially as formulated in the "law of the first wavefront" mentioned above), directional cues will be determined entirely from the first arriving wavefront, i.e. the direct sound, which will of course contain the correct ITD cues.
To cut a long story short, I think some experiments along the lines you've suggested would be very interesting
Last edited:
tuga
Major Contributor
Since stereo consists in creating phantom sources between the left and right speakers I don't think that it makes sense to extrapolate any observations concerning a mono source.
That's very interesting, though I'm not exactly sure what that graph is supposed to be plotting. Is the Y axis the angle the phantom image is supposed to be offset by a given reflection? And then is each curve is the angle of the reflection?
Also this would seem to take us back to "all reflections are bad".
Last edited:
- Joined
- Jun 19, 2018
- Messages
- 6,652
- Likes
- 9,410
The graph shows the angle of the ITD cue of the first lateral reflections plotted against frequency for a listener positioned between the two speakers.
The Y-axis is phantom image angle and the X-axis is frequency. Each line in the graph is the ITD cue of a given reflection for a phantom image of a given frequency.
For example, at 0° on both graphs (that is, when the phantom image is centred), both lateral and floor/ceiling ITD cues are completely correct (as you'd expect, because all path lengths are equal).
At 5° (that is, for phantom images panned 5° to either side), for floor/ceiling reflections, the ITD cues are close to correct (the bottom yellow trace, which varies from 4° to 6°)
But for lateral reflections of a phantom image panned 5° off centre, the variation in the ITD cues of the reflections varies from -7° at about 260Hz to 90° at about 100Hz, 160Hz, etc...
In other words, for floor/ceiling reflections, the relatively straight lines on the graph indicate that the ITD cues contained in the reflections are quite similar to the ITD cues contained in the direct sound at all frequencies/off-axis angles.
For lateral reflections on the other hand, the ITD cues vary wildly with frequency, and do not track the ITD cue contained in the direct sound.
Well, I was close enough for government work.
That graph for lateral reflections is so ridiculously extreme though. It seems like the ITD error must be significantly mitigated by other psychoacoustic factors or else speakers in any normal room wouldn't image at all.
My sensory evaluation below, having something with which to compare.
My Opinion:
Long ago, at one point in our play switching between the Martin Logans and the little JBLs when they were new...
Audio Buddy and I were all but unable to identify when the JBLs were playing out-of-phase vs in-phase with some stereo source music.
The low lateral dispersion panels didn't exhibit that unexpected result.
With mono pink source the JBL in/out of phase could be identified, but the ML had a much greater difference (tempted to say something like "rip your head off") when out of phase, as the "image" changed from virtually dead center to a discomforting 180 degree spread.
The reflections in my not specifically treated for lateral reflections room apparently blur the wide-dispersion "image" to an interesting degree. I have posted measurements which I interpret as contributing to that sensory effect, generally dismissed by those without a pair of beamy dipoles in their room with which to compare.
As for " It seems like the ITD error must be significantly mitigated by other psychoacoustic factors or else speakers in any normal room wouldn't image at all" --- I'll presume it is what folks are used to hearing, so it sounds "normal".
- Joined
- Jun 19, 2018
- Messages
- 6,652
- Likes
- 9,410
@maverickronin I have finally stumbled upon a study investigating the effects of early lateral reflections on image localisation with a stereo pair of loudspeakers.
Unfortunately, I can't get my hands on the original study, so all I have are descriptions of it in other literature.
The authors of this paper describe it as follows:
Toole goes into more detail in his book:
(@tuga and @RayDunzl you may also be interested.)
Unfortunately, I can't get my hands on the original study, so all I have are descriptions of it in other literature.
The authors of this paper describe it as follows:
Toole goes into more detail in his book:
(@tuga and @RayDunzl you may also be interested.)
Interesting. Seems like in a big enough room neither directivity nor first side wall reflections might matter all that much.
Hadn't got that far yet. I skipped ahead and read some of chapter 7 but I don't really have enough context to draw any conclusions.
- Joined
- Jun 19, 2018
- Messages
- 6,652
- Likes
- 9,410
Exactly. Taking the findings of the study at face value, the next question would be to ask where the threshold lies, i.e. how much narrower does a room need to be before early lateral reflections begin affecting phantom image localisation?
tuga
Major Contributor
One could arrive at that conclusion by resorting to good old common sense but it is nice to have a study which confirms my suspicions regarding large rooms or wide front walls.
For small rooms like my current one, in my anecdotal experience, wide dispersion splatters sound all over the place and I suspect that this could be creating unnecessary brain computation.
I have been working with rendering software for 24 years; they use ray tracing to create the efects of light over object and wall surface colour and texture.
Maybe one day someone will design a renderer that will calculate sound reflections.
- Joined
- Jun 19, 2018
- Messages
- 6,652
- Likes
- 9,410
I'm surprised that's your position now.
Earlier in the thread it was your position that any reflection earlier than 15ms must be absorbed or reflected away from the listening position to preserve proper imaging, if I'm not mistaken?
- Joined
- Jun 19, 2018
- Messages
- 6,652
- Likes
- 9,410
tuga
Major Contributor
I have defended from the beginning that narrow rooms need treatment early reflection zones in side walls or that dispersion need to be narrow.
Unless I'm mistaken, when the room is very large and/or the front wall is very wide early reflections coming from the side walls will be delayed and lower in level.
Perhaps Toole's study should have been performed in a narrow room.
Conspiracy theory: a study was performed but the results didn't please, thus study was not published?
- Joined
- Jun 19, 2018
- Messages
- 6,652
- Likes
- 9,410
In post #859 you said:
That seems to conflict with your more recent statement it is a matter of "common sense" that early reflections of 9ms delay should not interfere with imaging.
Perhaps Toole's study should have been performed in a narrow room.
Conspiracy theory: a study was performed but the results didn't please, thus study was not published?
Toole never studied this AFAIK. We're talking about a 2005 study by Choisel
About this Choisel study, Toole says:
tuga
Major Contributor
In post #859 you said:
That seems to conflict with your more recent statement it is a matter of "common sense" that early reflections of 9ms delay should not interfere with imaging.
Toole never studied this AFAIK. We're talking about a 2005 study by Choisel
About this Choisel study, Toole says:
Tax me, I got the numbers wrong.
- Joined
- Jun 19, 2018
- Messages
- 6,652
- Likes
- 9,410
Conspiracy theory: a study was performed but the results didn't please, thus study was not published?
Related:
File Drawer Effect
The idea that all early reflections below 15ms are extremely bad for the stereo image and inevitably lead to sound coloration is often found in forums.
Sometimes it is also demanded to suppress the reflections up to 15ms completely.
I think that a specification that is intended for studio recordings was transferred to the hifi world. For studio recordings the requirement is to avoid early reflections up to 15ms, because otherwise the recordings are "blurred" (without an increase in spatiality).
This is somehow plausible, since early reflections always occur when listening in living rooms, it is certainly not an advantage if the music signal contains additional early reflections.
A very important aspect of early reflections in living rooms is their masking. The negative effects of early reflections like comb filter effects or roughness and beating are masked by other reflections.
This means that a reduction of the reverberation time, for example, can lead to a stronger perception of the negative effects of early reflections.
Also very negative are single very early (little attenuated) reflections, such as those caused by a mixing console in the studio or by the living room table at home.
The whole context is much too complex for a simple rule like "avoid early reflections up to 15ms" or "...never feed it after midnight".
The idea comes from control room design, LEDE (live end dead end), RFZ (reflection free zone), non-environment room, Controlled Image Design etc etc. In a control room the engineer needs to hear the early reflections in the studio and recording, so they need to suppress the early reflections in the smaller listening space (control room). It was about being able to accurately hear what was being recorded, nothing about stereo image in the control room, and no explicit specification of 15ms as it depended on the early reflections in the recording.The idea that all early reflections below 15ms are extremely bad for the stereo image and inevitably lead to sound coloration is often found in forums.
Sometimes it is also demanded to suppress the reflections up to 15ms completely.
I think that a specification that is intended for studio recordings was transferred to the hifi world. For studio recordings the requirement is to avoid early reflections up to 15ms, because otherwise the recordings are "blurred" (without an increase in spatiality).
This is somehow plausible, since early reflections always occur when listening in living rooms, it is certainly not an advantage if the music signal contains additional early reflections.
A very important aspect of early reflections in living rooms is their masking. The negative effects of early reflections like comb filter effects or roughness and beating are masked by other reflections.
This means that a reduction of the reverberation time, for example, can lead to a stronger perception of the negative effects of early reflections.
Also very negative are single very early (little attenuated) reflections, such as those caused by a mixing console in the studio or by the living room table at home.
The whole context is much too complex for a simple rule like "avoid early reflections up to 15ms" or "...never feed it after midnight".
All that ctrl wrote above is true, and as far as I know there is little study around how the first 15ms of reflections affects stereo image in a stereo listening context. Most of the study seems to be around concert halls (or control rooms) and how clarity is determined (C50, C80 etc) because early energy versus late energy can impact on musical and speech intelligibility. But concert halls are completely different to stereo listening rooms, so I do not think much of that applies.
Last edited:
