https://users.aalto.fi/~ktlokki/Publs/JASMAN_vol_146_iss_5_3562_1.pdf
Selected excerpts:
"Ten assessors participated in the experiment as volunteers. The assessors are considered as expert listeners, with an average experience of 11.4 years [standard deviation (s.d.) 6 5.5] in acoustics research and development, spatial audio reproduction, and critical listening as part of their profession. All assessors reported proficiency in standard audio evaluation procedures. Seven assessors were familiar with standard sensory analysis protocols for evaluation of audio material, of which six had performed attribute elicitation procedures before."
"In the current study, the assessors were given the instruction of: “Imagine that you are in a typical residential room, listening to a 2-ch stereophonic reproduction over loudspeakers. Please rank the presented stimuli in a way that expresses your preference from ‘highly dislike’ to 'highly like'."
"The program materials included: (1) Shola Ama—You might need somebody, 0:09–0:25, (2) Anechoic Kongas African Rhythms—B&O Music For Archimedes, 0:24–0:33, (3) Female Speech English—EBU SQAM, 0:00–0:15."
"Rooms A, B, and D, comprise of wooden floors and ceiling, and acoustically treated rigid boundaries.24 They all include ordinary furniture and fully comply with IEC:60268-1311 specifications. Room C is a critical listening space10 built to host multichannel reproduction layouts.25 It is characterized by larger volumetric size, highly absorptive boundaries, no furniture, and a lower reverberation time (RT) than a typical room...In order to capture a wider range of possible acoustical fields than Rooms A–D, the interior of Room D was physically modified. Room D has a modular structure with adaptable acoustic panels which allowed the successive modification of the boundaries’ properties, such as the addition/removal of absorptive, diffusive, or reflective acoustic panels in all boundaries, including the ceiling. The modifications varied the RT30 as uniformly as possible across frequency. Modifications in the lateral plane were completed in symmetry. The reproduction system was fixed and the direct paths between the source, ceiling, floor, side walls, and the receiver were not modified. This ensured constant contribution of the first reflection points in these measurements, as shown in Table I."
"The data indicate that assessors were able to identify four perceptual constructs,59 and appropriately quantify the differences within the stimuli. In the first dimension, the factorial drivers relate to the perceived reverberance, width and envelopment, and proximity. The second dimension relates to the perceived bass as described by the assessors and the perceptually-relevant metrics...In our data, reverberance and width and envelopment have been identified as two separate perceptual constructs, yet the direction of the inertia explained by them on the major dimensions is similar."
"This analysis supports previous findings,60 that assessors systematically preferred the sound fields with lower RT. In our study, the most preferred acoustical conditions presented fields that evoked the sense of being less reverberant and less wide and enveloping. The sources were perceived as closer to the listener, exhibiting high levels of proximity."
"For example, one could attempt to alter the DRR (direct-to-reverberant ratio) within a field by means of directivity control in the loudspeakers, aiming to evoke certain perceptual aspects that would otherwise be dominated by the room’s natural acoustical field."
It's interesting to me that rooms C, D1, and D2 were the most preferred, when rooms A and B don't look so dissimilar to D2 at first glance. They do have slightly lower DRRs, but I wonder whether the stronger ceiling reflections worked against them, even though room B had weaker lateral reflections. Soren Bech was one of the authors of the 2019 paper, and one of his previous papers (the Archimedes paper: https://orbit.dtu.dk/files/4415806/Bech.pdf) had noted "The results have shown that the first-order floor and ceiling reflections are likely to individually contribute to the timbre of' reproduced speech. For a noise signal, additional reflections from the left sidewall will contribute individually. The level of the reve[berant field has been found to have an effect on the contribution of the individual reflections. An increase in the level of individual reflections are most likely to be audible for the first-order floor and ceiling reflections."
I also noticed that D3-6 had increasing RT30s with relatively constant first reflection point attenuation with respect to lateral, ceiling, and floor reflections. I could not tell if that was because of treatment of the front and rear wall or because of conversion of absorption to diffusion on the lateral, ceiling, and floor reflections.
As a layman, I have had the vaguest mental outlines of a possible framework for approaching listener preferences, which I mentioned here (https://www.audiosciencereview.com/...-audiophile-journey.14547/page-22#post-467125) after reading this other paper by Lokki et al (https://users.aalto.fi/~ktlokki/Publs/JASMAN_vol_140_iss_1_551_1.pdf):
1. "reverberance, loudness and width"
2. "definition and clarity"
3. "timbre"
I'm sorry that this isn't terribly scientific at all, but #1 and #2 does seem to correlate with Toole's discussion of audio professionals (I believe the assessors in the 2019 Lokki paper have similar preferences) and the role of lateral reflections. I don't know if "proximity" in the 2019 Lokki paper may approximate "definition and clarity" in the 2016 Lokki one. If so, perhaps we might further refine preferences and applications as:
1. "reverberance, loudness and width": wider dispersion loudspeakers? Less absorption? This would be reflected in the longer RT30 (with a similar trend in DRR but not specifically the lateral reflections) of the less preferred of rooms D2-6 in the 2019 Lokki paper.
2. "definition and clarity": narrower directivity speakers? Absorb lateral reflections or all first reflections, or mind the ISD gap?
3. "timbre"
For #3, I can only my speculations that I suspect that this may benefit from a relatively flat directivity index covering the majority of the key fundamental frequencies of most instruments, so starting perhaps as low as 100-200 Hz and extending to several thousand Hz. If wide baffle designs avoid the diffraction loss requiring compensation, maybe they have a flatter (what Toole might call "relatively constant") directivity index through most of this range compared with narrow baffle designs, which tend to have a rising (preferably what Toole calls "smoothly changing") directivity index through this range and thus may produce reflections with a frequency-sloping timbre compared with the direct signal, perhaps not sounding as natural or life-like in that respect for some listeners. For example, compare the JBL LSR6332 (https://jblpro.com/en/site_elements/lsr6332-spec-sheet) with its directivity index flat at around 5 dB between 300-1200 Hz and 7 dB between 1.5-6 kHz (so overall quite "flattish," as I think I've seen Toole use the term) with the Revel Salon 2 (https://speakerdata2034.blogspot.com/2019/03/spinorama-data-revel-ultima.html) with its series of changes in DI over this region. The so-called floor bounce dip falls into the lowest part of this key fundamental frequency range, which critics like Robert E Greene might describe as robbing orchestral music of its "weight" or power.
So, perhaps:
1. "reverberance, loudness and width": wider dispersion loudspeakers? Less absorption? This would be reflected in the longer RT30 (with a similar trend in DRR but not specifically the lateral reflections) of the less preferred of rooms D2-6 in the 2019 Lokki paper.
2. "definition and clarity": narrower directivity speakers? Absorb lateral reflections or all first reflections, or mind the ISD gap?
3. "timbre": wider baffle or boundary-adjacent designs? Anticipate floor bounce issues with specified speaker height, multiple or boundary-adjacent (or even cardioid radiation pattern) bass drivers? Treat ceiling reflections, possibly floor (though that is somewhat controversial)?
One could further distinguish #1 with early vs late reflections, also by differentiating direction of reflections, and as I said before, these preferences aren't necessarily mutually exclusive. For example, one could combine narrow directivity speakers with a delayed, less directional signal to add back perception of reverberance? The GGNTKT M1 might be described as prioritizing #3, then #1; the D&D 8C #2, then #3? Perhaps dipolar speakers try to balance #1 and #2 equally?
Anyway, just inchoate musings, based very loosely on science. Have a nice weekend!
Selected excerpts:
"Ten assessors participated in the experiment as volunteers. The assessors are considered as expert listeners, with an average experience of 11.4 years [standard deviation (s.d.) 6 5.5] in acoustics research and development, spatial audio reproduction, and critical listening as part of their profession. All assessors reported proficiency in standard audio evaluation procedures. Seven assessors were familiar with standard sensory analysis protocols for evaluation of audio material, of which six had performed attribute elicitation procedures before."
"In the current study, the assessors were given the instruction of: “Imagine that you are in a typical residential room, listening to a 2-ch stereophonic reproduction over loudspeakers. Please rank the presented stimuli in a way that expresses your preference from ‘highly dislike’ to 'highly like'."
"The program materials included: (1) Shola Ama—You might need somebody, 0:09–0:25, (2) Anechoic Kongas African Rhythms—B&O Music For Archimedes, 0:24–0:33, (3) Female Speech English—EBU SQAM, 0:00–0:15."
"Rooms A, B, and D, comprise of wooden floors and ceiling, and acoustically treated rigid boundaries.24 They all include ordinary furniture and fully comply with IEC:60268-1311 specifications. Room C is a critical listening space10 built to host multichannel reproduction layouts.25 It is characterized by larger volumetric size, highly absorptive boundaries, no furniture, and a lower reverberation time (RT) than a typical room...In order to capture a wider range of possible acoustical fields than Rooms A–D, the interior of Room D was physically modified. Room D has a modular structure with adaptable acoustic panels which allowed the successive modification of the boundaries’ properties, such as the addition/removal of absorptive, diffusive, or reflective acoustic panels in all boundaries, including the ceiling. The modifications varied the RT30 as uniformly as possible across frequency. Modifications in the lateral plane were completed in symmetry. The reproduction system was fixed and the direct paths between the source, ceiling, floor, side walls, and the receiver were not modified. This ensured constant contribution of the first reflection points in these measurements, as shown in Table I."
"The data indicate that assessors were able to identify four perceptual constructs,59 and appropriately quantify the differences within the stimuli. In the first dimension, the factorial drivers relate to the perceived reverberance, width and envelopment, and proximity. The second dimension relates to the perceived bass as described by the assessors and the perceptually-relevant metrics...In our data, reverberance and width and envelopment have been identified as two separate perceptual constructs, yet the direction of the inertia explained by them on the major dimensions is similar."
"This analysis supports previous findings,60 that assessors systematically preferred the sound fields with lower RT. In our study, the most preferred acoustical conditions presented fields that evoked the sense of being less reverberant and less wide and enveloping. The sources were perceived as closer to the listener, exhibiting high levels of proximity."
"For example, one could attempt to alter the DRR (direct-to-reverberant ratio) within a field by means of directivity control in the loudspeakers, aiming to evoke certain perceptual aspects that would otherwise be dominated by the room’s natural acoustical field."
It's interesting to me that rooms C, D1, and D2 were the most preferred, when rooms A and B don't look so dissimilar to D2 at first glance. They do have slightly lower DRRs, but I wonder whether the stronger ceiling reflections worked against them, even though room B had weaker lateral reflections. Soren Bech was one of the authors of the 2019 paper, and one of his previous papers (the Archimedes paper: https://orbit.dtu.dk/files/4415806/Bech.pdf) had noted "The results have shown that the first-order floor and ceiling reflections are likely to individually contribute to the timbre of' reproduced speech. For a noise signal, additional reflections from the left sidewall will contribute individually. The level of the reve[berant field has been found to have an effect on the contribution of the individual reflections. An increase in the level of individual reflections are most likely to be audible for the first-order floor and ceiling reflections."
I also noticed that D3-6 had increasing RT30s with relatively constant first reflection point attenuation with respect to lateral, ceiling, and floor reflections. I could not tell if that was because of treatment of the front and rear wall or because of conversion of absorption to diffusion on the lateral, ceiling, and floor reflections.
As a layman, I have had the vaguest mental outlines of a possible framework for approaching listener preferences, which I mentioned here (https://www.audiosciencereview.com/...-audiophile-journey.14547/page-22#post-467125) after reading this other paper by Lokki et al (https://users.aalto.fi/~ktlokki/Publs/JASMAN_vol_140_iss_1_551_1.pdf):
1. "reverberance, loudness and width"
2. "definition and clarity"
3. "timbre"
I'm sorry that this isn't terribly scientific at all, but #1 and #2 does seem to correlate with Toole's discussion of audio professionals (I believe the assessors in the 2019 Lokki paper have similar preferences) and the role of lateral reflections. I don't know if "proximity" in the 2019 Lokki paper may approximate "definition and clarity" in the 2016 Lokki one. If so, perhaps we might further refine preferences and applications as:
1. "reverberance, loudness and width": wider dispersion loudspeakers? Less absorption? This would be reflected in the longer RT30 (with a similar trend in DRR but not specifically the lateral reflections) of the less preferred of rooms D2-6 in the 2019 Lokki paper.
2. "definition and clarity": narrower directivity speakers? Absorb lateral reflections or all first reflections, or mind the ISD gap?
3. "timbre"
For #3, I can only my speculations that I suspect that this may benefit from a relatively flat directivity index covering the majority of the key fundamental frequencies of most instruments, so starting perhaps as low as 100-200 Hz and extending to several thousand Hz. If wide baffle designs avoid the diffraction loss requiring compensation, maybe they have a flatter (what Toole might call "relatively constant") directivity index through most of this range compared with narrow baffle designs, which tend to have a rising (preferably what Toole calls "smoothly changing") directivity index through this range and thus may produce reflections with a frequency-sloping timbre compared with the direct signal, perhaps not sounding as natural or life-like in that respect for some listeners. For example, compare the JBL LSR6332 (https://jblpro.com/en/site_elements/lsr6332-spec-sheet) with its directivity index flat at around 5 dB between 300-1200 Hz and 7 dB between 1.5-6 kHz (so overall quite "flattish," as I think I've seen Toole use the term) with the Revel Salon 2 (https://speakerdata2034.blogspot.com/2019/03/spinorama-data-revel-ultima.html) with its series of changes in DI over this region. The so-called floor bounce dip falls into the lowest part of this key fundamental frequency range, which critics like Robert E Greene might describe as robbing orchestral music of its "weight" or power.
So, perhaps:
1. "reverberance, loudness and width": wider dispersion loudspeakers? Less absorption? This would be reflected in the longer RT30 (with a similar trend in DRR but not specifically the lateral reflections) of the less preferred of rooms D2-6 in the 2019 Lokki paper.
2. "definition and clarity": narrower directivity speakers? Absorb lateral reflections or all first reflections, or mind the ISD gap?
3. "timbre": wider baffle or boundary-adjacent designs? Anticipate floor bounce issues with specified speaker height, multiple or boundary-adjacent (or even cardioid radiation pattern) bass drivers? Treat ceiling reflections, possibly floor (though that is somewhat controversial)?
One could further distinguish #1 with early vs late reflections, also by differentiating direction of reflections, and as I said before, these preferences aren't necessarily mutually exclusive. For example, one could combine narrow directivity speakers with a delayed, less directional signal to add back perception of reverberance? The GGNTKT M1 might be described as prioritizing #3, then #1; the D&D 8C #2, then #3? Perhaps dipolar speakers try to balance #1 and #2 equally?
Anyway, just inchoate musings, based very loosely on science. Have a nice weekend!
