What makes speakers "disappear " and can it be measured?

[Newman]

Have I misread the comments in this thread so far, or are they, instead of homing in on a majority consensus view on the cause, backed up with some evidence, in fact just expanding like a brain dump of ideas, soon to include the known and imaginary universe?

 

[dualazmak]

As a follow-up of my post #117 above, I am just testing the sound with several fur and down blankets as well as cushions between the SPs and the coffee table;

I have slightly better 3D sound perspectives and "SP disappearance", but with a little bit loss of bass sound which can be easily compensated by adjusting the gains for woofers and sub-woofers.

 

[Doodski]

As a follow-up of my post #117 above, I am just testing the sound with several fur and down blankets as well as cushions between the SPs and the coffee table;
View attachment 145009

I have slightly better 3D sound perspectives and "SP disappearance", but with a little bit loss of bass sound which can be easily compensated by adjusting the gain for woofers and sub-woofers.

With my previous tri-amp'd system I used sound absorption panels right behind my head. It improved everything. I see in your case that might be impractical. There was a wall directly behind my head as I sat on the sofa and I put the sound absorbing panels between the sofa and the wall.
These sort of sound absorbing partitions>>>

 

[dualazmak]

With my previous tri-amp'd system I used sound absorption panels right behind my head. It improved everything. I see in your case that might be impractical. There was a wall directly behind my head as I sat on the sofa and I put the sound absorbing panels between the sofa and the wall.
These sort of sound absorbing partitions>>>

Thank you for the nice information.

In my living room listening environment, I have another Japanese style tatami floor room, which is acoustically "well dead", behind me open-connected to the living/listening room;

I believe the rear "acoustically dead" space/room reacts just same as the sound absorption panels you have used in your system...

 

[dualazmak]

And as shared in my post here, I often place sound absorption sponge mattress in front of the white wall and the glass door leading to the corridor.

 

[Tangband]

It appears from the abstract that the ceiling had a negative effect on localisation; only the floor had a significant positive effect on especially height localisation. Which again suggests that a properly placed coffee table that blocks the floor reflection without adding new ones would decrease this localisation effect. Normally however, the floor bounce is there when some auditory event occurs in front of you. Either in real life or in the recording. Difference is however that the recording does not include the various angles. It will only relate to the height of the speakers.

Yes.
You dont want dual floor reflections - with real instruments playing, you always have the floor where the musicians are standing or sitting - its a part of the instruments timbre and it should be in the recording.

However, you dont want to have that ”twice” in your listening room . If you dont have a thick rug on the floor in the listening room , you gonna get floor reflections twice ( one from the recording event, one from listening ).
The result is bad sound.

 

[pogo]

Considering all the speakers I had, the most "invisible" ones were the ones with optimized enclosure (rounded corners, non-rectangular case), super low distortion and perfect timing of the chassis and crossovers.

From my experience, I can fully agree with you here. Super low distortion and perfect timing also plays a big role for the rest of the chain. I have had good experience with Purifi amps and with optimized and matched loudspeakers (x-overs).

 

[Thomas_A]

So far for ”disappering speakers”:

1. Speakers devoid of resonances and (larger?) frequency reponse irregularities. Probably certain frequencies are more revealing than others.
3. Early refections at sufficient level are bad.
3 Ratio of direct to late reflected sound.
4 Ratio of forward sound to side-wall reflections.
5. Floor reflection. If present it may reveal position, especially height localisation.
6. Lower treble - may relate to a more distant sound.

So to another question. Does the comb filtering effect of stereo setups contribute to localisation? This should only affect the central phantom image vs position of the sides. A speaker system that present a linear on-axis but total enegy whose reflections aid to even out the peaks and dips due to comb filtering. Will that disappear more or less? Or is it the other way around?

Frequency response corrections has already been mentioned for the floor reflection.

 

[Tangband]

I would like to add one thing to this interesting thread.

7. Cabinet coloration.

Some can be euphonic, like typical loudspeakers made of mdf material, that often has problems with resonances between 200-450 Hz . You can study this beaviour in the big number of loudspeakertests that John Atkinsson has made for stereophile.
Sometimes those resonances are even louder than the sound from the loudspeaker cones.
It makes the sound ”wooden” , in such a way that a cello sounds more alive and resonant. But its a coloration.

My experience with DIY building of loudspeakers is that if you can make a really stiff cabinet with cabinetresonances higher than 800 Hz its much easier to use damping material inside the loudspeaker to lessen the cabinet-effect.
Aluminum seems to be a good material . My Genelecs have its cabinet resonances above 1000 Hz, where itˋs low in amplitude.

Loudspeaker cabinets that has a lot of resonances between 100-400 Hz often makes it harder to hear what the pianist are playing in the left hand. The energy that comes from the cabinet is delayed energy thats gonna blur the perceived pitch.

A good cabinet with less resonances, in my experience, makes the perceived pitch of each piano tone clearer, and makes the whole loudspeaker dissapear.

 

[Tangband]

So far for ”disappering speakers”:

1. Speakers devoid of resonances and (larger?) frequency reponse irregularities. Probably certain frequencies are more revealing than others.
3. Early refections at sufficient level are bad.
3 Ratio of direct to late reflected sound.
4 Ratio of forward sound to side-wall reflections.
5. Floor reflection. If present it may reveal position, especially height localisation.
6. Lower treble - may relate to a more distant sound.

So to another question. Does the comb filtering effect of stereo setups contribute to localisation? This should only affect the central phantom image vs position of the sides. A speaker system that present a linear on-axis but total enegy whose reflections aid to even out the peaks and dips due to comb filtering. Will that disappear more or less? Or is it the other way around?

Frequency response corrections has already been mentioned for the floor reflection.

Good question. It would be very interesting with input from F. Toole , about this.

 

[sigbergaudio]

8. Consistent on- and off-axis response. This helps us understand reflections as reflections of the original sound, as opposed to a new sound, and thus ignore it.

 

[Floyd Toole]

Good question. It would be very interesting with input from F. Toole , about this.

Pleased to oblige with some highly simplified responses:
1. Speakers devoid of resonances and (larger?) frequency reponse irregularities. Probably certain frequencies are more revealing than others.
Detection thresholds of resonances are relatively independent of frequency but very dependent on the program material - pink noise is the most revealing. Low-Q resonances (least ringing) are more audible than higher Q for the same deviation in spectrum amplitude - counterintuitive, but evidence is that we don't attend to the ringing as much as we do to the spectral bump. See section 4.6.2 in the 3rd edition of my book, or Toole, F. E. and Olive, S.E. (1988). “The modification of timbre by resonances: perception and measurement”, J. Audio Eng. Soc., 36, pp. 122-142.
Re. cabinet resonances: they are included in the 3D spinorama measurements and are not better or worse than the equivalent resonant sounds radiated by any other component in the system.
3. Early refections at sufficient level are bad.
If the spectrum/timbre of a reflection is different from the direct sound it will be more easily detectable as a separate phenomenon. LEDE control rooms and the notion that lateral reflections must be eliminated came about as a result of loudspeakers having poor off-axis sound radiation. This seems to explain why loudspeakers with well behaved (smooth and fundamentally similar) off axis early reflected sounds (around 60 +/- degrees off axis) are awarded the highest sound quality ratings in double-blind listening tests - done in rooms with NO side-wall absorption. Here is an example of how bad things were when some of these practices were being added to the "rules" for good sound in control rooms, which migrated into homes. The UREI was a very popular monitor speaker of the period - obviously the far-off axis sounds need to be eliminated/absorbed before decent sound quality is possible. Fortunately things have greatly improved. With well-designed loudspeakers wide dispersion is generally much approved of, lending a friendly sense of space, especially for those soundstage components that are hard-panned to L & R speakers - i.e. mono sounds. See Section 7.4.2 in 3rd edition.
3 Ratio of direct to late reflected sound.
The important reflections need to have substantial time delays (longer than the short-delayed reflections in domestic rooms) to support illusions of great distance meaning that they need to be in the recordings. In stereo they don't get to be reproduced from the appropriate angles. This is the dominant advantage of multichannel - the difference in "envelopment" (the sense of being in a large space) can be profound.
4 Ratio of forward sound to side-wall reflections. See above.
5. Floor reflection. If present it may reveal position, especially height localisation. This will challenge your credibility. See Section 7.4.7 in the 3rd edition. The most definitive test I am aware of was done in the Fraunhofer Institute flexible acoustics room. They concluded: "Regarding the floor reflection, the audible influence by removing this with absorbers around the listener is negative - unnatural sounding. No normal room has an absorbent floor. The human brain seems to be used to this." Humans evolved with something reflective below the feet.
6. Lower treble - may relate to a more distant sound.
This would require familiarity with the sound at a known distance in order to recognize the sound at another distance. Air absorption is the physical mechanism responsible for the spectral change with distance - See Figure 10.12 in the 3rd edition. The perception of elevation is another case requiring familiarity with the sound, and even given some familiarity our precision in vertical localization is poor.

 

[Thomas_A]

8. Consistent on- and off-axis response. This helps us understand reflections as reflections of the original sound, as opposed to a new sound, and thus ignore it.

I would agree on general view. However speakers with good on-axis within say +/- 1.5 dB and trailing smooth off-axis may still diifer in terms of ”disappearance” depending on the distribution and Qs of the dips and peaks within those limits.

 

[sigbergaudio]

5. Floor reflection. If present it may reveal position, especially height localisation. This will challenge your credibility. See Section 7.4.7 in the 3rd edition. The most definitive test I am aware of was done in the Fraunhofer Institute flexible acoustics room. They concluded: "Regarding the floor reflection, the audible influence by removing this with absorbers around the listener is negative - unnatural sounding. No normal room has an absorbent floor. The human brain seems to be used to this." Humans evolved with something reflective below the feet.

So this indicates that having a large rug or wall-to-wall carpet may actually negatively affect the sound? And having a reflective (say hardwood) floor would be preferable?

 

[Floyd Toole]

So this indicates that having a large rug or wall-to-wall carpet may actually negatively affect the sound? And having a reflective (say hardwood) floor would be preferable?

I wish it were true that something as thin as a rug or carpet had any significant effect on a low-angle-of-incidence sound! Figure 7.6 in the 3rd edition is shown below and you can see that even for 2 inches of high-density fiberglass the absorption is substantial only at high frequencies when sound arrives from a moderate (45 deg) angle of incidence. Carpet will make a difference but it will really only change the timbre at high frequencies, not eliminate the reflection - the placebo effect may yield different results though, because it will reduce the overall reflected sound level in the room at higher frequencies. Bear in mind that high frequencies are usually substantially attenuated at off-axis angles relevant to floor reflections.

 

[sigbergaudio]

I wish it were true that something as thin as a rug or carpet had any significant effect on a low-angle-of-incidence sound! Figure 7.6 in the 3rd edition is shown below and you can see that even for 2 inches of high-density fiberglass the absorption is substantial only at high frequencies when sound arrives from a moderate (45 deg) angle of incidence. Carpet will make a difference but it will really only change the timbre at high frequencies, not eliminate the reflection - the placebo effect may yield different results though, because it will reduce the overall reflected sound level in the room at higher frequencies. Bear in mind that high frequencies are usually substantially attenuated at off-axis angles relevant to floor reflections.

I was aware that a thin carpet would have a limited effect. But as you imply, the visual impact of a carpet floor vs hard floor is big, so it's hard to intuitively grasp that the effect is that small.

I guess the placebo effect / experience of the room is good point. A full wall-to-wall carpet will likely calm / attenuate the energy in the room somewhat, making it more pleasant to sit and talk between songs. Perhaps to the point that it will positively impact the listening experience (more so than the actual sound) as well

 

[posvibes]

It has been my experience that the disappearing act has been best achieved by using infinite baffle (sealed) speakers. I don't have a skerrick of proof to back that position but it always seems to be a stark difference between that type of speaker as opposed to others. My preference is for sealed speakers but I am not fanatically adamant about it.

Closing my eyes while listeneing has been the other proven strategy I use to make speakers disappear.

 

[Thomas_A]

Pleased to oblige with some highly simplified responses:
1. Speakers devoid of resonances and (larger?) frequency reponse irregularities. Probably certain frequencies are more revealing than others.
Detection thresholds of resonances are relatively independent of frequency but very dependent on the program material - pink noise is the most revealing. Low-Q resonances (least ringing) are more audible than higher Q for the same deviation in spectrum amplitude - counterintuitive, but evidence is that we don't attend to the ringing as much as we do to the spectral bump. See section 4.6.2 in the 3rd edition of my book, or Toole, F. E. and Olive, S.E. (1988). “The modification of timbre by resonances: perception and measurement”, J. Audio Eng. Soc., 36, pp. 122-142.
Re. cabinet resonances: they are included in the 3D spinorama measurements and are not better or worse than the equivalent resonant sounds radiated by any other component in the system.

All agreed.

3. Early refections at sufficient level are bad.
If the spectrum/timbre of a reflection is different from the direct sound it will be more easily detectable as a separate phenomenon. LEDE control rooms and the notion that lateral reflections must be eliminated came about as a result of loudspeakers having poor off-axis sound radiation. This seems to explain why loudspeakers with well behaved (smooth and fundamentally similar) off axis early reflected sounds (around 60 +/- degrees off axis) are awarded the highest sound quality ratings in double-blind listening tests - done in rooms with NO side-wall absorption. Here is an example of how bad things were when some of these practices were being added to the "rules" for good sound in control rooms, which migrated into homes. The UREI was a very popular monitor speaker of the period - obviously the far-off axis sounds need to be eliminated/absorbed before decent sound quality is possible. Fortunately things have greatly improved. With well-designed loudspeakers wide dispersion is generally much approved of, lending a friendly sense of space, especially for those soundstage components that are hard-panned to L & R speakers - i.e. mono sounds. See Section 7.4.2 in 3rd edition.

*I agree fully andI enjoy the sound of a good and wide dispersion speaker.

3 Ratio of direct to late reflected sound.
The important reflections need to have substantial time delays (longer than the short-delayed reflections in domestic rooms) to support illusions of great distance meaning that they need to be in the recordings. In stereo they don't get to be reproduced from the appropriate angles. This is the dominant advantage of multichannel - the difference in "envelopment" (the sense of being in a large space) can be profound.
4 Ratio of forward sound to side-wall reflections. See above.

*Yes, agreed. And in small rooms it is difficult to get long time delays. So what you think about heavy toe-in to increase the lateral reflections of the opposing speaker wall and at the same time reducing the levels at the nearest speaker wall?

I use multichannel now and then as well, which works well with some but far from all recordings. Again, I think this is related to what model you wish to adapt to when listening to music. Do you want to be transferred to the event and ignore your room - go multichannel as far as possible. Do you want to transfer you room to the event, effectively using your room as a lounge with an transparent opening against the event? Go two- or three-channel audio, make sure that the wall behind the speakers are acoustically treated (within reasons!).

5. Floor reflection. If present it may reveal position, especially height localisation. This will challenge your credibility. See Section 7.4.7 in the 3rd edition. The most definitive test I am aware of was done in the Fraunhofer Institute flexible acoustics room. They concluded: "Regarding the floor reflection, the audible influence by removing this with absorbers around the listener is negative - unnatural sounding. No normal room has an absorbent floor. The human brain seems to be used to this." Humans evolved with something reflective below the feet.

*This was noted only according to another reference on the subject in this thread where I only read the abstract part:
https://pubmed.ncbi.nlm.nih.gov/2130378/
I have not made studied the subject in detail, but I agree that ground/floor is the only natural reflection in the open and should be expected. (And which may add to cues with respect to distance or height of the source.) The general recommendation here appears to have a carpet on the floor and the other question was the impact of a small coffee table blocking the primary floor reflection. With respect to timbre, reflections in the recording, including back wall, side walls, floor and roof is something that is already in many recordings (not all of course). So if I have a floor reflection in the recording, I will get the frequency response deviations already from start. The thing missing is of course the angle of reflection impossible to recreate with stereo. Having yet another floor reflection in my room, I will get an angle but also mask or get double effects of the recorded floor bounce and my own room. I am still not convinced that floor bounce is a good thing.

6. Lower treble - may relate to a more distant sound.
This would require familiarity with the sound at a known distance in order to recognize the sound at another distance. Air absorption is the physical mechanism responsible for the spectral change with distance - See Figure 10.12 in the 3rd edition. The perception of elevation is another case requiring familiarity with the sound, and even given some familiarity our precision in vertical localization is poor.

So how familiar should it be? A human voice in general or a particular human voice known to us?

 

[Floyd Toole]

I wish it were true that something as thin as a rug or carpet had any significant effect on a low-angle-of-incidence sound! Figure 7.6 in the 3rd edition is shown below and you can see that even for 2 inches of high-density fiberglass the absorption is substantial only at high frequencies when sound arrives from a moderate (45 deg) angle of incidence. Carpet will make a difference but it will really only change the timbre at high frequencies, not eliminate the reflection - the placebo effect may yield different results though, because it will reduce the overall reflected sound level in the room at higher frequencies. Bear in mind that high frequencies are usually substantially attenuated at off-axis angles relevant to floor reflections.

Oops, I forgot the attachment. Here it is:

 

[Floyd Toole]

All agreed.


*I agree fully andI enjoy the sound of a good and wide dispersion speaker.


*Yes, agreed. And in small rooms it is difficult to get long time delays. So what you think about heavy toe-in to increase the lateral reflections of the opposing speaker wall and at the same time reducing the levels at the nearest speaker wall?

I use multichannel now and then as well, which works well with some but far from all recordings. Again, I think this is related to what model you wish to adapt to when listening to music. Do you want to be transferred to the event and ignore your room - go multichannel as far as possible. Do you want to transfer you room to the event, effectively using your room as a lounge with an transparent opening against the event? Go two- or three-channel audio, make sure that the wall behind the speakers are acoustically treated (within reasons!).



*This was noted only according to another reference on the subject in this thread where I only read the abstract part:
https://pubmed.ncbi.nlm.nih.gov/2130378/
I have not made studied the subject in detail, but I agree that ground/floor is the only natural reflection in the open and should be expected. (And which may add to cues with respect to distance or height of the source.) The general recommendation here appears to have a carpet on the floor and the other question was the impact of a small coffee table blocking the primary floor reflection. With respect to timbre, reflections in the recording, including back wall, side walls, floor and roof is something that is already in many recordings (not all of course). So if I have a floor reflection in the recording, I will get the frequency response deviations already from start. The thing missing is of course the angle of reflection impossible to recreate with stereo. Having yet another floor reflection in my room, I will get an angle but also mask or get double effects of the recorded floor bounce and my own room. I am still not convinced that floor bounce is a good thing.



So how familiar should it be? A human voice in general or a particular human voice known to us?

Elevation information is dominantly spectral (HRTFS). The "rules" I concluded with in the relevant section (15.12.1) of my book are:
1. the sound must contain frequencies above about 6 kHz, and listeners must be able to hear them.
2. wide bandwidth and a dense spectrum help reveal the elevation cue in the spectrum.
3. familiarity with the sound is important.

Jens Blauert investigated vertical "localization blur" and concluded that the angular uncertainty was:
- about 17 degrees for an unfamiliar voice
- about 9 degrees for a familiar voice
- about 4 degrees for pink noise.
Plausibility plays a huge role - nothing is ever localized below the floor and helicopters tend to be localized above even without elevation speakers.