Monophonic vs. Stereophonic - Timbre change?

[ctrl]

I think it is worth to discuss this topic in more detail, because among others @amirm's subjective ratings and the Harman-Olive score are based on monophonic listening.

In the course of the discussion about a "blind-test preparation", I have described what difference it makes, in my experience, to hear a speaker freestanding in mono compared to stereo.

************** can be skipped ****************
To give a little bit of context, it was part of my criticism of Harman/Olive's listening tests with monophonic, completely free standing speakers, compared to the tests Toole did in 1984-86, with a speaker setup comparable to a typical listening room. If you want to do this to yourself, you can read it here:
post#246
post#317
Don't panic, I'm not claiming that the results of Harman/Olive are wrong, but that they are not simply transferable to typical listening rooms, where lateral reflections usually play a major role.
************** can be skipped end ****************

But here in this thread only one aspect of my statements shall be discussed, which brought me harsh criticism:

Several effects come together. Because the loudspeaker stands freely in the room, the influence of lateral reflections is greatly reduced - with corresponding consequences if the loudspeaker (and the lateral walls) does not show an "optimal" radiation pattern.

My experience is that if a free-standing loudspeaker, listened to in mono, is adjusted to the best possible transparency of the sound, then it sounds too bright when listened to in stereo.

This statement is of course only anecdotal, as it only reflects my experience in the development of loudspeakers. Therefore criticism was not long in coming:

No.

Simple fact is that they have tested the same speakers in mono and stereo configurations on multiple occasions. To quote Toole "without exception" the same speakers were preferred in both scenerios. This is the case regardless of what you are saying the speaker/room position configurations were.

I'm not really interested in arguing with forum members about this. It's a bit tedious TBH. Your opinion simply doesnt carry weight against that of Toole et Al.
....
Toole himself explicitly says in that video that without exception every time they tested in stereo the same speakers were preferred as mono.

I have not checked, but I assume that the forum user quoted Toole correct.
Unfortunately I do not know of any other papers from Toole or Olive that shed more light on the aspect of monophonic and stereophonic hearing and possible timbre change.


Here is a AES paper by Bernd Theiss "Localization Experiments in Three-Dimensional Sound Reproduction" (1996) that shows that there might be something to the change in timbre of mono/stereo listening.

How reliable are the statements in this paper? Unfortunately it is based on only five test participants, which is even less than the 8-10 participants in Tool's paper from 1984-86.
Nevertheless, it is "more valuable" than general statements from individuals.

The test setup is very simple. A monophonic signal is reproduced through a single reference loudspeaker and compared with a stereo signal whose phantom center is aligned to the reference:

Among others, the following can be read in the conclusions:

The differences between monophonic and stereophonic hearing were probably quite clear in terms of timbre. This coincides with my own experiences.

The reason for the differences given in the paper is also interesting, because it is generally assumed that comb filter effects are the cause:

so the popular belief that comb filtering in stereo reproduction remains unnoticeable should be rethought.

This of course includes reflections from boundary surfaces, but also extends the view to interference in direct sound. This conclusion, however, does not presuppose lateral reflections as a possible cause for the differences in timbre heard in mono/stereo.

 

[ctrl]

Did some browsing in Floyd Toole's book and found this statement from him:
Source: Sound Reproduction (1st ed), Toole

And more thoughts to take away: A loudspeaker that sounds good in a monophonic comparison is likely to sound good in a stereo comparison, but the reverse is not necessarily true.

When comparing loudspeakers of very different sound quality, I would agree with this statement. Especially if the comparison is made as Toole did in his 1986 paper (the loudspeaker is not listened to freestanding and centered) which is comparable to a typical listening room - i.e. the influence of lateral reflections is given.

(My thesis)
However, if two loudspeakers are very close to each other in terms of sound and LS-A sounds perfect in mono and LS-B sounds somewhat "dark/unspectacular", listening to loudspeaker A in stereo will no longer sound optimal due to the change in timbre when listening in stereo and LS-B will might be preferred - provided that for the monophonic sound reproduction a single, centered speaker in mono is used, as it seems Harman does it in their listening tests.

Cause are comb filter effects due to the different travel times of direct sound to the ears (if, for example, the loudspeakers are listened to at an angle of 30 degrees) and direct sound interference from the two speakers.

Floyd Toole has quoted interesting studies on this in his book:
Source: Sound Reproduction (1st ed), Toole

I copied from the figure above the part that shows the measured deviation between listening to a single centered speaker compared to a pair of stereo speakers and entered the absolute deviations for a few prominent points.
The sound level differences are considerable and would explain the conclusions of the paper from 1996 cited in post#1 quite conclusively. In the range 2.3-9kHz the sound pressure level increases by 1-2dB (compared to the average sound pressure in the range 0.2-1kHz) and in the upper midrange there is a 5dB dip (even in a "normal" reflecting room the diffuse sound cannot completely fill the dip).

However, only the change in timbre from centered monophonic to stereophonic should be considered generally valid. According to other sources, the change in timbre does not always have to go from mono - dark sound, to stereo - brighter sounding (see first post). This certainly depends to a large extent on the loudspeaker and the music material used.
Source: Sound Reproduction (1st ed), Toole

Listeners in experiments by Choisel and Wickelmaier (2007) reported a reduction in brightness when a mono center loudspeaker was replaced by a stereo phantom image .

Stereo also suffered in terms of sound quality (brightness) of the phantom center image due to acoustical crosstalk (see Figure 9.7).

To emphasize again, this is not to say that Harman/Olive's research on the sound evaluation of loudspeakers and the score derived from it is wrong, but that changes may occur due to the change in timbre when listening to loudspeakers in stereo.

For a final sound assessment of a loudspeaker, however, freestanding, centered, monophonic listening of the loudspeaker is, due to the listed sources, rather not suitable - unless the loudspeaker is to be used as center speaker

 

[aarons915]

To emphasize again, this is not to say that Harman/Olive's research on the sound evaluation of loudspeakers and the score derived from it is wrong, but that changes may occur due to the change in timbre when listening to loudspeakers in stereo.

For a final sound assessment of a loudspeaker, however, freestanding, centered, monophonic listening of the loudspeaker is, due to the listed sources, rather not suitable - unless the loudspeaker is to be used as center speaker

I was interested in this same question regarding the MLL setup at Harman and asked Dr. Toole about whether a single speaker in the center of the room would emphasize the direct sound and minimize the effects of the lateral reflections, here is the response.

The shuffler allows us to compare speakers in center or side locations and both have been done. A mono speaker on the one side is a distraction, and no changes in ratings were noted, so the center location was selected. The center channel is the most important channel in a multichannel system which is a consideration. Stereo images/sounds are corrupted by acoustical crosstalk and timbral/spatial differences related to the side-wall reflections, so ratings in mono have reason to be higher than those in stereo,

As seen in Figure 7.1, in a classic setup the side wall reflected sounds radiate at nearly 90 deg off axis, so sound levels are much reduced at medium and high frequencies. Also seen in the figure is the huge difference between sounds contributing to a phantom stereo center and a real center.

Pity that stereo is the norm

Also some other random comments he made regarding the mono vs stereo debate in that thread:

In my early experiments (1982) I used several randomized room locations, Figure 3.12, and also a positional substitution scheme (1985), Figure 7.11, which was used in stereo and mono, as explained in great detail. Recordings were found to be the dominant factor in imaging and soundstage. In the tests the speakers and room symmetry were very carefully matched.

As you have noted, he has also made comments that speakers that win the double blind tests in mono are always preferred in stereo. Regarding your graph of stereo's crosstalk cancellation, if anything it is a good reason not to listen to music in stereo in the first place.

 

[ctrl]

I was interested in this same question regarding the MLL setup at Harman and asked Dr. Toole about whether a single speaker in the center of the room would emphasize the direct sound and minimize the effects of the lateral reflections, here is the response.

Yes, this is an exciting topic. On the one hand we have the statement:

The shuffler allows us to compare speakers in center or side locations and both have been done. A mono speaker on the one side is a distraction, and no changes in ratings were noted, so the center location was selected.

So the rating does not change at all if the position is changed from lateral alignment to center alignment.

On the other hand we have this statement from Sean Olive:

The experiment takes place in a custom built room with a hidden mechanical speaker shuffler. Recently renovated with a slightly tighter width, Sean says the smaller room size helps increase a greater sense of 1st reflection thus making the test more sensitive to the off-axis performance of the loudspeaker.

Source: https://audio-head.com/the-harman-tour-part-1-loudspeaker-audio/

If the lateral reflections do not change ratings, then why reduce the width of the room to obtain a higher proportion of lateral reflections for the evaluation?

As seen in Figure 7.1, in a classic setup the side wall reflected sounds radiate at nearly 90 deg off axis, so sound levels are much reduced at medium and high frequencies.

I drew the room from Figure 7.1 and tried to place the speaker in a typical way (the distance to the side wall should actually be >=1m, but very few people can do this, so 0.9m as a compromise). The sitting position is on one corner of an equilateral triangle.

Due to the longer path length, the signal of the first sidewall reflection weakens by 2.3dB.
If the tweeter of the speaker is aligned exactly to the "nose tip", the angle for the first side wall reflection is 75° - thus still far from 90°.
To determine the total sound pressure drop for the side reflection one has to calculate additionally the sound pressure drop between the 0° frequency response and the 75° frequency response.

Think, however, that speakers are not typically aligned in this way. In most cases the tweeter might be aligned in such a way that it points slightly past the shoulder. This means that direct sound corresponds approximately to the -15° frequency response (which in most cases produces a very flat frequency response) and the first sidewall reflection corresponds to the +60° frequency response. So we have to calculate the sound pressure drop between the horizontal -15° frequency response and the +60° frequency response.

For an reasonably evenly radiating (vertical centered alignment of the chassis) loudspeaker it looks like this (normalized on-axis, horizontal FR):

This results in a "typical" total attenuation for the first sidewall reflection, between 1-10kHz, from -4.3dB to -9.3dB. One could call this "much reduced at medium and high frequencies", but in terms of audibility or influence on sound attributes of the speaker, I would argue, it is not.

With poorly designed speakers it can happen that certain frequency ranges of the 60° frequency response even reach the level of the on-axis FR. In this case, the first sidewall reflection would probably have a direct influence on the timbre of the speaker. Here a few examples (normalized on-axis hor FR):


Whereby, as said in post#1, the timbre change between monophonic and stereophonic hearing is mainly assigned to comb filter effects due to the ear distances and direct sound interference from the two speakers.
The sidewall reflections are likely to influence attributes like "width, spaciousness and envelopment".

As you have noted, he has also made comments that speakers that win the double blind tests in mono are always preferred in stereo.

Sorry, I have not found such a statement. What I know is much more carefully formulated with much more room for interpretation.

As pointed out previously [1], highly rated loudspeakers receive closely similar ratings in both stereo and mono tests, but loudspeakers with lower ratings tend to receive elevated ratings in stereo assessments.

Source: AES - Loudspeaker Measurements and Their Relationship to Listener, 1986, Toole - (highlighting by me)

I could not find a paper for the alleged statement "speakers that win the double blind tests in mono are always preferred in stereo". Do you have a link where Toole investigated it - some AES paper?

Regarding your graph of stereo's crosstalk cancellation, if anything it is a good reason not to listen to music in stereo in the first place.

This certainly does not apply to modern recordings, since the sound engineer himself listened to the recordings in stereo when mixing and should therefore have automatically corrected the coarsest comb filter effects.
I would rather say that the other way round, listening to a modern stereo recording monophonic, is rather problematic. Because then exactly this effect occurs:

 

[pozz]

I could not find a paper for the alleged statement "speakers that win the double blind tests in mono are always preferred in stereo". Do you have a link where Toole investigated it - some AES paper?

In the third edition of his book Toole discusses the difference between stereo and mono listening directly.

Pg. 41

 

[pozz]

In the description "very close to zero hearing level" refers to 0 dB HL, which refers near-ideal normal hearing showing little loss of sensitivity.

In that section Toole writes and cites others who say that hearing loss correlates with increased inconsistency of preference (note: not difference of preference, but inconsistency of preference).

 

[aarons915]

If the lateral reflections do not change ratings, then why reduce the width of the room to obtain a higher proportion of lateral reflections for the evaluation?

That was actually my original question to Dr. Toole, apparently the room and setup have changed a few times over the years. My theory was that since they placed the speakers in the center of the room, the shorter sidewall distance in the smaller room created more typical lateral reflections. Intuitively it would seem like these weaker sidewall reflections would affect preference but I don't think it makes as big of a difference as it seems, I would think the added distance traveled in the center of the room to the sidewalls would be offset by the now reduced angle. You also have the front, rear, floor and ceiling reflections so when you average all of that together I don't think the center placement changes things that much.

This results in a "typical" total attenuation for the first sidewall reflection, between 1-10kHz, from -4.3dB to -9.3dB. One could call this "much reduced at medium and high frequencies", but in terms of audibility or influence on sound attributes of the speaker, I would argue, it is not.

Agreed and when you add in all of the early reflections together, I believe the total SPL becomes higher than the direct sound which to me is a good reason to prioritize the early reflections over the direct sound if necessary. I don't see how this would affect a stereo vs mono comparison though because each reflection is impacted the same, a speaker with stronger sidewall reflections will have that advantage regardless of where the speaker is placed.

Whereby, as said in post#1, the timbre change between monophonic and stereophonic hearing is mainly assigned to comb filter effects due to the ear distances and direct sound interference from the two speakers.
The sidewall reflections are likely to influence attributes like "width, spaciousness and envelopment".

Agreed but again a speaker with an advantage in sidewall reflections will still have that advantage in a mono setup or a stereo setup regardless of placement.

This certainly does not apply to modern recordings, since the sound engineer himself listened to the recordings in stereo when mixing and should therefore have automatically corrected the coarsest comb filter effects.
I would rather say that the other way round, listening to a modern stereo recording monophonic, is rather problematic. Because then exactly this effect occurs:

I don't think most recordings have a peak in the area of comb filtering and it wouldn't make sense if they did. The reasoning is that many people these days upmix to a multi channel setup and that would cause the recording to sound too bright for them. Even in a stereo setup, the comb filtering is only a serious problem for a central listener and even that effect is lessened by all of the early reflections and not a big deal in practice.

Just to be clear about the whole mono vs stereo debate, I don't think anyone is claiming we should be listening to music in mono, only that comparisons should be done in mono because differences are a lot easier to discern.

 

[Tim Link]

I really noticed the timbral change between center phantom image and hard pan left and right a couple days ago while listening to a stereo test track with a male voice. That's why I found this thread! I listened again today on my little desktop system at work and could hear it there also. Using an absorptive divider panel in front of my face to reduce crosstalk makes the center phantom image timbre a lot more like either side channel playing by itself. I've played with divider panels before to get better imaging, but never paid much heed to the timbral effects. Upmixing seems like a good solution but I don't know that the methods to do that from stereo recordings are all that good yet. I have some 3 channel SACDs that I used to listen to on my old 3 channel setup. At the time I really couldn't hear much of a difference between the 3 channel SACD and the standard 2 channel redbook CD version. This timbral effect may not be something likely noticed unless pointed out. I was only listening for a clear center image and was getting that with 2 channels.

 

[levimax]

Just to be clear about the whole mono vs stereo debate, I don't think anyone is claiming we should be listening to music in mono, only that comparisons should be done in mono because differences are a lot easier to discern.

I agree that the comparison of speakers is a different issue than timber change between stereo and mono. Recently I have listened to some of my older "real" mono recordings (LP's made before stereo even existed) in "real" mono (Channels summed, one speaker centered) and it is certainly a different experience than listening to the same LP in "two channel mono". For the old mono recordings mixed for one speaker I almost always prefer listening to it on one speaker (although I seldom do because I am too lazy) .... for recent mono recordings mixed for stereo I usually prefer two speakers. If you have never listened to a "real" mono recording (originally mixed and mastered for one speaker) on one speaker it is an interesting experience that should not be missed and gives some interesting perspectives on imaging and image stability and center phantom images and the like.

 

[JohnYang1997]

The point lies on these questions not whether it changes or not.
1, Does it make it easier to conduct listening test?
2, Does it make the result more consistent?
3, Does it change the outcome of preference?

 

[ctrl]

1, Does it make it easier to conduct listening test?

Economically speaking, in any case (significantly less effort). But also the listener should have it easier, because by monophonic listening certain effects, which play a role in stereophonic listening, are omitted.

2, Does it make the result more consistent?

The experiments of Toole and Olive clearly say yes.
Very few people are likely to have any real conscious experience of this.

3, Does it change the outcome of preference?

So, here it gets a bit more complicated.

If the listening tests only use music produced specifically for monophonic listening, it should not make any difference.

But if "normal" produced music (stereophonic mastering) is used and this music is converted to mono and then listened to monophonic, the case is different.
In this case, a timbre change occurs at the transition to stereophonic listening.
Will this completely change the order in which the speakers are rated? Certainly not.

But there are more and more loudspeakers that basically do everything right and then nuances in the crossover tuning are crucial.
This is where my criticism comes in, because if a loudspeaker sounds monophonic optimal under these conditions, it won't do so in stereophonic. A loudspeaker A may sound slightly worse monophonic than loudspeaker B, but better in stereophonic.

A second possible effect would be to hear a loudspeaker standing freely in the room (side walls are far away) monophonically, in contrast to the normal/typical stereophonic hearing of the loudspeaker in the listening room with the side walls at a relatively short distance.
This has already been discussed in the posts above and I have not yet come to a clear conclusion, as there are circumstances where the difference might not play too big a role - more details will be available from me in a separate post at some point ... in the future

 

[JohnYang1997]

Economically speaking, in any case (significantly less effort). But also the listener should have it easier, because by monophonic listening certain effects, which play a role in stereophonic listening, are omitted.


The experiments of Toole and Olive clearly say yes.
Very few people are likely to have any real conscious experience of this.


So, here it gets a bit more complicated.

If the listening tests only use music produced specifically for monophonic listening, it should not make any difference.

But if "normal" produced music (stereophonic mastering) is used and this music is converted to mono and then listened to monophonic, the case is different.
In this case, a timbre change occurs at the transition to stereophonic listening.
Will this completely change the order in which the speakers are rated? Certainly not.

But there are more and more loudspeakers that basically do everything right and then nuances in the crossover tuning are crucial.
This is where my criticism comes in, because if a loudspeaker sounds monophonic optimal under these conditions, it won't do so in stereophonic. A loudspeaker A may sound slightly worse monophonic than loudspeaker B, but better in stereophonic.

A second possible effect would be to hear a loudspeaker standing freely in the room (side walls are far away) monophonically, in contrast to the normal/typical stereophonic hearing of the loudspeaker in the listening room with the side walls at a relatively short distance.
This has already been discussed in the posts above and I have not yet come to a clear conclusion, as there are circumstances where the difference might not play too big a role - more details will be available from me in a separate post at some point ... in the future

I think for speakers that are not remotely close to the target/perfect speakers. It simply does not affect the outcome at all, in their original form. Meaning no EQ.
When there's eq and/or the performance of the speakers are already approaching sota then I think it makes sense to do in the correct positioning and doing it stereo. At the end of the day we listen stereo.

Perhaps, just perhaps. Once in a year, put some of the best performer (including those cheap but respond well to eq etc) together in stereo and shoot out that way. Even better, get some more people to do the evaluation. Then publish all the data of each individual for each speaker.

 

[ctrl]

Perhaps, just perhaps. Once in a year, put some of the best performer (including those cheap but respond well to eq etc) together in stereo and shoot out that way. Even better, get some more people to do the evaluation. Then publish all the data of each individual for each speaker.

This would be desirable, but would be very difficult to realize.
My hope is that one day there will be a binaural recording for each loudspeaker in a typical listening room.

A very good approach to this can be found at lowbeats.de who have developed a procedure that measures an impulse response of a pair of loudspeakers in a listening room and then can be convolved with any piece of music - Unfortunately, for binaural recording the loudspeakers are measured in a recording studio room, which of course is not comparable to a typical listening room and thus strongly emphasizes direct sound.
Theoretically, you could switch back and forth between each speaker and use your own music selection.
This does not yet exist, but the future belongs to such a virtual listening studios. Most loudspeaker manufacturers will certainly not support this

 

[JohnYang1997]

This would be desirable, but would be very difficult to realize.
My hope is that one day there will be a binaural recording for each loudspeaker in a typical listening room.

A very good approach to this can be found at lowbeats.de who have developed a procedure that measures an impulse response of a pair of loudspeakers in a listening room and then can be convolved with any piece of music - Unfortunately, for binaural recording the loudspeakers are measured in a recording studio room, which of course is not comparable to a typical listening room and thus strongly emphasizes direct sound.
Theoretically, you could switch back and forth between each speaker and use your own music selection.
This does not yet exist, but the future belongs to such a virtual listening studios. Most loudspeaker manufacturers will certainly not support this

I can try to realize that in the future.

 

[Tim Link]

If a voice is coming from the left speaker or the right it will have a different timbre than the voice coming from phantom center. This means to make a speaker sound good in stereo it needs to have proper timbre playing in mono, and the mastering person needs to do proper eq for the center phantom image sound. Otherwise if the speaker is corrected for center phantom timbre it will be wrong for voices to the side, unless the mastering person eqs the side voices to make up for the speaker's center phantom correction. Is this correct?

 

[pozz]

If a voice is coming from the left speaker or the right it will have a different timbre than the voice coming from phantom center. This means to make a speaker sound good in stereo it needs to have proper timbre playing in mono, and the mastering person needs to do proper eq for the center phantom image sound. Otherwise if the speaker is corrected for center phantom timbre it will be wrong for voices to the side, unless the mastering person eqs the side voices to make up for the speaker's center phantom correction. Is this correct?

Kind of. A single piece of music will typically contain multiple tracks, which are EQed separately, not globally. There are also mid-side tools for EQing the various parts of the full stereo image separately (you won't have total control, however). This is standard practice these days.

That said, phantom center timbres are subject to pretty severe comb filtering in a stereo setup (this is inherent to stereo) and will sound different from an actual center speaker.

Some resources and links in this post: https://www.audiosciencereview.com/...speaker-review-measurements.14310/post-443624

 

[andreasmaaan]

OTOH, when stereo music is mixed, it is done so on stereo speakers. So the phantom centre you are hearing when you play back the recording on stereo speakers is the phantom centre that the engineer intended you to hear when they mixed it on stereo speakers (just as a hard-panned voice/instrument is heard as the engineer heard it when they hard-panned it and then mixed it).

*Subject, ofc, to differences in distortions inherent to the speakers the engineers used and you are now using - but these do not tend to affect phantom centre images anymore than they do mono or hard-panned R/L images.

 

[Tim Link]

OTOH, when stereo music is mixed, it is done so on stereo speakers. So the phantom centre you are hearing when you play back the recording on stereo speakers is the phantom centre that the engineer intended you to hear when they mixed it on stereo speakers (just as a hard-panned voice/instrument is heard as the engineer heard it when they hard-panned it and then mixed it).

*Subject, ofc, to differences in distortions inherent to the speakers the engineers used and you are now using - but these do not tend to affect phantom centre images anymore than they do mono or hard-panned R/L images.

It may not be what the engineer ultimately would have liked for me to hear, but was as good as they could achieve considering the circumstance of the comb filtering. I don't think I've heard a center panned vocal that sounded worse on good headphones, or when listening to stereo speakers with a divider panel to eliminate crosstalk

 

[andreasmaaan]

It may not be what the engineer ultimately would have liked for me to hear, but was as good as they could achieve considering the circumstance of the comb filtering. I don't think I've heard a center panned vocal that sounded worse on good headphones, or when listening to stereo speakers with a divider panel to eliminate crosstalk

A valid point, although I can't say I've found anything to sound as good on headphones as on loudspeakers.

When I experimented with crosstalk cancellation with speakers, I was so enamoured of the spatial presentation that I forgot to pay much attention to questions of tonality. Will revisit it at some point

 

[Tim Link]

I also only noticed the spatial effects and not the tone so much when I first tried cross talk elimination. When I tried it last night listening to Norah Jones I preferred her vocal tone with the divider panel. The spatial information is not too interesting on that recording, pretty much instruments panned left and right and her in the middle - it sounded like to me. I generally like the sound of stereo speakers better than headphones too. When adjusting my home made speakers I use headphones for comparison and have noticed that vocal presence is better on the headphones no matter what I do. Seems this comb filtering is the problem. This might also explain why people complain that a center channel removes a sense of depth. Presence is the opposite of depth, I'd say. In real life as people move away from us there is more comb filtering to be heard. A vocal panned to center should still have depth if the microphone is moved away from the singer, even when fed to a center channel.