Why evaluating the sound of a single speaker is essential

[Thomas_A]

As long as there is a reference to compare to, single speaker evaluations have the highest resolution. Evaluating without a reference means that you are in the "circle of confusion" dance - you are relying on quite large errors and/or your "memory" of good speakers. When errors get small, it is difficult without comparing A-B to the reference.

That said, it should be a goal to start with a linear on-axis response with an even dispersion and width according to personal preference. Finally, mixing and mastering does not follow any strict rules to adjust for the centre phantom image timbre. Some here say they do (perhaps some do), but others I've spoken to, does not follow any "anti-Shirley" EQ curve at all. They use EQ to taste. IMO, the "linear response" only refers to what's in the final audio file. Some say the ideal is have the same signal at the ear drum as what the mixing/mastering engineers had. I don't know how to control for that, so I rather have a sound that please me (which is very close to a linear response).

 

[RickS]

@Duke really appreciate your craft but seems you might be talking the difference between evaluating the speaker independent of the room vs as part of the room. Unless shown, it is difficult for many to appreciate the difference between system engineering and product engineering too.

Also seems a common issue for this thread overall. Maybe even more so outside of ASR...

Am pretty sure that nobody is claiming that evaluating a single speaker is the endgame. Do you never build a single speaker first and evaluate it (to some degree) before building another?

 

[Duke]

@Duke really appreciate your craft but seems you might be talking the difference between evaluating the speaker independent of the room vs as part of the room.

I consider speaker/room interaction to be an inevitability, therefore I design with speaker/room interaction in mind. I never design with "speaker independent of the room" in mind. Maybe I'm doing it all wrong??

Do you never build a single speaker first and evaluate it (to some degree) before building another?

Single speaker for measurements; single speaker vs single speaker for sound quality (crossover) evaluation; both speakers in stereo for spatial quality evaluation.

 

[MattHooper]

That's a description straight out of a very subjective audio magazine.

Yes. Surprisingly language is a form of communication. The experience of hearing sound is subjective - it creates subjective impressions which we can put into words. You can look up what every word there means in the dictionary if helpful.

Frequency response

Total Harmonic Distortion (THD)

Sensitivity (dB/W/m)

Impedance curve

Phase response / Group delay

Directivity / Off-axis response

Resonances / Cumulative Spectral Decay (CSD)

Intermodulation Distortion (IMD)

Maximum SPL

Step response / Impulse response

Thanks, but how exactly will any of those predict the characteristics I’ve been describing? Not that they won’t… but just saying “ look at this stuff” doesn’t really clarify much.

Also, does this mean that the qualities I described can be evaluated by listening in mono or not?

If you still want to hear it in order to judge it, pink noise is a good option.

If you train your hearing for pink noise or how it should sound when all frequencies are reproduced correctly, you are 90% on the safe side when assessing loudspeakers by ear.

Again, it’s unclear how that is addressing my question. Will I be able to predict the type of characteristics I described? If so, what aspects of the sound of pink noise from one channel is going to tell me that?
Again, I’m not saying that it couldn’t: I’m just looking for a clearer connecting of dots as to how it would.

Completely without flowery descriptions and audio poetry.

This ongoing allergy to descriptive language…


The language is there because I’m trying to describe the subjective experience that would lead to asking this question in the first place. And if you don’t recognize anything that I’m trying to describe there, then I would just have to move onto somebody else who can.

Such subjective descriptions are used all the time to get real work done in the real world of sound production, sound mixing, etc. (not to mention everywhere else in life.)

One can talk about a resonance or frequency, emphasis between 80 to 200 Hz, but the question remains “ What does that sound like? What are the Sonic consequences, for instance how might this make male vocals sound versus a dip in that same area?”

If you can’t communicate about such experience then you’ve effectively stopped yourself just at the point the measurements become relevant.

Subjective language really can communicate about these things - so long as the recipient doesn’t reject the attempt outright, since communication is a two-way street.

If, through rejecting the very relevance of such descriptions, the point of the question cannot get through, then the replies aren’t going to be very helpful.

I certainly think measurements are going to be relevant, and perhaps all the dots can be connected to the character characteristics I’m talking about evaluating. Though perhaps that still leaves the question open whether one can subjectively protect from listening to one speaker the characteristics I’m talking about in stereo.

Cheers

 

[Arindal]

it looks as though you believe that these two things are the same.

No, this is seemingly a misunderstanding from your side.

The stereo effects are not the loudspeaker, they are the recording.

The underlying patterns like reverb and interaural similarities defining phantom localization are of course part of the recording. But how they translate to what we perceive, is to a great amount influenced by the loudspeakers, the room and how they interact with each other. One and the same recording can sound fundamentally different in terms of localization stability, proximity, depth-of-field, reverb envelopment of the voices/instruments, tonality of the reverb and angles from which the listener is enveloped, in different monitoring facilities.

The evaluation of those effects are not an evaluation of the loudspeaker but of the stereo effects.

I dispute that. How many stereo recordings of a complex work using a main mic arrangement in a concert hall or church (preferably non-intensity stereophony) plus spot mics have you created and listened to on let us say 5 different monitors in 5 different studios? If you did, did they all give the exact identical perception in terms of imaging and ambience? I doubt that.

Here is what AI says;

I was interested in your experience with different monitoring facilities and subjective evaluation of loudspeakers, not what AI says.

 

[SIY]

@Duke really appreciate your craft but seems you might be talking the difference between evaluating the speaker independent of the room vs as part of the room.

That's part of it. But particularly, if L-R placement is baked in to the design, mono evaluation is problematic. As I asked in my previous post, how would you set up a Ken Kantor speaker like the 3.3? It has to go up against the wall to perform properly, and once you do that, what's the axis for listening? "Straight ahead" has no meaning here- if the listener axis is normal to the baffle, the speaker will be either on the left or on the right. If the speaker is placed directly in front of the listener, the first arrival is off axis AND partially baffled by acoustic foam.

Some of us think that designing a speaker to work with the room using designed placement (Kantor, Allison...) is a superior approach to home audio. But such speakers will be crippled using the methods discussed here.

 

[RickS]

I consider speaker/room interaction to be an inevitability, therefore I design with speaker/room interaction in mind. I never design with "speaker independent of the room" in mind. Maybe I'm doing it all wrong??

Depends on the goal ofc. Might approach differently if I have a known fixed room for the speaker(s) than otherwise. Most mass market speaker designers have to design “independent” of a specific room. Although I do know some that claim they design without measuring below 200 Hz. I would hope all listen in mono and stereo.

As I mentioned, do not think anyone is endorsing solely evaluating a speaker in mono. Even if one does not consider stereo imaging, the interaction of 2 speakers in a room is much different than a single one. Some would seem to demean the value of evaluating a single speaker, but given what I have learned, seems to isolate other issues and allow the speaker to be heard in a more discriminating conditions.

Shy of having access to fancy measurement equipment and/or anechoic chambers, a nearfield listen of a single speaker seems a key evaluating tool for many audiophiles. It has not been accepted practice but seems more likely a gap in understanding the benefit.

Single speaker for measurements; single speaker vs single speaker for sound quality (crossover) evaluation; both speakers in stereo for spatial quality evaluation.

Thanks for sharing. Gather single speaker sound quality involves some amount of listening to music?

 

[Arindal]

No, that was just a few little hints on how you can get a feel for the changes.

´feel for the changes´ defines a discrimination test in my understanding. And I don't think I need a feel for how pink noise sounds, have heard it numerous times under anechoic conditions as well as in concert venues and listening rooms. But I have never met anyone trying to evaluate imaging quality with pink noise. That is counterintuitive in my understanding, as pink noise is a frequency mixture which is continuous and decorrelated over time, containing no meaningful transient events for our ears.

But within that, they’re also the only speakers I’ve ever owned where I don’t perceive a fall-off in their sound quality from center-panned to hard-panned sounds.

I would take that as an indication that something is far from ideal with those speakers when the task is to reproduce naturally recorded phantom sources and their enveloping reverb. Mono material intensity-panned to the very flanks or center-panned as a 0deg phantom localization with little or just decorrelated reverb must sound different because it is very different for our ears.

It is pure speculation, and I do not know if this is applicable to yours, but I have heard a lot of speakers which make monaural material (both in mono or as a phantom source in stereo) sound less artificial, reduce annoying proximity of the ´birds on a wire´ localization and give a slight feeling of depth and natural localization. These are typically speakers which I would expect to sound superior in a monaural testing. Without any exception, these were the ones that gave a kinked, unnatural and artificial imaging in stereo, distorting proximity and depth-of-field alike, separating localizable sources from their reverb pattern on the recording.

he plays pink noise, and then plays pink noise run through EQ that simulates the speaker' anechoic Klippel response. It comes via compressed YouTube audio of course, but the difference is always quite obvious.

That is indicative of a massive difference between direct sound and indirect sound in the room where it is recorded, as the EQ simulation does solely include the alteration of the direct sound. If such difference occurs, it is pretty likely that the directivity is very uneven hence the diffuse soundfield severely colored.

I wonder how big the difference would be with a true constant directivity speaker. If the room´s RT60 is more or less balanced, the difference should be much smaller, if not negligible, in this case.

I haven't found, for example, SACD multichannel remastering/remixes of classic mid century jazz to be to my liking vs the originals.

I am not aware of the production details of the recordings in question, but ´mid-century jazz´ might be hinting to albums for which no usable multitrack master exists that allows a remix based on discretely separated direct sound and ambience tracks.

If I am not mistaken, for the overwhelming majority of recordings made before 1965, at best there is a couple of downmixed 3 or 4-tracks without separated ambience existing. Any surround remix made from these would be sounding fundamentally different from the original stereo mix, and I have heard a few pretty artificial ones.

 

[MarkS]

As I mentioned, do not think anyone is endorsing solely evaluating a speaker in mono.

Floyd Toole, Sean Olive, @amirm absolutely do endorse evaluating a speaker solely in mono. That's how @amirm does all his reviews here.

 

[watchnerd]

I am not aware of the production details of the recordings in question, but ´mid-century jazz´ might be hinting to albums for which no usable multitrack master exists that allows a remix based on discretely separated direct sound and ambience tracks.

If I am not mistaken, for the overwhelming majority of recordings made before 1965, at best there is a couple of downmixed 3 or 4-tracks without separated ambience existing. Any surround remix made from these would be sounding fundamentally different from the original stereo mix, and I have heard a few pretty artificial ones.

It's a whole mishmash of techniques that changed quite a bit from the mid 50s (mostly mono), early hamfisted stereo, and better stereo.

In addition, due to tape aging, in many cases we don't have the mix tapes anymore, anyway. We're just left with post mix masters.

 

[JPA]

OK for dispersion pattern, but - "no particular tones sticking out"?
How do you now it's not the room acoustics that sticks some tones up or down?
And in relation to what? Again - what is the sound of ideal pink noise?

First, listen to pink noise over a good set of headphones. Use that sound as your reference when playing pink noise over speakers.

 

[Duke]

Might approach differently if I have a known fixed room for the speaker(s) than otherwise.

Painting with broad strokes here, imo there is usually enough acoustic similarity between small-to-medium, not-overdamped rooms with non-nearfield speaker setups that the same general principles apply. So imo having a "known fixed room" isn't a requirement for taking speaker/room interaction into account at the speaker design stage.

Gather single speaker sound quality involves some amount of listening to music?

Very much so. Sometimes pink noise too, but usually not.

 

[aarons915]

Not that I have read every statement he has made on the subject, but my understanding is that Dr. Toole has said the speaker that "wins" sound quality in mono always wins sound quality in stereo, but I do not recall him saying that the speaker that "wins" spatial quality in mono always wins spatial quality in stereo. And the way he uses the two terms, spatial quality is not a subset of sound quality.

Somebody please correct me if I am mistaken.

(I first recognized and began using single-speaker evaluations as more revealing of sound quality differences back in the mid-80's, as an enthusiastic amateur.)

He said the speaker that wins in mono always wins in stereo, spatial qualities are part of that of course but blind listeners are simply choosing which speaker they prefer with spatial qualities, tonality, lack of resonances and extended bass being the main factors in choosing a winner. I think one thing that is interesting is how the quad mostly catches up in spatial qualities when used as a stereo pair in one of the earlier studies on mono listening, this tells me that ultra wide dispersion speakers aren't necessary when you have at least 2 in stereo.

 

[Arindal]

First, listen to pink noise over a good set of headphones. Use that sound as your reference when playing pink noise over speakers.

Sounds like a very misleading idea to me. Headphones partly circumvent your own HRTF, so pink noise must sound completely different on headphones compared to speakers, which will give you the typical tonal balance having undergone the HRTF. Note that the latter will be very different whether you listen to a stereo setup, a mono speaker at 30deg horizontally or one in front of you.

There are headphones which are labelled as ´diffuse field corrected´, which in theory should be some kind of averaged speaker-emulating HRTF correction, but in reality noise tonality still sounds pretty different on them.

 

[tmtomh]

I would take that as an indication that something is far from ideal with those speakers when the task is to reproduce naturally recorded phantom sources and their enveloping reverb. Mono material intensity-panned to the very flanks or center-panned as a 0deg phantom localization with little or just decorrelated reverb must sound different because it is very different for our ears.



That is indicative of a massive difference between direct sound and indirect sound in the room where it is recorded, as the EQ simulation does solely include the alteration of the direct sound. If such difference occurs, it is pretty likely that the directivity is very uneven hence the diffuse soundfield severely colored.

RE the first statement of yours I've quoted here, Incorrect. I never said the hard-panned sounds are identical to the center-panned ones. I said they sound just as good. Not the same thing. You're bringing a whole raft of unwarranted assumptions to your incorrect interpretation of what I wrote.

RE the second statement, you seem to have misunderstood or else are again bringing in irrelevant considerations. Erin's pink noise comparisons are not to my knowledge in-room microphone recordings. They are simulations: he plays pink noise direct, and then plays another passage of pink noise, also direct, but this time with EQ to simulate the speaker in question based on the Klippel response. Of course this is not a complete simulation because as you note we either are not getting the speaker's reflected sound characteristics, or if Erin is using the estimated in-room response rather than the anechoic response, then we're getting an approximation in a "typical" room and not in our specific room. But that doesn't meant the difference between the pure pink noise and the pink noise simulated via EQ from the reviewed speaker is about direct-vs-reflected sound. It's about how the speaker differs from flat. If you want to go chasing the small number of speakers that are intentionally - and successfully - designed so that their off-axis sound "fills in" holes in their direct sound, go ahead. But that's got nothing to do with the value of pink noise as a comparative evaluation tool for speakers.

 

[Duke]

Not that I have read every statement he has made on the subject, but my understanding is that Dr. Toole has said the speaker that "wins" sound quality in mono always wins sound quality in stereo, but I do not recall him saying that the speaker that "wins" spatial quality in mono always wins spatial quality in stereo. And the way he uses the two terms, spatial quality is not a subset of sound quality.

Somebody please correct me if I am mistaken.

He said the speaker that wins in mono always wins in stereo, spatial qualities are part of that of course

@Floyd Toole, could you clarify whether your observation that (paraphrasing) "the speaker which wins in mono always wins in stereo" includes always winning in spatial quality as well as always winning in sound quality?

Thanks!

 

[Salt]

@Floyd Toole, could you clarify whether your observation that (paraphrasing) "the speaker which wins in mono always wins in stereo" includes always winning in spatial quality as well as always winning in sound quality?

Thanks!

That should be clarified , indeed.

 

[Vuki]

@Floyd Toole, could you clarify whether your observation that (paraphrasing) "the speaker which wins in mono always wins in stereo" includes always winning in spatial quality as well as always winning in sound quality?

Thanks!

He said that stereo only diminish differences heard in mono sound quality test. Regarding spatial quality in relation to program and sound quality - research papers were lost...

 

[Duke]

I think one thing that is interesting is how the quad mostly catches up in spatial qualities when used as a stereo pair in one of the earlier studies on mono listening, this tells me that ultra wide dispersion speakers aren't necessary when you have at least 2 in stereo.

Agreed, this is interesting!

If sufficient test data were available, might the preferred radiation pattern characteristics for spatial quality be different for mono than for stereo? Might the preferred radiation pattern characteristics even be somewhat room-and-setup dependent? (I'm talking about two-channel stereo playback, not upmixed stereo or multi-channel playback.)

 

[RickS]

Floyd Toole, Sean Olive, @amirm absolutely do endorse evaluating a speaker solely in mono. That's how @amirm does all his reviews here.

The research shows more consistent evaluation results when you listen to a single speaker. The research also shows that stereo listening introduces more variables and makes it more difficult to discern speaker flaws. What I have not seen stated is not to evaluate in stereo ever.

What if 2 single speakers evaluated the same and listening in stereo yielded a preference? I think there is a tendency in many cases to try to turn these situations into absolutes. While I should have been more careful about who endorsed what, my context was about what happens when a second speaker is introduced. At the very least, you will get more output and more reflections. Toole’s book goes on to discuss this and more as well.