Why evaluating the sound of a single speaker is essential

[MAB]

I have a question.

If a speaker is designed to be listened to off-axis e.g. at approximately 30 degrees, then at what angle should it be evaluated in mono?

Such as previously mentioned, but if you want some examples, see below.

Dali speakers are designed to be listened to off axis.

Andrew Jones has mentioned he has designed with off axis in mind as the presumption is most customers will use this way.

Kef recommend starting with no toe in to the listener, so approximately 30 degrees to the main listening area best spot.

Maybe someone should do research on whether the manufacturer's stated off-axis recommendations are legit.

Lots of manufacturers have turned mismatched directivity into a game. Like B&W, do people really like the treble or have they just not found the perfect angle...

 

[Satanica]

Where are you going with this off axis stuff? Do you own one of these speakers that are recommended to be listened to off axis?

Usually those speakers that are recommended to be listened to off axis have an on axis problem, e.g. heightened on axis treble.

When a speaker is tested in mono, how is it typically tested, on/off axis?
I assume that mono speaker testing is done almost or entirely on-axis.
Is any consideration given that one could be testing a speaker outside of it's intended use case when on-axis?

But, regardless, when testing a speaker on-axis, it may or may not be the best representation of a speaker, example as linked to some speakers have fundamental issues on-axis that are at least (somewhat) remedied off-axis.

Have a guess as to whether my left and right speakers have the best anaechoic response on or off-axis and what the manufacturer recommends in regard to toe in? It might not be what you think most likely.

 

[Satanica]

Maybe someone should do research on weather the manufacturer's stated off-axis recommendations are legit.

Lots of manufacturers have turned mismatched directivity into a game. Like B&W, do people really like the treble or have they just not found the perfect angle...

Perhaps before any given speaker is tested in mono, it should be first determined what is the most preferred angle for listening to it (0, 10, 20, 30)? Pie in the sky stuff I know.

My quick AI lookup says that B&W are generally designed to be listened to on-axis.

 

[NTK]

The tests were sometimes conducted with more than 1 listeners, which means some were listening off-axis.

...
Only a single listener in the optimal seat is used in our product benchmarking tests. For group or tour demonstrations we will use multiple seats. We analyze the data separately for each seat and can see how the ratings vary across seat. The consistency in ratings across seats is quite remarkable except when a speaker has a significantly different frequency response on and off axis producing a much different direct sound for listeners in different seats

 

[Floyd Toole]

I assume that mono speaker testing is done almost or entirely on-axis

In my many years of evaluating loudspeakers I have always respected the manufacturer's "reference axis" for both measurements and listening - mono or stereo. To do otherwise would be irresponsible. Over the years the vast majority of loudspeakers performed best on-axis, so in the absence of a stated reference axis, that was used.

That said, most manufacturers don't specify a reference axis, or reveal any useful measurements, so either they don't know or don't care. With today's well-designed loudspeakers the on-axis and listening window (+/- 30 deg hor, +/- 10 deg vert) the curves are very close together which liberates the user to choose. As seen in several publicly available spinoramas there are problem loudspeakers on the market and the user is left to his/her own devices to decide what to do. There are also loudspeakers that are well behaved on and off axis, but with reduced HF output at increasing off-axis angles - this information allows the listener to choose intelligently how to optimize performance in any particular setup. Having trustworthy anechoic data is the entry point to understanding what loudspeakers to buy and how best to use them.

The tests were sometimes conducted with more than 1 listeners, which means some were listening off-axis.

This was done for demonstrations of the process. All critical listening evaluations were done solo - e.g. the ones used in product development and those in the subjective/objective correlations. Opinions collected from groups are often influenced by one or more "opinionated" listeners who verbally or otherwise communicate their feelings. Listening off axis is not necessarily the worst of the problems. Humans are easily biased.

 

[Duke]

If a speaker is designed to be listened to off-axis e.g. at approximately 30 degrees, then at what angle should it be evaluated in mono?

30 degrees

@Satanica, that was a concern I had as well, as I've been designing for off-axis listening for twenty years. My understanding is that Harman followed the manufacturer's recommendation regarding toe-in angle. My recollection is that @Floyd Toole said this in a post but now I can't find it.

Edit: Looks like Dr. Toole posted while I was searching for that post so I could quote it. Now I don't have to.

 

[Buckchester]

When a speaker is tested in mono, how is it typically tested, on/off axis?
I assume that mono speaker testing is done almost or entirely on-axis.
Is any consideration given that one could be testing a speaker outside of it's intended use case when on-axis?

But, regardless, when testing a speaker on-axis, it may or may not be the best representation of a speaker, example as linked to some speakers have fundamental issues on-axis that are at least (somewhat) remedied off-axis.

Have a guess as to whether my left and right speakers have the best anaechoic response on or off-axis and what the manufacturer recommends in regard to toe in? It might not be what you think most likely.

It doesn’t sound like you appreciate people making assumptions about your equipment, yet you are making assumptions about methods used in the “Harman” testing. How ironic.

 

[Duke]

This was done for demonstrations of the process. All critical listening evaluations were done solo - e.g. the ones used in product development and those in the subjective/objective correlations. Opinions collected from groups are often influenced by one or more "opinionated" listeners who verbally or otherwise communicate their feelings. Listening off axis is not necessarily the worst of the problems. Humans are easily biased.

I was under the mistaken impression that critical listening was done by three or four people at a time, probably because that's how many seats I saw in a photo somewhere. Thank you for clarifying this.

Usually those speakers that are recommended to be listened to off axis have an on axis problem, e.g. heightened on axis treble.

Maybe someone should do research on weather the manufacturer's stated off-axis recommendations are legit.

I use round waveguide-style horn horns, and such devices tend to have an on-axis anomaly: There will be a dip in the frequency region where the mouth reflection arrives 1/2 wavelength behind the direct sound. This dip goes away at about 10 or 15 degrees off-axis.

I also design with time-intensity trading in mind, which calls for a fairly narrow and well-behaved radiation pattern with the axes criss-crossing in front of the main listening position, this again putting the designed-for listening axis somewhat off-axis.

 

[MAB]

I use round waveguide-style horn horns, and such devices tend to have an on-axis anomaly: There will be a dip in the frequency region where the mouth reflection arrives 1/2 wavelength behind the direct sound. This dip goes away at about 10 or 15 degrees off-axis.

I have a few speakers with concentric drivers. All of them are terrific sounding speakers at all angles, the on-axis anomaly is barely noticeable and (I think) only stands out because the speakers are so well-behaved on- and off-axis. The on-axis behavior is barely noticeable in the measurements and unless I am listening critically I don't notice. I wonder how much of this I actually hear and how much I imagine given how many people say you need to listen to coaxial speakers 10-15° off-axis.

The thing I do notice is that across listening angles they have a predictable change in tone due to the smooth off-axis behavior, which is way more noticeable than the on-axis anomalies to me.

I would like to see a comprehensive test of the audibility of the on-axis resonances that show up in measurements, and the alleged audible effects. I had a pair of Seas Prestige Coaxial speakers I built for my dad that weren't good sounding, including poor on-axis behavior. I am pretty sure they would have audible issues, but honestly the on-axis behavior was just one of the issues with them. On the other hand so many good coaxial speakers sound great on-axis and off-axis too. I really wonder what people can hear, and if the Seas Prestige coax just sounds like a crummy speaker, and a Genelec One just sounds like a good one, irrespective of on- or off-axis.

 

[MAB]

Perhaps before any given speaker is tested in mono, it should be first determined what is the most preferred angle for listening to it (0, 10, 20, 30)? Pie in the sky stuff I know.

Or measure it on a near-field-scanner to determine what is actually happening on- and off-axis. In mono.

My quick AI lookup says that B&W are generally designed to be listened to on-axis.

Measurements exist of B&W on- and off-axis behavior. They tell a different story than AI scraping the internet for anecdotes.

 

[Duke]

I have a few speakers with concentric drivers. All of them are terrific sounding speakers at all angles, the on-axis anomaly is barely noticeable and (I think) only stands out because the speakers are so well-behaved on- and off-axis. The on-axis behavior is barely noticeable in the measurements and unless I am listening critically I don't notice. I wonder how much of this I actually hear and how much I imagine given how many people say you need to listen to coaxial speakers 10-15° off-axis.

Yes, "measurable" doesn't necessarily mean "audible"; and "audible" doesn't necessarily mean "objectionable".

In the case of the on-axis dip, it's an interference effect (i.e. NOT a resonance) and therefore ONLY present in the on-axis direct sound, and could easily be inaudible if the direct-to-reverberant ratio is sufficiently low at the listening position.

Imo it would be a mistake to correct for this on-axis cancellation dip via equalization, especially since the center frequency of the dip changes with listener (or microphone) distance.

I would like to see a comprehensive test of the audibility of the on-axis resonances that show up in measurements...

A resonance will show up in the on-axis as well as off-axis curves, and that's a reliable way of determining whether a particular response anomaly is a resonance and not something else.

 

[MAB]

Yes, "measurable" doesn't necessarily mean "audible"; and "audible" doesn't necessarily mean "objectionable".

In the case of the on-axis dip, it's an interference effect (i.e. NOT a resonance) and therefore ONLY present in the on-axis direct sound, and could easily be inaudible if the direct-to-reverberant ratio is sufficiently low at the listening position.

Thanks. I am throwing around the term 'resonance'.

Imo it would be a mistake to correct for this on-axis cancellation dip via equalization, especially since the center frequency of the dip changes with listener (or microphone) distance.

Yes, I wasn't suggesting correction. I was just saying I seem to be uncritical to the slight on-axis dip.

 

[Satanica]

In my many years of evaluating loudspeakers I have always respected the manufacturer's "reference axis" for both measurements and listening - mono or stereo. To do otherwise would be irresponsible. Over the years the vast majority of loudspeakers performed best on-axis, so in the absence of a stated reference axis, that was used.

That said, most manufacturers don't specify a reference axis, or reveal any useful measurements, so either they don't know or don't care. With today's well-designed loudspeakers the on-axis and listening window (+/- 30 deg hor, +/- 10 deg vert) the curves are very close together which liberates the user to choose. As seen in several publicly available spinoramas there are problem loudspeakers on the market and the user is left to his/her own devices to decide what to do. There are also loudspeakers that are well behaved on and off axis, but with reduced HF output at increasing off-axis angles - this information allows the listener to choose intelligently how to optimize performance in any particular setup. Having trustworthy anechoic data is the entry point to understanding what loudspeakers to buy and how best to use them.


This was done for demonstrations of the process. All critical listening evaluations were done solo - e.g. the ones used in product development and those in the subjective/objective correlations. Opinions collected from groups are often influenced by one or more "opinionated" listeners who verbally or otherwise communicate their feelings. Listening off axis is not necessarily the worst of the problems. Humans are easily biased.

Thanks. These days, I think there is evidence that a significant number of speakers are superior/most preferred off axis (X degrees) rather than on-axis. Such speakers are either designed specifically with off-axis listening in mind and/or just happen to be a product of somewhat haphazard design and implementation. Therefore, it actually seems irrelevant the intention of the designer/producer.

What is relevant is what is the most preferred/best axis for any given speaker. This could be determined with a pre-testing regime that discovers it and for the sake of simplicity, lets just think about 0 and 30 degrees although other angles could be relevant as well. In a truly anechoic environment, this could be determined by listening to a single mono speaker equal distance to the listeners ears at multiple angles. I a non-anechoic environment, evaluating a single mono speaker at lets say 30 dregrees would mean lopsided reflections in a symmetrical room and that does not seem close to ideal. So, perhaps in a non-anechoic environment this would be better discovered if the speaker was listened to in pairs in stereo in a symmetrical room. I foresee that some are going to have issues with this, although at this stage I would not understand why.

I think if a speaker is superior/most preferred off-axis, then it can only be considered a less than ideal version of itself when used as a centre channel for example (on-axis at the main listening position). I don't think it is outrageous to think that many speakers, perhaps even the majority, are not given any consideration by the manufacturer that they will be used a centre speaker, none. They are designed to be used in stereo pairs as left and right channels, but the apparent gold standard of testing them is if they are or could be used as a centre speaker on axis to the main listening position.

I'm not against single mono speaker evaluation in a non-anechoic environment, but it does not seem to be the perfect ideal standard that many seem to believe.

 

[Buckchester]

Thanks. These days, I think there is evidence that a significant number of speakers are superior/most preferred off axis (X degrees) rather than on-axis. Such speakers are either designed specifically with off-axis listening in mind and/or just happen to be a product of somewhat haphazard design and implementation. Therefore, it actually seems irrelevant the intention of the designer/producer.

What is relevant is what is the most preferred/best axis for any given speaker. This could be determined with a pre-testing regime that discovers it and for the sake of simplicity, lets just think about 0 and 30 degrees although other angles could be relevant as well. In a truly anechoic environment, this could be determined by listening to a single mono speaker equal distance to the listeners ears at multiple angles. I a non-anechoic environment, evaluating a single mono speaker at lets say 30 dregrees would mean lopsided reflections in a symmetrical room and that does not seem close to ideal. So, perhaps in a non-anechoic environment this would be better discovered if the speaker was listened to in pairs in stereo in a symmetrical room. I foresee that some are going to have issues with this, although at this stage I would not understand why.

I think if a speaker is superior/most preferred off-axis, then it can only be considered a less than ideal version of itself when used as a centre channel for example (on-axis at the main listening position). I don't think it is outrageous to think that many speakers, perhaps even the majority, are not given any consideration by the manufacturer that they will be used a centre speaker, none. They are designed to be used in stereo pairs as left and right channels, but the apparent gold standard of testing them is if they are or could be used as a centre speaker on axis to the main listening position.

I'm not against single mono speaker evaluation in a non-anechoic environment, but it does not seem to be the perfect ideal standard that many seem to believe.

Much testing was done in stereo and mono. The results were the same. The only difference was that people arrived at their conclusions more quickly in mono. There are reasons for this should you care to read about them.

 

[Satanica]

It doesn’t sound like you appreciate people making assumptions about your equipment, yet you are making assumptions about methods used in the “Harman” testing. How ironic.

It seems you like to play the man a bit more than the ball. But whatever, but please correct me if I'm wrong that Harman testing of a single speakers is/was not done on-axis directly in front of the listener?

 

[Newman]

What is relevant is what is the most preferred/best axis for any given speaker. This could be determined with a pre-testing regime...

I expect it could be determined from the anechoic spinorama.

I think if a speaker is superior/most preferred off-axis, then it can only be considered a less than ideal version of itself when used as a centre channel for example (on-axis at the main listening position).

EQ it.

I don't think it is outrageous to think that many speakers, perhaps even the majority, are not given any consideration by the manufacturer that they will be used a centre speaker, none. They are designed to be used in stereo pairs as left and right channels...

What about pro audio speakers? Desktop speakers? Multichannel suites of speakers?

Let's not constantly be dragging everything back to Bad Old 2.0 with no EQ. It doesn't have to rule the world, not any more.

Much of the available audio science points us away from sticking with two full range speakers used 'as is' in the listening room, à la 1960, and more towards more than two main channels, separate subwoofers placed in different locations than the mains, more than one subwoofer, and equalisation of the bass frequencies.

I'm not against single mono speaker evaluation in a non-anechoic environment, but it does not seem to be the perfect ideal standard that many seem to believe.

At last we see your central hypothesis. It would have been easier to be a bit more upfront about it, instead of engaging us in a game of Twenty Questions.

please correct me if I'm wrong that Harman testing of a single speakers is/was not done on-axis directly in front of the listener?

Didn't Dr Toole answer that in reply to you, a few posts up? And in the same post, his reference to the vast majority of speakers being at their best on-axis, seems to deal with your central hypothesis, IMHO.

cheers

 

[Satanica]

I expect it could be determined from the anechoic spinorama.

I agree, but it would still be untested.

What about pro audio speakers? Desktop speakers? Multichannel suites of speakers?

Let's not constantly be dragging everything back to Bad Old 2.0 with no EQ. It doesn't have to rule the world, not any more.

Much of the available audio science points us away from sticking with two full range speakers used 'as is' in the listening room, à la 1960, and more towards more than two main channels, separate subwoofers placed in different locations than the mains, more than one subwoofer, and equalisation of the bass frequencies.

I agree, but stereo is "the hobby". I think Dr Toole has mentioned it is the non ideal (stereo) situation we have to live with and I agree. But, and this is a very big but, stereo translates perfectly to headphone listening as we only have two ears. So there we are, stereo mixes are going nowhere as the primary mix, unless humans evolve more ears. I predict that many "normal" people wouldn't even know that stereo even exists and assume the same thing always comes out of both headphones or speakers (dual mono). But anyhow.

At last we see your central hypothesis. It would have been easier to be a bit more upfront about it, instead of engaging us in a game of Twenty Questions.

Oh I thought that would have been very obvious from the start, but perhaps I was wrong about assuming that.
But really I had and have no hypothesis, but rather I was/am questioning testing procedures/methodology.
"Twenty Questions" is quite an exaggeration.

Didn't Dr Toole answer that in reply to you, a few posts up? And in the same post, his reference to the vast majority of speakers being at their best on-axis, seems to deal with your central hypothesis, IMHO.

Yes, but I was replying to someone else. I maintain the reference axis of a designer (if they actually have one) might not necessarily be the actual preferred axis, at least to some listeners in some rooms.

 

[Buckchester]

It seems you like to play the man a bit more than the ball. But whatever, but please correct me if I'm wrong that Harman testing of a single speakers is/was not done on-axis directly in front of the listener?

If a speaker sounds better off-axis than on, then I would think there is a flaw in that speaker.

Really good speakers hardly sound different on vs slightly off axis anyways.

 

[Satanica]

If a speaker sounds better off-axis than on, then I would think there is a flaw in that speaker.

Perhaps in most examples, then I agree. I don't object to this statement, however I do object to testing a speaker in what can be in some instances sub optimal for a particular speaker i.e. on-axis is less preferred than off-axis and then calling the results of this test, absolute/perfect.

Really good speakers hardly sound different on vs slightly off axis anyways.

But they do, and the practice of off-axis listening being possibly preferred to on-axis in room, is endorsed by Dr Toole, see below.

There are also loudspeakers that are well behaved on and off axis, but with reduced HF output at increasing off-axis angles - this information allows the listener to choose intelligently how to optimize performance in any particular setup. Having trustworthy anechoic data is the entry point to understanding what loudspeakers to buy and how best to use them.

 

[Floyd Toole]

Perhaps in most examples, then I agree. I don't object to this statement, however I do object to testing a speaker in what can be in some instances sub optimal for a particular speaker i.e. on-axis is less preferred than off-axis and then calling the results of this test, absolute/perfect.



But they do, and the practice of off-axis listening being possibly preferred to on-axis in room, is endorsed by Dr Toole, see below.

Also, one must not ignore the fact that if a tweeter is slightly bright or dull on or off the listening axis there is a treble control that in many cases will be able to deliver the desired response. Loudspeakers must first not have resonances - that is the prime audible problem. The rest is spectral balance, which can be easily altered with tone controls or EQ (assuming that DI is not grossly flawed) and which is a requirement in some recordings because they are not consistently "neutral".