Why evaluating the sound of a single speaker is essential

[Jaxx1138]

While such criteria are judgeable in mono, this requires a designated mono mix standardized for the intended angle at which the speaker is placed, to come to a valid conclusion. I would not trust any verdict on tonal balance derived from a downmixed or channel-mapped stereo recording as this introduces a potential root for serious misjudgment

This is quite correct as a stereo signal mixed to mono will exaggerate correlated signals and reduce or cancel out of phase information.

Indirectly they are, as qualities of a speaker despite from just reproducing a tonally balanced and undistorted signal, greatly influence the ambience, localization and imaging aspect. Think of its following properties:

- pair tolerances (influence localization width and stability)
- dimensions and geometry of localization-relevant sound sources (localization stability and how a phantom source would be perceived within the ambience). This is particularly an issue under nearfield conditions or in a setup with pronounced early reflections
- evenness of frequency response over the potential listening window with the listener slightly moving or turning his/her head (again a typical nearfield issue)
- edge diffraction effects causing additional mirrored phantom localization or blurring the existing ones
- directivity index over frequency (greatly affects depth-of-field, perception of reverb tonality and direction, proximity and phantom source integration)

All of these items are certainly factors in how well the loudspeaker reproduces the signal. But all psychoacoustic effects are from the original mix. Any spatial effects from room reflections are technically a form of distortion. Even though some people may enjoy the effect, it was not likely intended in the original mix. The only part of the room that is consistent (even though it is a variable) is room gain at lower frequencies. Without it the recording will sound lean or lacking body or warmth. But I agree that all of these attributes are important and finding effective ways to subjectively evaluate those characteristics would be beneficial to the Hi-Fidelity journey.

Under anechoic conditions, certainly not. Nevertheless this test would be useful to separately judge localization stability, particularly when executed with a complex signal such as speech.

I agree that the better matched the two loudspeakers are the better all of the stereo effects will be. You often see issues with center channels that are considerably different from the Left or Right units. The loudspeakers would ideally be perfectly matched in all aspects in order to recreate the signal in stereo with an acoustically symmetric room.

 

[Dumdum]

Yesterday I listened to music for a good hour with just one speaker turned on.
I only noticed when I got closer to the system.

By the rational of this post and Floyd’s research that must be a nice sounding mono setup

 

[Matt_Holland]

I too would like to know what source material @amirm uses for his subjective mono speaker evaluations and also if anyone is aware of any dedicated mono test material available for evaluating speakers. I imagine that finding recordings of voice that would down-mix from stereo correctly from mono would be quite easy, but multitrack musical content, not so easy.

 

[lddo133]

If it's good in mono, it's good in stereo? Yes, that's somewhat true, but if you evaluate it with speakers, it's a wrong evaluation. Since all the songs are already in stereo, you should just listen to them in stereo and evaluate them in stereo. The original evaluation criteria are that stereo should be evaluated in stereo and mono should be evaluated in mono.

 

[Audionaut]

One of the qualities that I’ve always cared about and looked for in loud loudspeakers was a sense of sonic density, palpability, focus, and solidity to the Sonic images.

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

If you want, You can evaluate a loudspeaker on the basis of these measurement results without having heard it.

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


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.

Completely without flowery descriptions and audio poetry.

If the measurements in mono are good, stereo will not be bad.
You just have to make sure that there are no cancellations and boosts in the low frequencies due to room resonances and that the total energy in the high and mid frequencies corresponds to that which was also correct in mono.

 

[Hayabusa]

both both both both!
end of discussion!
pfff...

 

[Matt_Holland]

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

If you want, You can evaluate a loudspeaker on the basis of these measurement results without having heard it.

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


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.

Completely without flowery descriptions and audio poetry.

If the measurements in mono are good, stereo will not be bad.
You just have to make sure that there are no cancellations and boosts in the low frequencies due to room resonances and that the total energy in the high and mid frequencies corresponds to that which was also correct in mono.

I think @MattHooper makes a somewhat valid claim. It’s something have always appreciated from a good system. Buchardt S400mk2 are a good example of creating a solid, palpable stereo image.
A recent experience with KEF R and Q speakers gave a similar affect.
Now, I think room acoustics play a big role here so maybe it’s predictable using RT vs Frq.
The other factor I think it speaker pair matching.

 

[SIY]

Possibly I missed this discussion, but how would one test speakers which are deliberately designed for stereo, not mono, using a deliberately asymmetrical dispersion pattern, and have very specific placement requirements? E.g., Ken Kantor's designs? Is "on axis" the speaker directly in front of you so it beams off to your side (unlike when in use)? Or placed properly, L or R, with the speaker beamed toward you but displaced from center L to R? And would the speaker be placed on the wall as required by the bass loading?

(stolen picture)

 

[Dumdum]

Possibly I missed this discussion, but how would one test speakers which are deliberately designed for stereo, not mono, using a deliberately asymmetrical dispersion pattern, and have very specific placement requirements? E.g., Ken Kantor's designs? Is "on axis" the speaker directly in front of you so it beams off to your side (unlike when in use)? Or placed properly, L or R, with the speaker beamed toward you but displaced from center L to R? And would the speaker be placed on the wall as required by the bass loading?

View attachment 454703
(stolen picture)

Those speakers (if the crossover is designed well) will not sound like the low freq driver is behind the higher freq drivers, they will just sum to one single point source with all freqs being at eye/ear level… so asymmetry in the cabinet is not a thing to be concerned with

 

[Matt_Holland]

For stereo evaluation is there a standard speaker spacing and listening distance prescribed? How far apart the speakers are makes a big difference to the stereo presentation.

 

[Dumdum]

For stereo evaluation is there a standard speaker spacing and listening distance prescribed? How far apart the speakers are makes a big difference to the stereo presentation.

No, because every room and listening environment is different

 

[Arindal]

I imagine that finding recordings of voice that would down-mix from stereo correctly from mono would be quite easy,

In practice this is a pretty difficult one, as voices originating from a mono track (such as a single studio microphone) mixed for stereo will in this or that way being EQed and optimized for phantom localization, which means tonality for coming from two real sources at approx +30deg and -30deg while the perceived tonality should more or less match a real source coming in from 0deg. If you downmix such to mono, you will get level balance issues and imbalanced tonality alike, as the source is not coming in from 30deg anymore, nor is it a phantom source anymore.

but multitrack musical content, not so easy.

Maybe easier in practice, as some 5.1 and Atmos mixes use the center more in a way how a mono mix would be done. It might need some extensive testing, though.

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

Party yes, partly no. While the first two terms are raising eyebrows for sure, ´focus´ is pretty much another term for phantom localization width/accury, while ´solidity of the sonic image´ translates to phantom localization stability in different frequency bands. Both are standard criteria for evaluating stereo listening setups.

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.

I don't think so, as most of criteria belonging to ambience, imaging and transparency cannot be judged with pink noise.

For stereo evaluation is there a standard speaker spacing and listening distance prescribed?

Equilateral triangle is somewhat of a standard. The actual listening distance greatly varies with speaker type, the room´s reverberation time and grade of diffusion. I personal prefer a listening distance which will give an optimum of phantom localization accuracy and stability which would not change even when the distance is expanded by +10%. Other people might have different preference.

 

[watchnerd]

How much channel separation is good ve superfluous?

I’ve read basic stereo needs 20 dB, imaging improves with 30 dB, and the AES recommendation is 40 dB.

 

[Matt_Holland]

No, because every room and listening environment is different

My question was somewhat rhetorical, but I thought I would check if there was any standard setup guidelines (maybe from Toole’s test for example). This variable alone should make us favour mono testing for all speaker qualities except spatial presentation.

Does anyone know of some good mono recordings for speaker evaluation?

 

[MaxwellsEq]

For stereo evaluation is there a standard speaker spacing and listening distance prescribed? How far apart the speakers are makes a big difference to the stereo presentation

 

[SIY]

Those speakers (if the crossover is designed well) will not sound like the low freq driver is behind the higher freq drivers, they will just sum to one single point source with all freqs being at eye/ear level… so asymmetry in the cabinet is not a thing to be concerned with

It absolutely is. The polar pattern is HIGHLY asymmetric.

edit: For clarification, mirror symmetric L to R, but each speaker is asymmetric about its centerline. The drivers are offset "inward" for L-R placement, the baffles are cut at 110° rather than the usual 90°, and there's an HF absorption pad on the "outside" for L-R placement.

 

[Arindal]

a stereo signal mixed to mono will exaggerate correlated signals and reduce or cancel out of phase information.

That is one of the potential problems with downmixing to mono, but there are other implications such as the tonality of phantom sources vs. real sources and the different perception of monaural reverb vs. reverb spanning over the stereo base as intended.

But all psychoacoustic effects are from the original mix. Any spatial effects from room reflections are technically a form of distortion. Even though some people may enjoy the effect, it was not likely intended in the original mix.

I am not sure which spatial effects you are referring to, as I did not mean aspects of tonality except from the tonal balance of the added reverb in the listening room (which is maybe the easiest aspect to very using measurements).

Phantom localization, depth-of-field, proximity, ambience and how they all blend, are surely intensionally integrated by the mixing engineer. How this translates to an actual perception in a listening test, can only be verified with doing a stereo test.

There is a problem, though, that the original intension of the mixing engineer might be unknown. That is why I like to have such aspects judged by recording engineers who bring their own recordings, and I always prefer recordings which I have produced myself as well. In particular those concert recordings which I had the opportunity to listen to both in the auditorium (for example during general rehearsal), as well as in the control room, when the broadcast mix is done. Highly recommend this method as it gives the best idea of the mixing engineer´s intension IMHO.

I agree that the better matched the two loudspeakers are the better all of the stereo effects will be.

It is not only about matching the on-axis FR of the speakers, other important aspects factor in as well: deviations in FR over a broader listening window, interaural differences between early reflections for different frequency bands, tonal balance of the reverb added in the listening room, grade of diffusion thereof, and a few more. It is all a very complex picture, and you cannot judge even parts of it without doing a proper listening test in stereo.

 

[Audionaut]

I don't think so, as most of criteria belonging to ambience, imaging and transparency cannot be judged with pink noise.

Just give it a try.
Change the filters of the crossover circuit, the EQ, walk around the room.
There's really almost nothing that pink noise can't tell you about a loudspeaker.

 

[Arindal]

Just give it a try... There's really almost nothing that pink noise can't tell you about a loudspeaker.

Okay, let us imagine I have to choose new studio monitors for a rather narrow control room used for all sorts of stereo recordings incl. broadcast of acoustic concerts. I have three potential models, all within a range of linearity that they do not exhibit any tonal difference within in the listening window, measurements are flawless and no audible disturbance like distortion. But when listening to concert recordings in a stereo setup, I notice:

A - perfectly accurate localization of phantom sources, proximity thereof exaggerated with direct localizations overly close to the listener, ´dry, holographic picture´ staging, while ambience and reverb detached from the voices/instruments, distant, decorrelated and overly enveloping the listener. When doing mixes, I notice that mixing engineers are putting an overdose of ambience/reverb on the mix in order to balance the flaws of the loudspeakers

B - decent distance of imaging, depth-of-field as intended, wide ambience, good localization of vowels, localization instability problems with sibilance and the typical ´SSSS´ and `T´ of choirs flying around being localized in a different position compared to the voices themselves. Recording engineers are tempted to reduce sibilance bands or increase the relative level of spot mic signals with their fader to compensate for the localization issue.

C - deep, homogeneous ambience with stable localization and fair amount of depth-of-field, voices being enveloped by their resulting reverb and proximity as intended.

Could you draft an experiment please, with which I can reliably determine which of the three speakers is delivering correct imaging in terms of localization and proximity, solely by listening to pink noise in mono?

Change the filters of the crossover circuit, the EQ, walk around the room.

You are talking about discriminance testing, or am I getting it wrong? I am not interested in that. I want to know which of the three studio monitors is most likely to reproduce imaging and ambience as intended by the mixing engineers.

 

[Vuki]

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

If you want, You can evaluate a loudspeaker on the basis of these measurement results without having heard it.

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


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.

Completely without flowery descriptions and audio poetry.

If the measurements in mono are good, stereo will not be bad.
You just have to make sure that there are no cancellations and boosts in the low frequencies due to room resonances and that the total energy in the high and mid frequencies corresponds to that which was also correct in mono.

How do you train your hearing for pink noise? What is the ideal sound of pink noise?