Research Project: Infinity IL10 Speaker Review & Measurements

[pozz]

I don't have the burden of proof , this is your task to point out how what you hear correlate with the data to released and the data Olive produced.

That's what we are asking. You just never anwser this question.

It's a community effort. Contribute through research or data analysis.

I disagree.
Elevated treble would not account for "grunginess", it's a tonal matter and easy to eliminate with EQ.
This is something else.

Until people acknowledge that there life beyond frequency response (and directivity) we aren't going

What you've posted is for engineering/design only. It makes no claim on audibility.

 

[bobbooo]

It is not just that. It was a comparison testing of a dozen or so speakers against each other. That is how the listeners heard and evaluated the speakers and gave them scores. So if you want to be strict, then the only valid outcome is for those set of speakers being tested that way and nothing else. You can't extend the results or model to other speakers. And even the same speaker cannot be rated against other speakers not in the same test set.

As a result, you cannot use anything from the research to invalidate my listening results because it simply doesn't read on it. Nor would it read on the experience of anyone here going out to buy a speaker as they would most likely not be upgrading from one of the speakers in the Olive research.

You want to live by the rules? Those are the rules. Shall we throw out all the research, preference score, etc. now because of this?

This isn't a binary choice. I never said there should be absolute zero extrapolation of the model. Obviously there has to be some in order for it to have practical utility. But expecting a scientific model that was devised based on controlled, double-blind listening tests to explain all casual, sighted, unscientific listening impressions is an extrapolation too far, because these impressions are subject to innumerable additional uncontrolled variables (cognitive bias being a major one).

The strength of your impressions as data points isn't binary either. I never said they have zero importance, just very limited scientific utility in their current form, and so nothing can really be confirmed about Olive's model from them. There are some ways in which your listening tests could be made more scientific though, which could incrementally improve their reliability and import as data points, and again this is not an all or nothing affair.

 

[bobbooo]

You are assuming that only frequency response and directivity matter.

Of course there could be other explanations that would also have to be ruled out before possibility 2 could be confirmed, but in order for any of them to be confirmed, possibility 1 (placebo / cognitive bias) has to be eliminated. That should always be the first step in any scientific investigation.

 

[GXAlan]

Here’s my non-statically based analysis.

1) Dominant effect is FR. Amir didn’t like the JBL 4319 but he was OK with the Canon. While expectation bias (lowered for a fun speaker like Canon) plays a role, the Canon actually matches the Harman target curve once 1/3 octave smoothed. The JBL 4319 has a flat in-room response rather than down sloping and the tweeter was even hot in Amir’s room due to the L-pads.

We see subjectively, people not liking the Audyssey default curves (which have a 3kHz dip and does not follow the Harman curve).

We see subjectively that many customers prefer the Revel or JBL over KEF, likely due to the bass output.

2) Next is the off axis response. Old school was eliminating all reflections. Science based response is that reflections actually can enhance the sound. The impact of off axis response gets amplified by listening into mono.

3) That is all science based and no difference from Dr. Toole’s research.

4) To go beyond what has been published, there are a few branching points.

- Distortion
We have our classic THD which the Klippel measured and can be assessed on test tones. What isn’t captured is the Intermodulation Distortion. Then what isn’t captured is the Amplitude Distortion that represents the sum.

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4172235/

I would hypothesize that while THD for the M16 and the IL10 are below audible thresholds, total amplitude distortion is different.

@amirm
It would be interesting to use a 32 tone IMD test signal and capture speaker response on and off axis. Can then FFT to figure out actual response. Just for the M16 and IL10. Can do it in-room with moving mix. If the spin is similar but the in room 32 tone test being played back is vastly different, it would give us clues.

- Stereo/Mch imaging
Imagine if I had a DAC with SINAD of 130 dB for $120 and a DAC with a SINAD of 150 dB for $25,000. Everyone would go for the 130 dB “already better than Mola Mola” since the difference with 150 would be inaudible.

The argument for listening in Mono is that it allows differences in speakers to more easily be discerned but we see how the differences diminish with stereo. Potentially with Mch, the differences get even smaller. There may be a point where incremental improvements audible in mono are below the threshold of audibility in stereo.

I have not listened to Salon2’s in my own home but O have compared the 100 degree dispersion of the Canon’s against the asymmetrical horns of my JBL S/2600 and I still like the stereo imaging of the S/2600 more. It is possible that modern wide dispersion speakers with a center will trump the two channels. But the same concept applies.

There may be a point where a three channel front setup will outperform the performance of a two channel setup composed of speakers that are ordinarily higher performing in mono.

 

[Robbo99999]

Here’s my non-statically based analysis.

1) Dominant effect is FR. Amir didn’t like the JBL 4319 but he was OK with the Canon. While expectation bias (lowered for a fun speaker like Canon) plays a role, the Canon actually matches the Harman target curve once 1/3 octave smoothed. The JBL 4319 has a flat in-room response rather than down sloping and the tweeter was even hot in Amir’s room due to the L-pads.

We see subjectively, people not liking the Audyssey default curves (which have a 3kHz dip and does not follow the Harman curve).

We see subjectively that many customers prefer the Revel or JBL over KEF, likely due to the bass output.

2) Next is the off axis response. Old school was eliminating all reflections. Science based response is that reflections actually can enhance the sound. The impact of off axis response gets amplified by listening into mono.

3) That is all science based and no difference from Dr. Toole’s research.

4) To go beyond what has been published, there are a few branching points.

- Distortion
We have our classic THD which the Klippel measured and can be assessed on test tones. What isn’t captured is the Intermodulation Distortion. Then what isn’t captured is the Amplitude Distortion that represents the sum.

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4172235/

I would hypothesize that while THD for the M16 and the IL10 are below audible thresholds, total amplitude distortion is different.

@amirm
It would be interesting to use a 32 tone IMD test signal and capture speaker response on and off axis. Can then FFT to figure out actual response. Just for the M16 and IL10. Can do it in-room with moving mix. If the spin is similar but the in room 32 tone test being played back is vastly different, it would give us clues.

- Stereo/Mch imaging
Imagine if I had a DAC with SINAD of 130 dB for $120 and a DAC with a SINAD of 150 dB for $25,000. Everyone would go for the 130 dB “already better than Mola Mola” since the difference with 150 would be inaudible.

The argument for listening in Mono is that it allows differences in speakers to more easily be discerned but we see how the differences diminish with stereo. Potentially with Mch, the differences get even smaller. There may be a point where incremental improvements audible in mono are below the threshold of audibility in stereo.

I have not listened to Salon2’s in my own home but O have compared the 100 degree dispersion of the Canon’s against the asymmetrical horns of my JBL S/2600 and I still like the stereo imaging of the S/2600 more. It is possible that modern wide dispersion speakers with a center will trump the two channels. But the same concept applies.

There may be a point where a three channel front setup will outperform the performance of a two channel setup composed of speakers that are ordinarily higher performing in mono.

Regarding Stereo & Multichannel, wouldn't introducing another variable into Amir's listening test further confuse and complicate the aim to determine why the spinorama doesn't always match Amir's subjective listening test perceptions? Or are you saying that listening testing in stereo & multichannel should be the norm because that's how we use speakers anyway, and therefore listening in mono on one speaker is fundamentally non-representative & thereby not elucidating enough on the true potential of the speaker, which I think could be a fair point (and perhaps that way there could be more matching of the expected spinorama to the subjective listening test result).

 

[ROOSKIE]

But there is some logic in devaluating a speaker with HD3 that is about 5 times (in HD-percent) above the perception threshold. If there are other speakers that do not exceed this limit.
Just like an amplifier with SNR 90dB is not rated as good as one with 110dB - even if I would most likely not hear the difference.

Yes, well this is part of the issue. If you can't hear it you can not claim it is the source of the quality of your subjective experience. So if I don't like the sound of an amp with a SNR of 90Db, I can't likely claim that was the reason vs the 110db amp. (yes in very specific cases that may matter)

The fact that HD has a measurement does not by itself make it valuable. It has to have a reference. If I have 10 units of Guatemalan currency I can buy far less than if I have 10 units of USD or EUR.
Since most music is thousand of times more complex than a single test tone, we need to be careful treading the Harmonic Distortion subject. 5 times more may still not be meaningful in the exact same way most of us would agree that a frequency response error in a loudspeaker that is five times a .05db error (.25db) is not all that meaningful.
Anyway, again I say that we have not ruled out that that amount is audible to AMIR and others but your logic is a false one unfortunately. There is no logic yet in devaluing the speaker directly due to the HD measured.

There is of course a logic in devaluing it because it didn't sound good. That is the subjective component that can not be easily simplified.

 

[GXAlan]

Regarding Stereo & Multichannel, wouldn't introducing another variable into Amir's listening test further confuse and complicate the aim to determine why the spinorama doesn't always match Amir's subjective listening test perceptions? Or are you saying that listening testing in stereo & multichannel should be the norm because that's how we use speakers anyway, and therefore listening in mono on one speaker is fundamentally non-representative & thereby not elucidating enough on the true potential of the speaker, which I think could be a fair point (and perhaps that way there could be more matching of the expected spinorama to the subjective listening test result).

I'm just saying that another "next step" beyond the published Harman work is to study distortion and to study stereo/Mch.

For ASR, attention should be on IMD/THD since that's something that Amir is noticing/hypothesizing to differ. Maybe Klippel can offer a free trial of the IMD module.

 

[ROOSKIE]

Regarding Stereo & Multichannel, wouldn't introducing another variable into Amir's listening test further confuse and complicate the aim to determine why the spinorama doesn't always match Amir's subjective listening test perceptions? Or are you saying that listening testing in stereo & multichannel should be the norm because that's how we use speakers anyway, and therefore listening in mono on one speaker is fundamentally non-representative & thereby not elucidating enough on the true potential of the speaker, which I think could be a fair point (and perhaps that way there could be more matching of the expected spinorama to the subjective listening test result).

Maybe Amir should label his subjective reviews "a brief listen in mono".

This title or a similar one would be accurate and give folks a far clearer sense right off the bat of what amir's subjective component is going to consist of and its possible caveats.
He is not spending week with the unit, nor using stereo, nor spending a lot of time adjusting for all the EQ variables, nor playing with toe and such, nor measuring in room responses, nor adjusting for SBIR as the speakers are positioned, nor testing paired with a sub, ect.
All completely fine (and would be extremely time consuming, taking away from the measurements) but all things one might do with a week or a month on a set of speakers.
This is a fairly quick listen, a snapshot of how the speaker strikes him. Obviously still valuable, but since it is very different from what other reviewers usually do it might need to be made obviously clear as part of the review.
Call it something different.

 

[bobbooo]

I don't think that bit of jaggedness would affect the score that much; hard to imagine it'd do more than 0.1 points, but I'd have to mess around with the curve to be sure.

@MZKM calculated the score from the Harman measurements to be 0.5 higher than Amir's here. Of course the extra jaggedness might not be the only contribution to this discrepancy, but it seems to be one of the major differences between the measurements. The only way to confirm if this is indeed due to the mic cage (and not e.g. unit variation, age, other measurement discrepancies) and exactly how much it affects the score is to remeasure this (and a couple of other speakers) without the cage and compare the results.

 

[tuga]

2) Next is the off axis response. Old school was eliminating all reflections. Science based response is that reflections actually can enhance the sound.

What science are you referring to, preference polls? That's not science perhaps scientifically gathered data.
And no, reflections are not universally preferred.

The impact of off axis response gets amplified by listening into mono.

If you listen on-axis to a single speaker in the middle of the room what impact does the room/directivity have in the sound at the listening spot?

 

[ctrl]

Thanks. This supports the suspicions I raised earlier in regard to the audible distortion not having to necessarily be of the harmonic kind.

And again, we need to look beyond frequency response and directivity if we are to seriously address performance/sound quality assessment (both technical and listening).

One would have to look beyond frequency response and directivity if other factors were to have a significant average influence on the sound of the loudspeakers investigated.

But there is no evidence for this, since the algorithm used is already very reliable (ignoring whether the collected auditory impressions can be transferred to hearing in average living rooms).

If you could determine a possible influence of distortions of any kind with another score, put it in relation to existing scores and calculate an overall score, it probably wouldn't change much, because then you would only capture single outliers correctly, because a reasonably constructed loudspeaker actually has no hearing impression decisive distortions (at least at 85dB sound pressure).

The Spinorama and the score give us a measure of the probability of how the loudspeaker in question will sound to most consumers and how flexibly one can influence the sound of the loudspeaker.

Those who have tuned crossovers themselves know that 0.3 to 0.5dB over 1-2 octaves in important frequency ranges often decide whether a speaker sounds "okay" or "really good".
A score or spinorama cannot depict such a thing, but it can show trends and probabilities.

But when discussing a review of a loudspeaker, one should of course examine possible harmful sound influences down to the last detail in order to detect possible outliers and complete the evaluation.

Holy shit, this forum has a bad influence on me, I'm already talking like a hifi salesman

Discussion was not about if that HD3 spike is problematic or not but if it is related to his description about "grunginess and lack of clarity to everything it played". While it is probable that such HD3 spike is audible with test tone I can't say if it is audible with music he was listening with IL10, but I wouldn't expect that narrow HD3 spike to cause such effects as he described. What I would expect is, if audible with music, that it would add some harshness with the tones in that narrow range but certainly not to mess up the clarity of everything been played. Wouldn't you expect the same?

I'm with you there. The sole cause for the sound impressions might not be the increased HD3. On the other hand the HD3 could definitely have a share in it.

It was important to me to show that the different weighting by Shorter/GedLee is automatically included in the "correct" interpretation of HD diagrams.

What Geddes and Lee suggest in their formula (and others before) makes everyone "intuitively" evaluate the diagrams of harmonic distortions:
At what sound pressure and order do the HDs occur, pay attention to the Q factor of the HD.
Shorter and GedLee provide for each order (of harmonic distortions) a weighting (see above) which makes the evaluation easier and uniform.

Anyway, again I say that we have not ruled out that that amount is audible to AMIR and others but your logic is a false one unfortunately. There is no logic yet in devaluing the speaker directly due to the HD measured.

Yeah, the SNR comparison was unfortunate to stupid. 90dB SNR are not five times above the perception threshold, like the 1.5% HD3 at the measured speaker.

If you like the sound of two different speakers equally well, but one has a 1.5% HD3 distortion peak but costs one dollar less, would you buy the cheaper one because there is no "logic" to consider harmonic distortion above the hearing threshold until these distortions are perceptible with any kind of music? Who are you trying to kid?

 

[tuga]

One would have to look beyond frequency response and directivity if other factors were to have a significant average influence on the sound of the loudspeakers investigated.

But there is no evidence for this, since the algorithm used is already very reliable (ignoring whether the collected auditory impressions can be transferred to hearing in average living rooms).

If by algorithm you mean the Preference Rating then I don't agree that it is taking into account small yet audible problems, only tonal balance on- and off-axis.
In fact I find that it is an overly blunt tool and using it to assess performance instead of estimating preference is lowering expectations to a farcically shallow level. Enough for supermarket-level consumers but not hobbyist who care about sound quality and accurate reproduction.

I understand the attraction – if only life could be that simple – but if people believe that the Spinorama is capable of providing the necessary answers then we might as well decided that all speakers that are flat on-axis and have smooth directivity sound the same and be done with it. In any case I would consult with Toole, I'm sure he'll set you straight.

 

[GXAlan]

What science are you referring to, preference polls? That's not science perhaps scientifically gathered data.
And no, reflections are not universally preferred.

Yes. That was implied. Speaker preference has parallels to food preference. There are objective measures (nutritional value, temperature) and subjective measures. Clearly, nothing is universally preferred.

My favorites actually defy the classic Harman science. I really enjoy listening to the JBL 4319 which has a flatter in room response than the downward sloping Harman curve. My home theater uses asymmetrical horns up front (JBL S/2600) and the Canon S50 for rear speakers.

But my preference doesn’t negative the science of what sounds (and sells) the best to most customers. Nor the science that most customers cannot hear phase changes or harmonic distortion.

If you listen on-axis to a single speaker in the middle of the room what impact does the room/directivity have in the sound at the listening spot?

I am not actually sure that Amir listens with the speaker in the center of the room. His subjective tests are still affected by room reflections.

 

[Blumlein 88]

snip
I understand the attraction – if only life could be that simple – but if people believe that the Spinorama is capable of providing the necessary answers then we might as well decided that all speakers that are flat on-axis and have smooth directivity sound the same and be done with it. In any case I would consult with Toole, I'm sure he'll set you straight.

If all speakers have the same exact on axis and smooth off axis directivity you would be right. They don't however. Two identical speakers sound identical obviously. Two somewhat different speakers that radiate sound exactly the same way would also sound the same. Two speakers that are very close in sound radiation might sound close, but not the same.

Harman has shown at least that most people prefer speakers within a certain window of flat on axis and smooth directivity off axis.

Is there one most perfect speaker nearly everyone would prefer in all situations based upon flat on axis and some certain smooth off-axis directivity? Probably not though it appears the window for acceptibility is small enough to be helpful.

 

[Darkweb]

Kudos for keeping an open mind in your journey.

 

[tuga]

If all speakers have the same exact on axis and smooth off axis directivity you would be right.

That's the thing I don't agree with. Drivers and cabinets produce colouration and frequency response measurements of not illustrate all types of colouration. I am think we need a lot more research focusing on the audibility of the latter, with proper methodology.
The Spinorama is an oversimplification of a complicted matter.

 

[tuga]

But my preference doesn’t negative the science of what sounds (and sells) the best to most customers. Nor the science that most customers cannot hear phase changes or harmonic distortion.

In the scope of high fidelity or accurate reproduction of the recorded signal it essential that we raise our expectations; the general public couldn't care less about sound quality, they listen to mp3 and stream from YouTube.


I would rather people not used the word science when referring to preference polls. Available research is immensely more accurate. Science gives it a false sense of authority...
Researching what sells most requires models and sales numbers, I've not seen numbers of the most sold speakers.

 

[fredoamigo]

@tuga

That's the thing I don't agree with. Drivers and cabinets produce colouration and frequency response measurements of not illustrate all types of colouration. I am think we need a lot more research focusing on the audibility of the latter, with proper methodology.
The Spinorama is an oversimplification of a complicted matter.

I don't know if you've read toole's book? but what he said some time ago on the forum is very interesting.
https://www.audiosciencereview.com/...r-perception-of-audio.501/page-16#post-130632

""""The consolation is that comprehensive anechoic data, suitably processed and presented (the spinorama), offer considerable insight, as is explained in great detail in my book. Can physiological/psychoacoustic factors add value to these analyses? Obviously yes, because such data are already part of the analysis - the audibility of resonances being one of our (Toole and Olive) contributions. Applying simple interpretations of this information Olive was able to show very high correlations between real (from double-blind tests in a normal room) and predicted (from anechoic spinorama data) sound quality ratings on loudspeakers - 0.86 for loudspeakers of all sizes and prices, and 0.996 for similar bandwidth bookshelf speakers (p < 0.0001). This is not accidental.

The big missing ingredient in the analysis of loudspeaker performance is a measure of non-linear distortion that reliably correlates with perception. The common harmonic and intermodulation metrics are almost useless. That is where perceptual (simultaneous and temporal masking) factors are absolutely required in the metrics. ""

 

[tuga]

@tuga

I don't know if you've read toole's book? but what he said some time ago on the forum is very interesting.
https://www.audiosciencereview.com/...r-perception-of-audio.501/page-16#post-130632

Like I said we need more research with better methodology.
In my view a lot of the existing research/testing was oversimplified, trading resolution and effectiveness for practicality.

 

[TimVG]

At least we have Dr. Olive's attention.