Equalizing loudspeakers based on anechoic measurements (community project)

[Robbo99999]

That's spot on and what's giving me the best listening experience so far. I'd like to add that depending on how you're measuring and what the distance and room qualities are like, the predicted slope may not match the measured slope, especially in the treble. Going by the estimate does give us an 'untainted' picture that doesn't include potential measurement artifacts (let's say a reflection from the back of the couch).

I added an EDIT too after you quoted my post, a potential Step #5. What do you reckon to my Step#5??

 

[bobbooo]

If I wanted to EQ my speaker based on anechoic measurements, I think I would simply EQ the LW to flat and call it a day. Any remaining problems would be caused by uncontrolled directivity, which can't be fixed with EQ anyway.

I suspect the 'ideal' LW slope from Olive's paper being -0.2 (for both the bookshelf and all-speaker tests) and not exactly zero is not a coincidence, so I don't think EQing the LW slope to zero would be wise.

 

[TimVG]

Equalizing to the PIR is the only universal approach that works.

It does not work -audibly- on two of the three pair of speakers that have been tested by Amir and are in my possession. The score does not match the listener preference in this case.

I added an EDIT too after you quoted my post, a potential Step #5. What do you reckon to my Step#5??

That's a good strategy, I doubt it will make an audible difference but not everyone has room for 10 filters - I usually end up somewhere around 6.

 

[TimVG]

Equalizing to the PIR is the only universal approach that works.

How many speakers from your spreadsheet have you actually in your possession and heard with your filters? Equalizing to the PIR is not something that can be universally applied as explained a couple of times already. Just because the score increases doesn't mean it sounds better, unlessI'm deaf.
As others have also pointed out, there is a limitation to the model which we've stumbled upon. It's as simple as that.

 

[Robbo99999]

That's a good strategy, I doubt it will make an audible difference but not everyone has room for 10 filters - I usually end up somewhere around 6.

Gees, with my JBL 308's and my room I think I'm gonna need double that! Ha, but yeah, that was an idea I had just now on how to consolodate lots of layers of potentially conflicting filters. Actually, that approach would really help me in the last stage of converting it to my miniDSP (for TV/movie watching) that only has 2 banks of 6 filters (so 12 filters), whereas for playing music from my laptop I use Equaliser APO so unlimited number of filters there. (Perhaps less filters means better sound, I don't know, maybe it means less ringing....that's just my intuition based on some snippets of knowledge I tried to string together).

 

[Robbo99999]

This is my main concern about this whole "EQ based on score" project. The Olive model was designed to correlate with preference on a typical sample of loudspeakers. Just because something correlates with good sound in a typical speaker doesn't mean that it causes good sound, and for that reason, using the model as an optimization target might be missing the point entirely. (Though I will gladly concede that it's likely better than doing nothing.)

This is a good point, and it tallies with points I've heard from experienced people in the field like Oratory1990. One of his points he's made in the past is that conforming exactly to Headphone Harman Curve is not the main reason for headphones sounding good or bad, ie a higher Preference Rating once you go past 90 does not net you better sound. Other factors come into play, but he concedes it's important to be close enough to the Harman Curve, but that there is no point in just chasing ever increasing accuracy to the Harman Curve beyond about 90 odd Preference Rating. I guess there's a similar logic applied to speakers. It distills to the fact that it's futile to chase an approximation to the nth decimal place.

P.S. I'm not pouring water on this thread, as this thread is broadly about using spinorama data to EQ a speaker, and I think that's a very valid & worthwhile approach. I think that's probably different to chasing Preference Ratings to the nth decimal place.

 

[Hactar]

Yes, I also cut the relevant collumns and make a new txt file. I'm not sure if any further corrections need to be performed, perhaps the microphone cal file needs to be embedded, which I hadn't considered up until now - perhaps someone can provide the answer.

Are you able to share your spin data for the F206?

 

[Maiky76]

@Maiky76 Do you need the Estimated In-Room .txt file for your calculation or will the ON, LW, ER and SP suffice? I have them for the F206 and C208 which are setup permanently and allow me to test for longer periods of time without cluttering up the house.

ON, LW, ER and SP suffice.
Are they scanned from pictures?
That is going to be much more difficult as I believe that Harman to designing their speakers with a certain target in mind that they probably have thoroughly tested. In particular the larger ones that have to cope with far less constrains. If you calculate the PIR (and have a close look at the SPDI) of the F228Be is does resemble one of the preferred curve from the In-room EQ study. I don't think this is a coincidence...
https://speakerdata2034.blogspot.com/2019/03/spinorama-data-revel-home.html

Just a point,
The data that is measured by @amirm is something that is invaluable for the hobby crowd, most maker don't use such advanced gear.
It a the proverbial shift in paradigm for the hobby and a different league compared to discussing ad nauseam unsubstantiated opinions in other places.
This data is very interesting to select speakers and see which one are the best starting point, EQ friendly.
I am of the opinion that anechoic data can only get the user so far as people usually listen is a real (semi-reverberant) rooms.
The anechoic Spinorama data is just a partial insight that might enable the users get more out their speakers but that still needs to be coupled with in-room EQ to fulfill their full their potential. Arguably, the in-room EQ could be seen as more critical.

I do not pretend that the EQ I make (still in WIP!) is the end of it.

JL Ohl and Igor have been doing that at a much higher level (Igor is in the music business and JL is in the installation business) including the room for years.
https://www.audiosciencereview.com/...ts-community-project.14929/page-4#post-468791

 

[Maiky76]

Do you mean if the overall magnitude of the PIR slope increases? If so, this means the formula (incorrectly) rewards extreme PIR slopes, both bright and dark (positive and negative). In practice there will likely be very few speakers with a PIR slope above 0, so the excessively bright scenario is likely less of a problem, but the overly dark scenario might be more common. I think this all means it's vital to do a 'sanity check' on each speaker's computed score by cross-referencing it with its PIR slope value, which should be close to the 'ideal' -1 value. Alternatively, as there is a slight discrepancy between the ideal PIR slope of the bookshelf (-1.2) and the 'all-speaker' (-1) tests, which Olive suggests could be down to the latter on average having wider dispersion, maybe the best slope value to look at would simply be the on-axis, which for both tests have an ideal value of exactly 0. (A further cross-check could even be done with the listening window slope, which again agree on an ideal value, -0.2, between the two tests.)

For EQing, I think the best approach would then be to optimize for the highest Olive score, but only if the slope of the (in order of importance) on-axis / listening window / PIR move further toward their 'ideal' values, or stay the same. Otherwise the above-mentioned shortcoming of the Olive formula might be 'gamed' for a higher score at the expense of worse overall tonality. @Maiky76 is this something that could be implemented in your ideal EQ computation?

That's more or less what I have doing lately, optimizing the score while not compromising the ON/LW or vice versa.
Anything else could be potentially implemented however with no validation and no target to work towards; it's all moot.
That's why one can't separate the EQ from the complete design of the speaker. No amount of EQ can fully compensate for a poor directivity.
The Spinorama is one of the best way to see that. One can't arbitrarily optimize all the curve separately as one EQ acts the same way on all direction, unless one control the directivity at the same time (DSP beam forming).

 

[bobbooo]

That's more or less what I have doing lately, optimizing the score while not compromising the ON/LW or vice versa.
Anything else could be potentially implemented however with no validation and no target to work towards; it's all moot.
That's why one can't separate the EQ from the complete design of the speaker. No amount of EQ can fully compensate for a poor directivity.
The Spinorama is one of the best way to see that. One can't arbitrarily optimize all the curve separately as one EQ acts the same way on all direction, unless one control the directivity at the same time (DSP beam forming).

Could you explain how DSP beam forming would work?

 

[Maiky76]

Could you explain how DSP beam forming would work?

see there:

• Controlled and constant directivity down to 100Hz via large drivers, waveguides and/or beam forming

https://www.stereophile.com/content/bang-olufsen-beolab-90-loudspeaker-measurements
https://www.stereophile.com/content/dutch-dutch-8c-active-loudspeaker-system-measurements
https://www.stereophile.com/content/kii-audio-three-loudspeaker-measurements
https://www.audiosciencereview.com/forum/index.php?threads/apple-homepod-measurement.8425/

In particular the Kii and B&O you can change the directivity of the speaker to shape the PIR somewhat independently from the the ON.
If the ASR community wanted to make a DIY speaker I would suggest this kind of design as opposed to another shoebox.

 

[richard12511]

So this only works if you have anechoic measurements? From what I understand, with REW it's possible to get anechoic measurements above a certain frequency, right?

 

[QMuse]

This is my main concern about this whole "EQ based on score" project. The Olive model was designed to correlate with preference on a typical sample of loudspeakers. Just because something correlates with good sound in a typical speaker doesn't mean that it causes good sound, and for that reason, using the model as an optimization target might be missing the point entirely. (Though I will gladly concede that it's likely better than doing nothing.)

I beg to disagree here.

Yes, Olive model was designed to correlate with preference on a typical sample of loudspeakers, and as such it presents a well defined design target for loudspeaker designers. It is based on the measured output of the speaker, not on technology used to generate that output, so if DSP in the form of the speaker EQ helps to make the score better it presents an equal value if better box design was used or DSP XO was used to overcome shortcomings of analog XOs. It doesn't matter, all it matters is that measurable output was enhanced, which was reflected in the score. So yes, reverse enginerring the score in a way to EQ a speaker in such a way that will make the score higher means it will have better measurable output and that it will indeed sound better. Point here is that getting the spinorama measurements right in order to achieve a better score is universal design target and EQ-in a speaker is just another way to help doing it.

If I wanted to EQ my speaker based on anechoic measurements, I think I would simply EQ the LW to flat and call it a day. Any remaining problems would be caused by uncontrolled directivity, which can't be fixed with EQ anyway.

I beg to disagree with this as well. Olive's scoring system, while surely not being perfect, is the best we have so far and to my knowleldge it is the only one that has been verified following proper scientific principles. As such, it clearly favours smooth PIR more than it favors flat/smooth LW making @flipflop 's logic right.

So far we don't know if the formula would stand a test with sample made of modern speakers, like the ones tested here. But we also don't know that it wouldn't. Hopefully a better formula will be invented in the near future, but so far I have seen absolutely no proof that the formula should be modified in a way you're suggesting so that LW takes absolute priority over ER/SP/PIR. On the other hand current formula verion has been properly verified with a set of sample speakers. It may easilly turn out that it would stand a test with set of modern set of speakers, but even if it had to be modified there's absolutely no proof that it will be in a way you're suggesting, making the LW curve of all curves.

So, here's what I think: as long as the current version of the formula is the only one being validated we should stick to it and we should set speaker EQ target to achieve better score because it does mean the speaker would sond better, because the fomula has been verified to correctly assign better score to better sounding speaker. This means that making PIR smooth, even on account on on-axis flatness with speakers with flawed directivity, we can expect a better sound due to the better score. So yes, EQ-ing a speaker to a smooth PIR is IMO definitely a valid speaker designed target, as much as making a smooth PIR via XO improvements or baffle step optimisation. And as much as it sounds counter-intuitive, smoothing PIR is a better speaker optimisation stratgey than flattening on-axis and/or smoothing LW, so in the situations with flawed speaker when you have to choose what to EQ choosing to smooth PIR is your best choice, because scoring system says so.

 

[QMuse]

@TimVG If your idea here is to check via series of blind tests which EQ strategy (smoothing the PIR vs flattening the ON) gives better results I really think you should be comparing measured instead of simlated in-room response with various filters.

 

[TimVG]

@TimVG If your idea here is to check via series of blind tests which EQ strategy (smoothing the PIR vs flattening the ON) gives better results I really think you should be comparing measured instead of simlated in-room response with various filters.

Well the thread is based around EQ on anechoic data. In-situ EQ has never left me very satisfied I'm afraid.


In any case. Toole has written and talked about often how the direct sound is dominant in terms of listener perception and never advises to optimize the response in favour of the PIR.

At the 34 minute mark:

and indeed, if I were to optimize my F206 according to the PIR, I'm screwing up the listening window (which is already an average) in not uncertain terms. This is audible in a blind test and again, not preferred.

 

[QMuse]

Well the thread is based around EQ on anechoic data. In-situ EQ has never left me very satisfied I'm afraid.

PIR is calculated based on anechoic data, but you should always check it vs actual in-room response measured with MMM. If your actual in-response north of 400Hz significantly differs from PIR you have room/positioning issues which need to be adressed.

In any case. Toole has written and talked about often how the direct sound is dominant in terms of listener perception and never advises to optimize the response in favour of the PIR.

I'm aware of it, but the simple truth is that contradicts Olive's scoring formula.

 

[TimVG]

ON, LW, ER and SP suffice.
Are they scanned from pictures?

Yeah. I've come up with some EQ files of my own - would be interesting to see if they match your matlab version closely again.

 

[TimVG]

Are you able to share your spin data for the F206?

Please see the post above ^

 

[QMuse]

That's more or less what I have doing lately, optimizing the score while not compromising the ON/LW or vice versa.
Anything else could be potentially implemented however with no validation and no target to work towards; it's all moot.

I fully agree with both of your points..

 

[JIM_82]

Hello everyone, first contribution (sorry for my English)

PIR is a prediction of the response for a "standard" room but not real.
The part of direct sound at the listener position depend of many parameters in real life, reverberation level for each octave band is not equal and distance listen depend of room size, listener prefence. Directivity of loodspeaker is variable with the frequency.
The only invariant parameter in the spinorama is the direct sound (in axis and in listening windows).

The speaker can be oriented out of axis, it's so a part of possible optimization.

The other part of spinorama contribute to the many reflexions and diffuse sound in the room.

In practice, the weight of the ON vs LW and PR is variable and so the EQ.
MMM measurement is a simple way to compare the measurement of the IR vs PIR of spinorama.

For example, X curve in theater (equivalent to IR) has different slope depending of room size.