What is Alan Shaw on about? (is "coloration" unmeasurable?)

[sq225917]

To be fair, he's laying a voyage of discovery pathway for potential customers to get on board with as it and they progress. No doubt there's a new product at the end of this for those able to make the jump and the satisfaction of sticking with technology they understand for those who can't.

A bit of an all things to all men story.

 

[fpitas]

View attachment 191855

The cute owl is the best post in the thread!

 

[CapMan]

Think it's pretty widely accepted to primarily be an artifact of trying to correct speakers with nonlinear off axis anomalies combined with the way 2 ears and a brain "filter" room reflections being rather different from the way an omindirectional mic and computer tries to do it. Perhaps there is something yet to solve with the later, but I'm not sure a self professed golden ear trying to measure the golden is going to wind up there.



Erin's 365C measurments show a broad and quite pronounced (maybe +3dB) on-axis rise and shelf starting at around 7Khz and extending to ~17Khz, relative to the broad average of the preceeding region (~700Hz to 7Khz). Would not at all be surprised at someone descibing subjective sibilance there.

Predicted in room response was a nice gentle slope down - I guess that is what I hoped I’d hear

 

[DSJR]

Harbeth do have active 'form'. They made a limited number of active M40s, which weren't pursued as a commercial product. The active M40s used opamps in a 'classic' active arrangement, much as Meridian did before their DSP-based actives.

Going DSP-based seems to me the only sensible progress a company like Harbeth can make. Passives have gone about as far as they can.


S.

Don't forget the HHB Circle models with purple cones active and passive. Harbeth made and can still service them I gather and they had the upper mid lift that many circa mid 90's pro speakers had..

 

[mhardy6647]

The cute owl is the best post in the thread!

Owls Are Impressed With Massive Stick | 'Woah, check out his stick!' | By LADbibleFacebook

'Woah, check out his stick!'

www.facebook.com

We now return you to our regularly scheduled topic.

 

[fpitas]

Owls Are Impressed With Massive Stick | 'Woah, check out his stick!' | By LADbibleFacebook

'Woah, check out his stick!'

www.facebook.com

View attachment 221564

We now return you to our regularly scheduled topic.

I like where this thread is headed!

 

[majingotan]

He already did that, and in one of his post in this tread he says that they are disassembling it: https://www.harbeth.co.uk/usergroup/threads/digital-signal-processing-and-loudspeakers.80536/page-2 (you need to be member to view the forum, membership is free and does not mean you need to be a customer)

View attachment 221482

He probally means he wants to measure the drivers, the amp and the dsp seperatly like he said further in the disucssion

Just inverse his definition of “coloration”. He probably meant that the DSP applied is so clean (just inverse this as coloration) that his system doesn’t have that giant 2nd and 3rd schlong of harmonic distortion and many grass of IMD distortion sprays from non oversampling DAC (pretty much opposite of DSP)

 

[thewas]

That would suggest he hadn’t done a lot of reading before making DSP adjustments, if he’s doing narrow-Q adjustments in higher frequencies. I guess he’s got to take this journey on his own, though, and document along the way.

Also that linear on-axis response is not enough if the directivity is not continuously smooth, if all loudspeaker engineers would have read Toole's book it would have spared them a lot of unnecessary trial and error paths that have been passed too many times already from most.

 

[krabapple]

I bought a Monoprice 365 THX centre ‘blind’ based on Erin’s review and positive measured response with klippel. It’s certainly a lot of speaker for £365, but vocals were sibilant and spitty sounding to me.

.....in your room. With some unnamed recording(s)

I had to override the bias the that good measurements created and go with my ears on this one

Alternately , you could have considered what factors might cause the sound you heard.

 

[CapMan]

.....in your room. With some unnamed recording(s)

With all the broadcast TV and the movies I watched.

Alternately , you could have considered what factors might cause the sound you heard.

Listening was done on axis (it’s a centre speaker after all). The room is acoustically treated.

I ran speaker calibration on my Marantz AVR , but found the vocal presentation was sibilant irrespective of whether this was turned on or off. I don’t have any other control on the AVR to manually EQ the centre so if Audyssey doesn’t fix it I’m at a dead end I think.

Perhaps I could have run REW and looked for anomalies, but I am already on that hamster wheel with my mains! My room is acoustically treated and with no placement or manual EQ options on the centre I’m not sure there would have been any point ..

 

[krabapple]

It's good to present such details when posting on AScienceR.

Now the question would be, does the Klippel data indicate that the speaker is inherently sibilant?

If it does not, either the measurement was taken wrong, or something else is the cause. Sibilance is not a mystery in terms of transducer frequency response. It shows up in measurements from 3-8 kHz.

A perception of 'sibilance' can also be a byproduct of high frequency hearing loss.

 

[DSJR]

I noticed on one of the 'energy window' pics that there was a bit of a 'spread' in the mid kHz region

Yes I know, it's the vertical one -

I wonder?

 

[Chromatischism]

I noticed on one of the 'energy window' pics that there was a bit of a 'spread' in the mid kHz region

Yes I know, it's the vertical one -

I wonder?

That will affect the predicted in-room result, so look there.

 

[Soniclife]

The compression test for this speaker is most odd, and could be the issue.

 

[DSJR]

That will affect the predicted in-room result, so look there.

I did...

There seems a little 'lift' above 3kHz while still following the gentle downward slope and the wide dispersion may well offer more room for room reflections to add to the sibilance area? I agree the compression test shows something odd at 2khz or so.

All good fun, but obviously the subjective effects will change from room to room. I may well benefit from the greater sense of 'top' here but others with better hearing could well find it more oppressive. Smaller rooms and listening distance will also have some effect as well as in this case, the basic volume the speaker is played at, depending on the centre channel content.

I suspect this is one reason why any 'new' Harbeth goes through hours of final listening to make damned sure the balance and driver 'blending' suits. A dB or two either way in tweeter level may well make all the difference here.

 

[CapMan]

I suspect this is one reason why any 'new' Harbeth goes through hours of final listening to make damned sure the balance and driver 'blending' suits. A dB or two either way in tweeter level may well make all the difference here.

I would like to see Harbeth introduce a centre speaker - it would seem to marry well with their design focus on vocal reproduction.

 

[Willem]

Alan Shaw just posted an update on their work. The first thing they did was to connect their own dsp module to the competitor speaker to see if they could remove the "coloration" that he objected to. The second thing they did was to collect all prototype woofers they developed over the years, successively mount them in a C7 box, create a dsp crossover for maximum flatness, and then measure and listen to the result. The aim is to find out if there is anything in the driver designs that may be responsible.

 

[ThoFi]

Alan Shaw just posted an update on their work. The first thing they did was to connect their own dsp module to the competitor speaker to see if they could remove the "coloration" that he objected to. The second thing they did was to collect all prototype woofers they developed over the years, successively mount them in a C7 box, create a dsp crossover for maximum flatness, and then measure and listen to the result. The aim is to find out if there is anything in the driver designs that may be responsible.

…are there, 2 months later, any updates from Alan Shaw…?
Did he share any new information, any measurements?

 

[ROOSKIE]

The view that an electronically flattened response does not immediately sound right is not an exceptional view either: many have argued that e.g. Dirac used to flatten response above the transition frequency does not sound quite right, without providing much explanation either. So let us see what he comes up with.

Toole touches on this well enough in his book and it is clear that at the time it was published he is not a big fan of micromanaging in room response nor has doing so been shown to correlate with improved subjective sound in a blind test.
Sure using it as a broad band tone control should be fine above the transition zone - expecting to fix design and tonal flaws with something that can only see the in room steady state is rather polar to what Floyd Toole's whole premise is in his book - that speakers that measure well anechoically sound the best in a real room and the mind of the listener adapts to the room.

The explanation is clear, if you have a 2db bump at 1200hrz in room that is not there anechoically and then Dirac or PEQ removes this bump you now have a speaker with an approx -2db dip anechoically. But blind testing thus far has shown that a speakers with the best anechoic measurements are the ones that win, it has never shown that speakers with such and such in room response win.
Speakers with great anechoic responses do have a general downward trend(that varies in 'look' based on several factors especially dispersion) in real rooms but this not the goal - the goal is flat anechoic on axis responses with smooth, even falloff.

The OP is not correct in terms of Toole that Steady State in room responses can be enough information to understand what is being heard. It can't be more clear from his book that you first need a complete spin. That is the whole premise of the book.

 

[ROOSKIE]

Predicted in room response was a nice gentle slope down - I guess that is what I hoped I’d hear

This is unfortunately not what you will hear with any speaker.
It is simply a side effect and often overly weighted, likely as folks can take these type of measurements at home and hobby up(I do it). Also Dirac and the like have us focused on this as Dirac can adjust in room responses for $.
That is why there is a spinorama, which consists of measurements of major importance.

The predicted in room response is really just to show some correlation between the Spinorama and a real room. It is not what is heard, the hearing of which is happening with a much more complex system that is not even remotely as limited as the in room steady state. (your ears, brain & mind)

I have had a hard time being able to tell exactly where sibilance will come from in the test of your THX-365C, though the measurements to my eye are meh - in terms of the horizontal off axis which you will hear bouncing off the walls and recombining in some various forms with the direct sound whether you are sitting on axis or not. That said it is possible your copy has a defect, a bad tweeter or a lose part. Are you able to measure HD & some sort of gated response, also you could take a steady state in room and compare it with the prediction which could be fun.

As other stated the perception of sibilance can be sourced to a variety of issues. Is it there when you listen to content that you know was recorded without it? Many, many recordings contain sibilance and therefore any accurate replay will also contain it.

The location of a speaker in a cabinet or causing something nearby to resonate could also be a source.

Anecdotally, I have experienced extra sibilance it some speakers that appeared they would not have it, not sure why. Also some times I am far more bothered by it vs others again not sure why.

IMD and Doppler distortions are almost never published or measured well, they can be so complex and attempts to recreate them may far short of reality. (though, like HD tests they still are helpful to point in the right direction)