Rational Acoustics



PhillipIvanPietruschka
February 20th, 2011, 05:51 AM
Has anyone observesed (or aware of relevant research) a correlation between a highly non linear phase response (particularly in mid band frequencies) and poor speech intelligibility?

Recently I worked on a system that had markedly worse inteligibility than I am used to. There could be a wide variety of reasons for this, granted, but the following at least is true: the main system which I supplemented with a variety of fill speakers, had two full phase wraps above three hundred Hz. The system as a whole had impressively even coverage and a more than acceptably even frequency response. All the fill speakers were treated with allpass filters, measured in-situ, to achieve maximum consistency with the main (in house) system.

So am I chasing a red herring here, when the actual culprits are complex room acoustics problems (it IS a weird space, but not overly reverberant, echoy, etc) and poor technique on behalf of the actors? Or has anyone else had a similar experience?

Best,
PhiilipIvan

PhillipIvanPietruschka
February 20th, 2011, 05:59 AM
I should add, I started thinking along these lines because most of the systems I work on have extremely even phase responses above 200 or 300 Hz, and I have repeatedly achieved acceptable or better intelligibility in notably reverberant houses, even in seats which are really somewhat offaxis and deserving of additional fill systems (but for time and money etc).

Ferrit37
February 20th, 2011, 03:23 PM
Hey Phil,
Prehaps it has more to do with Listening difficulty, a metric introduced by Sato, et al., 2005 whilst studing classrooms/word intelligibility.
It seems to me that the brain has less work to do sorting through the phase stuff and relaxes more.
I've noticed less hearing fatigue after spending a few hours in front of more coherent systems.

Arthur Skudra
February 20th, 2011, 09:48 PM
There's a two part article that appeared in the August and October 2006 issues of Live Sound International written by John Murray on "Tuning Phase" that are definitely worth a read. Consider that a non-linear phase response is "distortion" of some sort in the time domain. Thus, it would make sense that the less "distortion" the better for intelligibility.

Dr. J
February 21st, 2011, 05:17 PM
There's a two part article that appeared in the August and October 2006 issues of Live Sound International written by John Murray on "Tuning Phase" that are definitely worth a read. Consider that a non-linear phase response is "distortion" of some sort in the time domain. Thus, it would make sense that the less "distortion" the better for intelligibility.

Arthur -- I have tried to get these articles but they only have one of them. They also want to charge for them even though their current editions & archives are free. If I remember right -- the August issue was the only one they had & the other one sold out and was out of print.

I asked -- "You archive these don't you?" Their response was sometimes. The whole conversation was just wierd.

I think I am going to try to get a hold of John myself to see if he can make those available some how. I would rather give him the money anyway.

If anyone else has had any luck -- please let us know how we can get a hold of those articles. I will do the same.

Arthur Skudra
February 21st, 2011, 08:08 PM
Live Sound International has gotten strange lately, their parent company has made a royal mess of the Live Audio Board with the transition to new software, seems like the magazine is not faring any better online. I stopped paying attention to the magazine when they stopped the complimentary print subscriptions to Canada. They used to have back issues archived digitally, available online, now nothing. I can't even login (even though I'm a subscriber to the digital edition) to see the current issue! What a mess! :mad:

PhillipIvanPietruschka
February 23rd, 2011, 07:47 AM
Well, having re-read it, Arthur, although this tangential to my primary query here, do you have any more articles or advice to offer on the application of FIR filters?

I believe the Soundweb Londons I regularly use are capable of 'doing' them but really am not sure about how I would go about using or programming one.

For sometime I, as an intellectual experiment if nothing else, have been curious to sit down with Soundweb and one of the speakers I am most familiar with (so probably a UPJ) and see what happens if I use all pass filters to achieve a flat phase response through the entire passband. I am aware there would be non trivial latency involved. Whilst I have an idea of how I might do this with IIR all pass filters, FIR filters I am unfamiliar with.

Arthur Skudra
February 23rd, 2011, 08:03 AM
In this situation, perhaps a UPJ might not work as well, since it already has a relatively linear phase response, kinda hard to make something that good noticeably better, and defeating the filters already built in would be challenging.

If you have access to the Fulcrum Acoustic loudspeakers, you can switch between their non-FIR and FIR processing settings to hear what I think is a very noticeable difference particularly in intelligibility. You can do the same with some older EAW products comparing an older tunings with one of their "focused" settings that they have produced. Some remark that it's similar to taking a veil off the loudspeaker. Maybe you can convince Ferrit to come down there with a Martin MLA system and give you a demo, I'm still waiting for mine, and I live only an hour away from their North American headquarters! :) (Hint hint Ferrit :D )

PhillipIvanPietruschka
February 23rd, 2011, 09:37 AM
Arthur, I agree wholeheartedly that the speaker I mentioned is very much 'not-broke' but it is one my bread and butter speakers. But since that is the standard of phase Linearity / speaker pedigree I typically work with, I confess to being more fascinated to if the Very Good can be improved, even marginally, than to if I can raise other speakers to a standard I typically expect by default.

With regards to the Fulcrum speakers and the Martin MLA I am not aware of the presence of either anywhere on my continent. But it might be that I just wasn't paying attention (I gave up on most of the local industry publications some time ago).

PhillipIvanPietruschka
February 23rd, 2011, 03:59 PM
If there is such a difference in quality why have Fulcrum made these filters switchable? Is there some other non-trival trade off involved (such as latency)?

Arthur Skudra
February 23rd, 2011, 08:51 PM
If there is such a difference in quality why have Fulcrum made these filters switchable? Is there some other non-trival trade off involved (such as latency)?
The only reason why they have/don't have FIR filter settings is to support processors that are capable or not capable of handling the advanced filters required. More info here:
http://www.fulcrum-acoustic.com/nm_webkit/libs/download.php?file=/downloadfiles/Processor%20Configurations/Implementing%20TQ%20Processing%20README.pdf

When I was auditioning the Fulcrum products, they did a demo of their speakers with and without the FIR filters, just to demonstrate the improvement in sound. EAW did something similar at a trade show when they first introduced their "focused" settings. Granted both Fulcrum and EAW are doing more with FIR filters than just smoothening out the phase response curve.

PaulTucci
February 23rd, 2011, 10:18 PM
Phillip,

I think we need a qualifier in your statement because where in time we view a loudspeaker's phase response from...means everything.
A perfectly flat phase loudspeaker can be made to measure with phase wraps IF your measurement offset is incorrectly set. Conversely, A poorly behaving loudspeaker can be made to have a flat phase response in PART of the frequency range. The flat phase area of the measurement readout is where the offset delay is most accurately set. So I ask you this. "What area of frequency was most flat in your measurement?"

Ferrit37
February 23rd, 2011, 10:25 PM
Hey Arthur,
working on Nashville area next, probably,
There's no MLA lurking in canada :)

PhillipIvanPietruschka
February 24th, 2011, 12:03 AM
Paul, i understand this very well. But without saved traces in front of me I can not offer a categorical answer except to say somewhere in the high-mid and treble, and I dont believe I still have the traces from that project still saved to disk. The characteristic phase deviations of a mis-set reference delay line are not what I was observing.

PaulTucci
February 24th, 2011, 08:53 PM
Phiilip,

So if you had a legit measurement in the mid/high area and still had a lower bandwidth that was wrapping, perhaps the xover timing for that entire bandwidth was wrong.

Don't know if that would measure as lower intelligibilty, but I agree that would be a type of distortion and fatiguing to listen to.


Regards,
PT

PhillipIvanPietruschka
February 25th, 2011, 12:35 AM
Paul,

The main speakers in question were crossed over by a rebadged XTA unit with most of the settings factory locked. The other occasion I have had to measure a self powered and processed version of the same speaker had much the same results, within acceptable errors margins typical for memories over 12 months old. I have to believe that that is how the manufacutrer intended the boxes to be; but that is really beside the point.