graaf said:
Any practical minded person would try to solve problems in the order of their significance/magnitude: biggest problems first.
If you compare room interaction with driver resonances visible in a CSD and amplifier generated distortions: For a loudspeaker driven in it´s intended passband there is at least an order of magnitude between each of those effects.
I for myself would not care too much for third order faults in audio if the first order ones haven´t been tackled properly.
Rudolf said:
To this I would whole heartedly agree, but how do you scale the subjective importance of these items. To me the loudspeaker and its room interaction are inseperable parts of the same problem. One is not more important than the other - they are tightly coupled together into a single perception Now amplifier sound quality - thats easily made insiginificant.
The CSD can be important, but you have to look at more than one point. Clearly a compact impulse reponse in all directions is desirable, but only if its directivity is controlled.
Rudolf said:
who is arguing? certainly not me
they have been tackled properly
I invite You to my thread "Loudspeakers and room as a system"
perhaps You might find it interesting
Rudolf said:
I am not convinced, all those problems in practice can bring equally annoying effects detrimental to the sound quality
BTW I don't believe in "driving a loudspeaker in it´s intended passband" - trying to do this is IMHO a source of major first order faults in audio
but this is of course another disscusion
best regards,
graaf
graaf said:
This is a nearfield CSD plot of a subwoofer and the same subwoofer measured at some farfield listening position in a room:
An externally hosted image should be here but it was not working when we last tested it.
To me it is clearly visible where to start with any otimization.
So I don´t see how we both could come to a mutual understanding of priorities in audio.
Rudolf said:
and the ear hears what?
Do You believe that our hearing works like the microphone and measuring equipment?
I am not talking about subjective hearing vs objective measurement
I am not talking about OBJECTIVE physiology of human hearing vs objective measurement
hearing mechanism is unlike electronic measuring equipment
Rudolf said:
I repeat - I am not questioning this - I agree with You - room-speaker interface is most important
Rudolf said:
So I don´t see how we both could come to a mutual understanding of priorities in audio.
to an agreement perhaps but an understanding?! why not?
let's talk!
best,
graaf
Rudolf said:
Both the CSD and room response are important, but have different audible effects when you improve each. The cleaner the CSD above 1KHz, the more realistic percussion instruments are. If you have a clean room decay, you will be able to hear more of the recorded room decay rather than your own room.gedlee said:
what about the kind of time distortion called by Goldmund "group delay distortion":
http://www.goldmund.com/technology/leonardo/
supposedly introduced by crossover filters and multiway loudspeakers in general
is it marketing pseudotechnobabble?
as is this: www.manger-audio.co.uk/PDFs/acoustical_reality.pdf (Manger got Diesel Medal not for pseudotechnobabble?)
and this: www.celticaudio.co.uk/articles/science.pdf (search Amazon for Watkinson's books)
is it all pseudotechnobabble?
best,
graaf
http://www.goldmund.com/technology/leonardo/
supposedly introduced by crossover filters and multiway loudspeakers in general
is it marketing pseudotechnobabble?
as is this: www.manger-audio.co.uk/PDFs/acoustical_reality.pdf (Manger got Diesel Medal not for pseudotechnobabble?)
and this: www.celticaudio.co.uk/articles/science.pdf (search Amazon for Watkinson's books)
is it all pseudotechnobabble?
best,
graaf
graaf said:
That is NOT marketing pseudotechnobabble. Group delay is a standard, well-defined term describing the phase response of a linear time-invariant system. It is the negative of the slope of the phase curve (definition) and the group delay at a given frequency tells you the input to output delay of the envelope of a wave packet at that frequency. Group delay distortion is just phase distortion viewed from this particular angle.
As far as I can tell, the way group delay distortion is percieved is not very well understood. Some work has been done to establish thresholds but I've never even seen something that tries to tie the shape of a group delay curve to perception.
A little while back I tried to determine if this could possibly be a factor in a "typical" situation, say an LR4 or LR8 crossover at some reasonable frequency. To do this I examined the ideal case: I used a computer program (Matlab) to process music and then A/B the processed/unprocessed over a good set of headphones. This way nothing except the phase response changes. I was never able to tell a difference with up to LR16 crossovers.
Of course at some point, you can screw things up so bad that it will be audible. Matlab has a nifty built-in function, iirgrpdelay, that creates a filter with an arbitrary (up to some limits) group delay. What I found was that this type of distortion is inaudible with anything (crossover-wise) I'd ever want to do and when it does become audible, it's not particularly offensive unless you very carefully select program material or you do something really stupid.
That investigation was primarily concerned with the group delay of crossovers and effects due to the springs and masses speakers are made out of. I make no claims about the spatial dependence of group delay distortion or group delay distortion from HOM in waveguides. I haven't studied the former and I don't even know how the latter behaves physically.
Graaf, I'll admit to being pretty fussy about the use of words, and the label pedant could well apply. Firstly, group delay errors and the need for correction of them is something that has been known about, and written about for a very long time. It was recognised in long distance telephony, and carrier telephony in particular ( feeding a number of channels down a pair of wires, otherwise known as multiplexing) To describe it as a distortion is a little confusing as it is a linear distortion, meaning that it does not change with signal level. It is generated in all analog filters, so crossover filters cannot escape it. For any system to be able to reproduce square waves the amplitude/freq reponse AND group delay curves have to be flat. The audibility of group delay and the need for "transient perfect" speakers is a much debated subject.
As I own a pair of Manger zerobox 109 speakers I guess I have experienced transient perfection? above the 140 Hz XO frequency, as has anyone listening to headphones. They demonstrate the folly of concentrating all the design effort into one aspect. They cannot compete with speakers that have directivity brought under control such as dipoles using conventional drivers.
There is no technobabble about anything John Watkinson writes on. He is a Fellow of the AES and the author of several highly regarded books on digital audio and television.
Sorry to be bursting in not having read the context that may have given rise to your interest in group delay.
Keith
As I own a pair of Manger zerobox 109 speakers I guess I have experienced transient perfection? above the 140 Hz XO frequency, as has anyone listening to headphones. They demonstrate the folly of concentrating all the design effort into one aspect. They cannot compete with speakers that have directivity brought under control such as dipoles using conventional drivers.
There is no technobabble about anything John Watkinson writes on. He is a Fellow of the AES and the author of several highly regarded books on digital audio and television.
Sorry to be bursting in not having read the context that may have given rise to your interest in group delay.
Keith
Group delay is much discussed. To get a feeling for it one should read the paper by Brian Moore some years back. He concludes that it is not audible "in most situations", but acknowledges that it would be in others. It is very instructive to sort out in what circumatances it would be audible according to his paper.
Lidia and I did a paper on the audibility of group delayed types of signals as would be present with diffraction (this is quite different than GD in a crossover). We found, as did Moore, that the audibilty was SPL dependent - more audible at higher SPL. Hence one must be careful in these discussions and when reading the literature to determine the SPL levels under consideration. Seldom are they defined or controlled.
I use GD to great advantage in my designs where the woofer and waveguide signals are time aligned by the GD of the LP crossover filter. GD need NOT be a bad thing, it can be quite useful.
The whole topic is extremely important, but also extremely complex with little relavent data to support much or anything.
Lidia and I did a paper on the audibility of group delayed types of signals as would be present with diffraction (this is quite different than GD in a crossover). We found, as did Moore, that the audibilty was SPL dependent - more audible at higher SPL. Hence one must be careful in these discussions and when reading the literature to determine the SPL levels under consideration. Seldom are they defined or controlled.
I use GD to great advantage in my designs where the woofer and waveguide signals are time aligned by the GD of the LP crossover filter. GD need NOT be a bad thing, it can be quite useful.
The whole topic is extremely important, but also extremely complex with little relavent data to support much or anything.
As Gedlee said, group delay distortion, and crossover GD distortion is generally not audible but, with special signal such as triple tone (cf Zwicker, Psychoacoustics), it is audible. You can to try yourself : http://www.ohl.to/about-audio/audio-softwares/phase-audibility/
Compared to other distortions, I suppose that it may generally be neglected.
gedlee said:
"negligable to inaudible" with every signal under any conditions?
were any tests conducted with musical signal under normal listening conditions?
so is it really what respected "AES Fellow" Watkinson write on the subject a pseudotechnobabble? or Manger - decorated by German authorities with prestigious Rudolf Diesel Medal?
that
can we simply disregard what they say because "it is negligable to inaudible"?
gedlee said:
well, You know the paper - instruct us please
best regards,
graaf
Dr. Geddes,
I noted in the thread "Setting up the Nathan", Post #33 that Markus posted the nearfield FR of the bass/mid-range driver. The response showed three dip/peaks located at 290Hz, 520Hz and 700Hz.
I assume these are due to standing waves in the enclosure and that without the foam damping material, they would appear much greater in amplitude. In your research on distortion perception have you established how audible these standing waves are ?
and have you established a minimum level which renders them inaudible ?
Regards
Peter
I noted in the thread "Setting up the Nathan", Post #33 that Markus posted the nearfield FR of the bass/mid-range driver. The response showed three dip/peaks located at 290Hz, 520Hz and 700Hz.
I assume these are due to standing waves in the enclosure and that without the foam damping material, they would appear much greater in amplitude. In your research on distortion perception have you established how audible these standing waves are ?
and have you established a minimum level which renders them inaudible ?
Regards
Peter
Our distortion studies were on nonlinear distortion not linear distortion. This is an entirely different thing.
I would take exception to your assumption of the causes of the peaks and dips. I believe that the dip just under 300 Hz is due to the spider. It resonates and draws energy away from the cone causing a dip.
I suspect that the other two higher up are likewise in the driver itself. I seriously doubt that they are in the cabinet and I am pretty sure its not internal standing waves. The upper two are barely blips in the far field frequency response and don;t worry me at all. The one at 280 Hz is more of a concern.
Acording to Toole, who has done the best work in this area, the dip at 280 Hz would be just audible, but not a serious problem. Virtually every driver that I have tested has this type of spider resonance. I have tried to get the manufacturers to do something about it, but to no avail. I'm just not a big enough buyer of the products to get their attention and the other users haven't complained (or so I have to presume).
I would take exception to your assumption of the causes of the peaks and dips. I believe that the dip just under 300 Hz is due to the spider. It resonates and draws energy away from the cone causing a dip.
I suspect that the other two higher up are likewise in the driver itself. I seriously doubt that they are in the cabinet and I am pretty sure its not internal standing waves. The upper two are barely blips in the far field frequency response and don;t worry me at all. The one at 280 Hz is more of a concern.
Acording to Toole, who has done the best work in this area, the dip at 280 Hz would be just audible, but not a serious problem. Virtually every driver that I have tested has this type of spider resonance. I have tried to get the manufacturers to do something about it, but to no avail. I'm just not a big enough buyer of the products to get their attention and the other users haven't complained (or so I have to presume).
I look at it from the point-of-view that a speaker which cannot reproduce the input waveform is simply not able to accurately reproduce what it is fed with.
OTOH I have also heard some very nice sounding speakers that are far from transient-perfect.
In the end of the day the total behaviour of a speaker is relevant and not only its behaviour in a specific domain. Unfortunately optimising the temporal response usually means worsening other things like FR linearity, non-linear distortion and directivity.
Regards
Charles
P.S. I have been able to listen to the Goldmund Leonardo system and some of John Watkinsons constructions.
OTOH I have also heard some very nice sounding speakers that are far from transient-perfect.
In the end of the day the total behaviour of a speaker is relevant and not only its behaviour in a specific domain. Unfortunately optimising the temporal response usually means worsening other things like FR linearity, non-linear distortion and directivity.
Regards
Charles
P.S. I have been able to listen to the Goldmund Leonardo system and some of John Watkinsons constructions.
phase_accurate said:
You are making a huge assumption here that the ear hears "waveforms". It doesn't. It deconstructs the waveform into a pattern of excitations in the ear (along the Cochlea) which are detected in complex ways. My point is simply that this "point-of-view" considers waveform reconstruction as "necessary" and its not - we don't hear that way and fitting the sound system to the way the ear hears is more effective.
phase_accurate said:
Would it surpise you that speakers will generate something that looks totally different from what it is fed if you compared the waves time aligned in real time? There is probably no speaker than can accuratly reproduce what it is fed.
This considers the imagination of the mind and very subjective. I think his original intent was to just point out what is accurate.gedlee said: