Now that's another debate... audibility of group delay.
But it has nothing to do with our prior discussion. As said, you can EQ away any group delay with FIR filters if so inclined. A linear phase crossover works similarly. It has a minimum phase part and a phase edit. At the cost of a tiny delay. EQ your frequency flat for a 1.5 or 2 octave past where it drops off and apply a linear phase crossover and you've repaired all that group delay you're so worried about.
I actually have/use a phase linear system as I did worry about similar things.
I've heard it both ways and prefer the phase linear behavior (*). As I don't play live, the small delay after pressing play is of no concern to me.
For video the pictures are delayed as well to restore audio/video sync. I prefer it that way, but stopped defending it. I do describe it and how
I got there for those that are interested on my thread.
I think we all hear a bit different. You don't have to like what I prefer and vice versa.
(*) Actually, my phase is following a minimum phase curve for the bandwith my speaker system plays. I did not linearize the natural roll of on both ends.
But it has nothing to do with our prior discussion. As said, you can EQ away any group delay with FIR filters if so inclined. A linear phase crossover works similarly. It has a minimum phase part and a phase edit. At the cost of a tiny delay. EQ your frequency flat for a 1.5 or 2 octave past where it drops off and apply a linear phase crossover and you've repaired all that group delay you're so worried about.
I actually have/use a phase linear system as I did worry about similar things.
I've heard it both ways and prefer the phase linear behavior (*). As I don't play live, the small delay after pressing play is of no concern to me.
For video the pictures are delayed as well to restore audio/video sync. I prefer it that way, but stopped defending it. I do describe it and how
I got there for those that are interested on my thread.
I think we all hear a bit different. You don't have to like what I prefer and vice versa.
(*) Actually, my phase is following a minimum phase curve for the bandwith my speaker system plays. I did not linearize the natural roll of on both ends.
Last edited:
Well there's some truth to this. Personally, I wouldn't go much above 24 dB/octave in the resulting acoustic slopes, just because of the excess group delay being audible at some point (speaking of analog/IIR filters). But I guess that hardly any horn+driver will have such a high natural slope that this should really bother you. Maybe some high-loading horns that exhibit a sudden drop in the response are worse than average in the knee region, I don't know. It's certainly not an issue with an average CD waveguide.
Last edited:
The audibility of group delay on itself isn't relevant either.
Just the fact that that it still a "debate", so apparently not so obvious and apparent (read: audible) says enough.
On the list of priorities of things that can go wrong in a speaker system definitely not in the top 10.
One can still estimate the contribution of things with thought experiments and interpolation of certain problems.
They do this in biology and sometimes in physics all the time, but when it comes to audio, all these techniques seem to go right out the window and all of a sudden THD+N seems to be as important as acoustics (reverb, standing waves etc etc), as directivity, as audibility or group delay for some people.
It just doesn't make any sense at all.
Just the fact that that it still a "debate", so apparently not so obvious and apparent (read: audible) says enough.
On the list of priorities of things that can go wrong in a speaker system definitely not in the top 10.
One can still estimate the contribution of things with thought experiments and interpolation of certain problems.
They do this in biology and sometimes in physics all the time, but when it comes to audio, all these techniques seem to go right out the window and all of a sudden THD+N seems to be as important as acoustics (reverb, standing waves etc etc), as directivity, as audibility or group delay for some people.
It just doesn't make any sense at all.
If anything, group delay can only be audible at certain jumps (from low to high or something).
Otherwise it is just basically the total system delay.
So as long as you keep all filters withing ± 2nd order system max difference, the damage is extremely minimal to begin with.
Just saying something about a filter without knowing the rest of the context doesn't say anything. If one filter is 2nd order and the other is something like a 6th order or so, yeah maybe.
But also nothing more than a maybe.
In that case I really wonder why someone needs such higher order filters to begin with to be very honest. It's not really needed, unless you're using speakers/drivers/woofers that are less optimum (= lots of dips, peaks, resonances)
In that case there are other issues to be solved first, go back to the priority list.
Otherwise try a param EQ first
Last edited:
It was just a general statement plus a little nuance on your 24dB/oct statement.
You didn't avoid it by crossing over at 1.5 to 2x fb. As that was easily to solve anyway. I've mentioned it in a couple of posts how to do that.
Cross where your driver can still reach the desired SPL level, or even better, where the directivity matches the driver beneath it.
To do it without causing you group-delay-heartaches, EQ the both the drivers flat well beyond where you want to cross them and after that apply linear phase crossovers.
After that you can do the experiments where which driver still sounds best to you, playing the mid-range. Does the compression driver win or the big 15".
(or is it better still, to use a synergy type MEH horn with multiple small mid-ranges that fill in that spot)
Last edited:
Group delay anomalies can change the attack character of snares and bassdrums ( transient info right?)
In order to use a low order xo dont you still have to avoid fb?
Last edited:
In order to use low order crossovers, you have to look at the acoustical end result, not the electrical order by itself or the named slope from a DSP.
That includes the natural roll off of the driver unless you use EQ to shape the driver to the desired shape acoustically, while making sure the
driver can handle whatever you're feeding it.
That includes the natural roll off of the driver unless you use EQ to shape the driver to the desired shape acoustically, while making sure the
driver can handle whatever you're feeding it.
Go have a look what the order of magnitude is of those transients, and go have a look at the order of magnitude of the difference in jumps in group delay.
Than also take the integration time of the human ear in mind, and all these other things happening in a recording PLUS the background noise.
I mean, yeah you can spend all day figuring that out.
But experience tells me that most of the time the acoustics of the rooms is many MANY order of magnitude worse.
Not to mention how messed up the recording already is by itself.
But that is a whole other can of worms.
Switching to pathetic mode?
This is humbug, and you know it.
It also shows contempt for other forum members.
Or do you need a list of comments to jog your memory?
Last edited:
Don't really see the need to be snappy and snarky at each other?
Doesn't seem contribute to the discussion.
Dammit you beat me to it...I even had it typed out then deleted to chase another reply.....the last time we debated group delay, I think Dr. Geddes pretty much spelled out the same story, that is, the room is going to destroy much of the attempt at perfect gd/phase....
Still you can't shun me for trying
Which reminds me, looking at phase is not like looking at time to Peak energy… phase has to do with where the waves are in time… Time to peak energy shows where those waves reach max
amplitude, in time
Last edited:
I just felt it as appropriate and polite to thank for something I appreciate.
Ok so this is your real motivation, not that I am incorrect, rather that I will not benefit from my correctness. I am paying attention. Dr. G recently stated, where there is resonance there will be group delay. I present to you evidence B, the second attachment of the measurements I posted...burst decay....at the 1st mode...tuning note, you will see excess decay...this, is, bad, mmk.... how will you explain away that this excess decay is not worry worthy? We have time to peak moved out in excess at a mid frequency, combined with excess decay...unlike group delay, moving a hp filter up above fb will not take this excess decay with it, or maybe it will, Ive never has burst decay before or investigated this But! Lets say xo filters do create excess decay...placing it within already existing excess decay will just compound it....no less than group delay...and to wesaysos point...the xo doesnt have to get into the cutoff for there to be exacerbation of existing gd/decay...in particular if the xo is of high enough order.
There, did I make a point yet? How many excess cycles are too many cycles?
Camplo
There are many things that we know about perception, and many we don't. What you are talking about seems purely hypothetical to me (but admittedly I don't understand your position.) All of us, including myself, will have hypotheses as to what is important, but until we do the tests, have the data, it's all just conjecture. I try to not deal with conjecture, I want facts and reliable data to back up any position.
Let me give you a prime example. Throughout the first half of my audio career I believed that nonlinear distortion was the major reason that speakers sound different. I studied nonlinearity intensely and did some of the first tests of nonlinear audibility. I got the facts and the data. But alas, my view then has not been shown to be the case. It was, in fact, completely wrong.
We may think that XYZ has to be true because it makes so much sense, but until one has the facts and the data everything is a candidate for being completely wrong.
There are many things that we know about perception, and many we don't. What you are talking about seems purely hypothetical to me (but admittedly I don't understand your position.) All of us, including myself, will have hypotheses as to what is important, but until we do the tests, have the data, it's all just conjecture. I try to not deal with conjecture, I want facts and reliable data to back up any position.
Let me give you a prime example. Throughout the first half of my audio career I believed that nonlinear distortion was the major reason that speakers sound different. I studied nonlinearity intensely and did some of the first tests of nonlinear audibility. I got the facts and the data. But alas, my view then has not been shown to be the case. It was, in fact, completely wrong.
We may think that XYZ has to be true because it makes so much sense, but until one has the facts and the data everything is a candidate for being completely wrong.
And I think that it is appropriate for experts to raise flags when falsehoods are expressed. That's why I said that. You were right, he was wrong and people need to hear that.
But what does that even mean? If I have a flat minimum phase (MP) system then the peak energy excess delay will be the same at all frequencies.
What we don't know is if the system is non-MP, a flat system can have excess group delay, then at what level is this audible? We do know that at some level it will be (Brian Moore.) I don't know the answer and I doubt that anyone else in this thread does either. But remember that most "excellent" system designs are going to be very close to MP, meaning that, at best, this is a small influence, maybe worth doing if everything else is ideal, but otherwise not worth worrying about.
What we don't know is if the system is non-MP, a flat system can have excess group delay, then at what level is this audible? We do know that at some level it will be (Brian Moore.) I don't know the answer and I doubt that anyone else in this thread does either. But remember that most "excellent" system designs are going to be very close to MP, meaning that, at best, this is a small influence, maybe worth doing if everything else is ideal, but otherwise not worth worrying about.