I think maybe it is the fact that "uncorrelated" sounds bad. In signal processing language correlated means the signals are the same and uncorrelated means they are different with a continuum between them.
If the signals are the same they will add together in a different way like with two sub drivers placed within a 1/4 wavelength of each other producing the same signal. They will on the whole be highly correlated and you will get a 6dB addition.
The opposite sides of an acoustic crossover, not so correlated. This is how it makes sense to me anyway.
Thanks fluid,
Bill makes a well said statement.
And i think it kinda illustrates a problem our measurement programs are forced to handle.
That being, they are forced to generate continuous mag and phase curves from end to end, like it or not. Even when there can be regions of chaotic discontinuity that don't really fit with the curve.
(As perhaps from the synergy notch.)
I think the measurements don't really show the discontinuous breakups, other than maybe with really steep or reversing phase and mag anomalies.
This has inspired me to try to spend some time today with Smaart's coherence blanking threshold. It can be set to simply leave out the mag and phase regions where insufficient coherence exists. Will be interesting to see how it handles the notch.
Switching gears, any idea why Self says the outputs can be regarded as uncorrelated? It seems like it's the same signal just being sent through a voltage divider, albeit a divider that varies by freq.
Anyway, it's clear to see there is an electrical power dip going to the speaker.
But how does that translate into an acoustical power dip coming out of the speaker?
An acoustic power dip seems to ignore the efficiency gain provided by two sources as per my prior example (unless lack of correlation is at the heart of it).
I have yet to see what i think is credible research that isolates phase audibility as the only variable........across the spectrum on a full range 20-20k, high dynamic system, in listening conditions that would allow a good test (outdoors). Along with source material that is from a single mic recording of natural full spectrum, high dynamic, sound.
Anything less seems flawed, and just continues the debate, imho.
Well, mark100, that's some depressing post for researchers. Sort of like proving the null hypothesis.
Let me come at the issue sideways. When a speaker "introduces" HD, it is adding something that you hear. Likewise, when the tweeter is set loud, the treble sounds bright.
But how would you "hear" a driver that had a big drop between 1 and 1.2 kHz? Or if there were a phase anomaly in that region? To hear stuff like that with music playing, it is assumed you have some absolute judgment of loudness or a prior knowledge of what a music piece ought to sound like in that narrow band.
If someone compared two crossovers which differed a lot in phase behaviour but only a little in other aspects, and nobody could tell them apart tested blind, wouldn't that be convincing?
B.
Don't understand why you would make a distinction between group delay and phase.
Since group delay equals phase, just expressed in different terms than normal phase graphs using degrees rotation vs freq.
Nope Ben, not like proving the null hypothesis at all.
More like truly isolating the variable under test...the variable being phase.
Where the test is flat phase vs when does phase rotation become audible.
Not just one level of phase rotation vs another.
And yes, to answer your question. That would be convincing if the test were made, various phase rotations against flat linear phase head 20-20k.....in conditions like i previously mentioned.
The one condition I didn't mention, because it is implicit in being able to test against flat phase, is the actual ability to reproduce it.
Until relatively recently, that ability simply hasn't been feasible with the multi-way speakers necessary and capable of reproducing the full audible bandwidth with convincing dynamics and SPL.
This is the crux of why older research fails imo.
So, to answer yes again, ....flat mag and phase, comparing the same flat mag with nothing but phase rotation as the variable, would convince me.
But again, it would need to be under all the other conditions to be good science imo.
Gotcha.
And I agree, group delay is known to be generally accepted as audible.
Whereas phase is generally not accepted to be audible.
This always makes me laugh, and makes my eyes roll around....cause they are the same dang thing !!!!!!!!!
My experience while trying to align the phases between the elements in the crossover sections, in my case 400 Hz and 2.5k Hz, is that the "phase warping" introduced by the filters typically not is audible. What is audible is the "jump" in the phase when the elements not are well enough aligned in the crossover section where both elements can be heard.
Im my case I can easily hear a 11 microseconds phase difference, in both crossover sections. At 400 Hz it is detectable as an "unclear" or "double" tone when playing a 400 Hz sinus wave.
At 2500 Hz I find it more revealing to listen to live recordings with emphasis on information in the crossover region.
Even though the phase warping not is directly audible, I experience a clear difference between LR4 and LR2 filters, which I connect to different "warping characteristics" (both have pros and cons and are good at different aspects).
In this sense "phase in general" also seems to have a kind of audibility.
Last edited:
I have now tried both LR4 and LR2 (w. inverted mid), all other settings similar, and the results are interesting. Both filter types really bernefit from good phase alignment in the crossover sections, but LR4 is of course more sensitive since the steeper phase warping.
The filter types have different characteristics, and there is no clear winner.
LR4 has a wider sound stage and more life-like reproduction of voices and instruments. LR2 has a much better transient response and handles rythm better.
The Hypex plate amps can store three settings to switch between with a one button selection on the remote. This means that I don't have to choose, I can have both!
(If I really had to make a choice I think I eventually would end up with LR4.)
Hi Rob, I've tried that several times with Sony and Stax headphones.
Sometimes i think i can hear a difference, most of the time not.
In any case, if headphones were a definitive test, I'd quit with the idea phase is a big deal.
I think the reason headphones don't work, is they can not impart the sound and feel of strong bass and/or strong transients like a speaker can.
Besides the greater auditory impact of speakers, I feel bass notes and transients in my body, even at pretty modest levels.
When I get tuning to flat phase across the spectrum, I feel an increased sense of rhythm to the the music.
And ripping bass, along with slamming transients, are surely stronger and clearer. Sometimes i just start laughing with glee at the sound when crankin
Good stuff johanhaggstrom, don't you love the ability to quickly swap/try things with dsp presets !
Thank you.
I have not tried to midranges, but I would guess that it does not make things easier. And 6" midranges do require lower crossover and more careful layout.
What about the front baffle? How do you control dispersion in the crossover between tweeter and midranges?
Yes I suppose, like an open field is the same as the side of a cliff because they are both flat.
Before going über-sarcastic could someone who thinks that phase and group delay are not the same, please explain to me WHY they are not the same?
From what I understand the term "group delay" describes the delay a wave packet experiences by going through a system, a filter for example. This delay might be dependent on the frequency of the wave packet. in this case the system does not have a linear phase response.
So in my understanding there is a clear relationship between phase and group delay, one might even say they are two terms for the same concept.
If group delay is constant for all frequencies the system has a linear phase response. I phase response is not linear, group delay will vary with frequency.
Or did I completely mess things up?
From what I understand the term "group delay" describes the delay a wave packet experiences by going through a system, a filter for example. This delay might be dependent on the frequency of the wave packet. in this case the system does not have a linear phase response.
So in my understanding there is a clear relationship between phase and group delay, one might even say they are two terms for the same concept.
If group delay is constant for all frequencies the system has a linear phase response. I phase response is not linear, group delay will vary with frequency.
Or did I completely mess things up?
Hi Tfive,
Your question asking why phase and group delay are not the same thing demands a valid answer.
I say they are the same thing, because in essence they are.
Both are completely a function of phase, phase simply being itself
Group delay is not exactly phase: it is the negative first derivative of phase plotted on the Y-axis vs frequency on X-axis.
You may well know this ...pls don't be offended if i'm talking like a damn teacher.
But when the counter arguments are: it is all about context, or open field vs cliffs, or a tree vs a forest ...well, sigh....
Anyway, why i say they are the same thing, is because they are both 100% tied to phase and both frequency dependent with no
variable other than phase.
Group delay is nothing more than an expression of phase change against freq. It's all about phase, nothing else.
Your question asking why phase and group delay are not the same thing demands a valid answer.
I say they are the same thing, because in essence they are.
Both are completely a function of phase, phase simply being itself
Group delay is not exactly phase: it is the negative first derivative of phase plotted on the Y-axis vs frequency on X-axis.
You may well know this ...pls don't be offended if i'm talking like a damn teacher.
But when the counter arguments are: it is all about context, or open field vs cliffs, or a tree vs a forest ...well, sigh....
Anyway, why i say they are the same thing, is because they are both 100% tied to phase and both frequency dependent with no
variable other than phase.
Group delay is nothing more than an expression of phase change against freq. It's all about phase, nothing else.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.