Still beating that dead horse ...
Let's say I show you two (2) time-domain impulse responses :
1. The impulse response of a first-order, LOW-pass filter ... with single pole (-3dB) corresponding to 4kHz.
2. The impulse response of a first-order, ALL-pass filter ... with pole/zero pair corresponding to 1kHz.
The first thing you may notice, is that the impulse response "tail" for #1 is SHORTER. It doesn't look like an impulse, mind you, but it "arguably" looks more like an impulse than #2, because of the shorter tail.
Which one sounds better? Which one has higher "fidelity"?
If your answer is : i can't really be sure, until i take a Fourier Transform, and examine the magnitude response component ... then we have a winner
(THIS is a really good post, i think ... in my humble opinion)
Sure, I would expect the all-pass to sound better....Why the need to look at the Fourier transform? All-pass shouldn't effect mag, just phase, right?
Isn't the running contention phase doesn't much matter ?
I guess the real reason I was equating linear phase to flat magnitude, and to excellent sound, is first, the best speakers I've heard have had very nice magnitude response, and also very smooth phase traces usually with gentle downward slopes.
Second, my own speaker development/tuning efforts continue to show me when I fix magnitude variations, I also fix phase (for minimum-phase variations). When I add in linear phase crossovers to the corrected drivers, I continue to hear improved transient response and clarity ....all the while seeing both mag and phase flatten up, and impulse get cleaner.
So I've come to the fix one, fix all belief...
Again, thx for the all pass example and your patience
Last edited:
I'm sure that happens sometimes, but it's not what I was talking about which was phase nonlinearity in the presence of linear frequency response.
No disagreement that frequency response matters more. My only point is that phase may or may not be negligible, it depends.
For example, in music as a harmonically rich glissando passes through significant phase shifting as frequency changes, phase relationships between the partials shift around unnaturally. Similar for pitch sequences adhering to rules of voice leading (perceived as one melodic line). It has a sound somewhat reminiscent of a wha wha pedal. Not good at all for some sounds. It can sound harsh, IMHO.
Last edited:
True, but being sensitive to group delay is not a cause. It's explaining one observation in terms of another. Why is the ear sensitive to group delay?
Group delay corresponds to time delay (although the two are not necessarily inter-changeable). So the answer may be more simple than you think ... and you may have already supplied the clues. If certain frequencies ... or harmonic overtones ... are delayed in time ... compared to a fundamental, for example ... it may become apparent to the ear (somewhat) that there's some "time smearing" at play. Phase and group-delay are intimately related by a simple first derivative, of course ... but it may be that the 'time-delay' connection to 'group-delay' means that we should look at group-delay, more so than pure phase, to determine sensitivities in hearing.
I am not sure that there is a good reason "why" the ear is sensitive to group delay, perhaps it's what you said before the harmonics getting messed up relative to the fundamental. Griesinger says that it is the harmonics that make up what we sense more than the fundamental so group delay of these matters. The fact is that we know that the ear is sensitive to group delay and it appears to be more sensitive to it as the SPL rises. Moore made a point of this in his paper on the audibility of group delay. It can be masked by many other aberrations, but as the SPL goes up it invariably will get unmasked and become audible.
that's a neat parallel and may be a not bad way to convey perceived sonic anomalies!
i like the wahwah pedal analogy where a notch sweeps with changing frequency or better yet an old Aphex Aural exciter and anyone who's played around with those can emulate the signature i've heard from badly tuned arrays! phase by itself is benign but when it starts "sweeping" it's audible!
i like the wahwah pedal analogy where a notch sweeps with changing frequency or better yet an old Aphex Aural exciter and anyone who's played around with those can emulate the signature i've heard from badly tuned arrays! phase by itself is benign but when it starts "sweeping" it's audible!
That makes total sense to me.
I think alot of the reason studies have had a hard time determining if phase matters is because we haven't had speaker systems tuned well enough to hear differences. I think that is beginning to change, largely due to FIR.
An analogy to me was with the beginning of digital photography.
As megapixel counts were rising, countless debates broke out about whether greater resolution was needed unless enlargement was the goal....because it was taken as a given that the human eye can only see 200 lines-per-inch resolution. This was in textbooks, based on air-force charts going way back.
Well, it was true, we could only see 200 lpi,....... but because freakin printers could only cleanly produce 200 lpi
Point is, I think our speakers have been in the same boat as the printers..
we just couldn't tune them any better.
Now, with ever better tuning techniques appearing, no telling what we will find we can hear from them.
I mean, in nature, if you whack a kick drum, all the frequency components, all the harmonics, arrive in time, in their particular phase. How can a speaker replicate that without linear phase?
In science, things are rarely black or white.That makes total sense to me.
I think alot of the reason studies have had a hard time determining if phase matters is because we haven't had speaker systems tuned well enough to hear differences. I think that is beginning to change, largely due to FIR.
An analogy to me was with the beginning of digital photography.
As megapixel counts were rising, countless debates broke out about whether greater resolution was needed unless enlargement was the goal....because it was taken as a given that the human eye can only see 200 lines-per-inch resolution. This was in textbooks, based on air-force charts going way back.
Well, it was true, we could only see 200 lpi,....... but because freakin printers could only cleanly produce 200 lpi
Point is, I think our speakers have been in the same boat as the printers..
we just couldn't tune them any better.
Now, with ever better tuning techniques appearing, no telling what we will find we can hear from them.
I mean, in nature, if you whack a kick drum, all the frequency components, all the harmonics, arrive in time, in their particular phase. How can a speaker replicate that without linear phase?
The notion that ... "it's linear phase, or it's WRONG!!" is much too simplistic to be useful, to engineers or hobbyists.
How SENSITIVE is our hearing, to phase deviations from linearity? Compared to ... how SENSITIVE is our hearing, to magnitude deviations from flat?
MOST COMMONLY, when two signals are added (or summed) ... at crossover, for example, when 2 drivers are playing the same frequency. The MAGNITUDE of the combined signal is STRONGLY dependent on BOTH the magnitude AND phase of the individual components. This is just simple vector math, really.
I just see linear-phase as a desirable, logical goal, .....nothing right or wrong about it IMHO.
Completely agree, ear is most sensitive to linear magnitude, and linear mag rules...I just think it has a kissing sweetheart named linear-phase
Completely agree, ear is most sensitive to linear magnitude, and linear mag rules...I just think it has a kissing sweetheart named linear-phase
As a speaker design goal or at the listening spot?
Regarding pitch perception as occuring in the ear rather that the brain, a topic we touched on earlier, here is some research regarding the brain's role: http://www.physiology.org/doi/pdf/10.1152/jn.00104.1999
A brief quote from the article:
"... we report psychophysical and anatomical
evidence in favor of the hypothesis that fine-grained frequency
resolution at the perceptual level relies on neuronal frequency selectivity
in auditory cortex."
A brief quote from the article:
"... we report psychophysical and anatomical
evidence in favor of the hypothesis that fine-grained frequency
resolution at the perceptual level relies on neuronal frequency selectivity
in auditory cortex."
Hi wesayso, at the listening spot / measuring spot / design spot or better yet, designed coverage area.
But that's kinda the same thing as speaker design, isn't it?
This is a good test, also of absolute phase of the membrane's direction and all the elements arriving at the same time. I'm not sure what you mean by "in their particular phase" though
If phase doesn't matter this one should sound perfect shouldn't it but case is rhythm falls total apart and foot really stops tapping, suggest try it yourself :
This is better than above but still rhythm is not right and foot tapping is not very often because one sense bass attacks comes a micro bit too early, but it works very well and sounds natural for the case where one listen from outside the room via a open door:
Think this one is much better than the two above in phase follow amplitude for system band-pass and FIR DSP help discuise where XO point finds place:
but case is rhythm falls total apart and foot really stops tapping, suggest try it yourself :This is better than above but still rhythm is not right and foot tapping is not very often because one sense bass attacks comes a micro bit too early, but it works very well and sounds natural for the case where one listen from outside the room via a open door
:Think this one is much better than the two above in phase follow amplitude for system band-pass and FIR DSP help discuise where XO point finds place:
Attachments
I guess what I meant, is picture the striker hitting the kick, the membrane moves, waves start originating, but don't further waves keep originating in all kinds of ways apart from the primal membrane movement ...? Each with their own continued mag and phase?
And aren't harmonics originating off that primal movement...? Not to mention the further movements......all with their own mag phase relationship to the primal movement?
And how about internal reflections bouncing out of the kick? And their harmonics? Again, all with their own mag/phase relationship back to primal
All these waves have both a mag and phase relationship with each other., don't they?
Sure we all agree on the importance of mag, but how can they possibly be accurately reproduced without phase also?
Linear phase.....a desirable goal I think.....
Last edited:
Fortunately, everyone agrees that phase mattersIf phase doesn't matter this one should sound perfect shouldn't it
Certainly, when 2 or more signals are added. Even, for a single speaker through a single driver. The question is, how sensitive is the ear, to deviations from phase linearity? How much of a deviation from flat amplitude would you tolerate, to improve phase linearity?- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.