MRC01
Major Contributor
Simply put, you could say that non-flat group delay means some frequencies are delayed more than others. But you can't say that without mentioning frequency!
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
What is group delay?
- Thread starter KTN46
- Start date
Simply put, you could say that non-flat (non-constant) group delay means non-linear phase vs frequency function, as only linear phase vs frequency function will result in a constant group delay.
Bottom line guys. Can you explain it for the end user customer? How does it affect my headphones? The BIG question, can I change it to do better? If I as the customer of Amazon can't change it, why do I care? In other words for the buyer of headphones in just a few sentences with NO MATH, no engineering speak, explain (like to a 10 year old) why do I as a headphone buyer need to know this? If I can't change it then it seems irrelevant. I "think" that is what the OP is asking. I'm only trying to get the answer out of the engineering realm into the customer buying headphones on Amazon realm. I know Amir can explain it well to a 10 year old so you guys can too. I have faith in you guys cause your all pretty smart! Looking forward to who gives the explanation using my parameters and wins the prize of best answer for the day. 
Bottom line guys. Can you explain it for the end user customer? How does it affect my headphones? The BIG question, can I change it to do better? If I as the customer of Amazon can't change it, why do I care? In other words for the buyer of headphones in just a few sentences with NO MATH, no engineering speak, explain (like to a 10 year old) why do I as a headphone buyer need to know this? If I can't change it then it seems irrelevant. I "think" that is what the OP is asking.
Ok, here it is: when considering speaker/headphone measruements non-linear group delay should be at the very bottom of your list as it pretty much doesn't affect perceived SQ. What you should consider is linear frequency response (also directivity with speakers) and distortion figures, but with far less weight than FR because what we measure distortion wyse is miles away from how we perceive distortion. So, it pretty much boils down to FR.
Please also note that with speakers, apart from directivity, women acceptance factor (WAF) can also come heavilly to play.
OMG! You might have won the prize for the best post!Ok, here it is: when considering speaker/headphone measruements non-linear group delay should be at the very bottom of your list as it pretty much doesn't affect perceived SQ. What you should consider is linear frequency response (also directivity with speakers) and distortion figures, but with far less weight than FR because what we measure distortion wyse is miles away from how we perceive distortion. So, it pretty much boils down to FR.
Please also note that with speakers, apart from directivity, women acceptance factor (WAF) can also come heavilly to play.
Thus , the casualness with which the term is used by some, belies its complexity and poorly understand nature?
While those who really know insist that it is of no consequence, especially in the lows, where so many purport that is the main difference between type of subwoofers, ie sealed , vs ported vs band pass , etc ?
Color me thoroughly confused
While those who really know insist that it is of no consequence, especially in the lows, where so many purport that is the main difference between type of subwoofers, ie sealed , vs ported vs band pass , etc ?
Color me thoroughly confused
I think what everyone what you need to know as the end user/customer it is not an issue. You can't change it as the user. It is something that the engineering team in design will be looking at but is not an issue for buyers of headphones to think/worry about. You just buy what you buy based on many other parameters and don't worry about group delay as it just isn't that important. Price, FR, weight, looks, quality of bass and other issues will be very important, the group delay is for the very smart engineers in the world. Buy what you like and enjoy!
Last edited:
- Joined
- Jul 23, 2019
- Messages
- 517
- Likes
- 544
I must admit I am confused by this. Can you offer a more detailed explanation?
It is possible to design an equi-ripple linear-phase FIR digital bandpass filter using the Parks–McLellan filter design algorithm (see here), with a passband of say 200 Hz to 2000 Hz. The phase response of this filter is linear in the passband of the filter. As was pointed out earlier, group delay can be mathematically calculated from the derivative (slope) of the phase response curve as a function of frequency: Group delay = –δφ/δf.
Because the slope of the phase response of the FIR bandpass filter is a straight line (i.e., linear phase), the computed group delay will be a constant value in the filter's passband.
Now, if we construct a windowed 800 Hz toneburst signal, the main frequency components of that signal will lie largely within the bandpass filter's 200–2000 Hz passband. When we apply the FIR bandpass filter to that windowed 800 Hz toneburst signal, the output signal will look more or less identical to the input signal, albeit delayed by a constant amount of time (the time delay is equal to 1/2 the length of the FIR bandpass filter).
So how is it that the constant group delay in the passband of the FIR bandpass filter does not constitute a time delay?
- Joined
- Jul 23, 2019
- Messages
- 517
- Likes
- 544
Where do headphones come into it? The original question was: "What is group delay? I've done a lot of Googling and searching... And I still don't understand what it is or what it does!".
Having skimmed this thread, I can confidently say that nobody knows what it is and that what it does is confuse audiophiles.
Raindog123
Major Contributor
I do. I think.
But my upbringing does not allow me to repeat the same thing more than thrice. 
Holmz
Major Contributor
Bottom line guys. Can you explain it for the end user customer? How does it affect my headphones? The BIG question, can I change it to do better? If I as the customer of Amazon can't change it, why do I care? In other words for the buyer of headphones in just a few sentences with NO MATH, no engineering speak, explain (like to a 10 year old) why do I as a headphone buyer need to know this? If I can't change it then it seems irrelevant. I "think" that is what the OP is asking. I'm only trying to get the answer out of the engineering realm into the customer buying headphones on Amazon realm. I know Amir can explain it well to a 10 year old so you guys can too. I have faith in you guys cause your all pretty smart! Looking forward to who gives the explanation using my parameters and wins the prize of best answer for the day.
In amazon speak… users of this headphone also purchased FIR based DSPs.
Many people are confused with math and physics.
Those that understand group delay have a general correlation with understanding what a DSP does, and they probably also do not buy kilo-buck power cords.
I do. I think.
ahh… so that s that the 1-2-3 part?
Last edited:
That means you're a nobody.
You are correct. For some reason I thought he mentioned them in a post later on. But, I'm too lazy to look now. Good catch!
Where the sound of one frequency arrives at your ears at a different time, to the other frequencies, even though on the original recording, they are in alignment. (ie: simultaneous)
In the bass / low end, it can cause or alleviate nul points in the room - and is used to overcome room generated issue (or can cause them)
In the high end, it can (potentially, theoretically?) contribute to or detract from imaging / space.
It is a form of distortion or non-linearity, in that the final outcome is different from the input...
Then we get involved in the intricacies of whether this is audible, and whether as a consequence it is worth spending time and effort trying to correct it.
My default position is, first do no harm... so try to keep the original signal and the original relationship between amplitude, frequency, and time intact.... - BUT lots of things act as filters / equalisers and change those relationships - sometimes, they can be corrected with intentional EQ (which includes things like crossovers, as well as equalisers, and digital Room or Speaker EQ systems)... But trying to correct some of the more complex group delay issues, can cause even more issues.
Raindog123
Major Contributor
Holmz
Major Contributor
It is normal to show it as ”Time versus Freq.”
@Habu I got one of those patches shown in your avatar when I was a kid.
Because high order systems, like 6th order sub enclosures, have a high group delay.
^This^ Is only true if we want to ignore the time domain.
Since the 70s people have had graphic equalisers with the sliders to alter the frequency response.
There are a whole other set of equalisers that affect the time domain and the impulse response.
And there are a whole set of speakers with good time and phase performance. And a whole bunch of people that say that it doesn’t matter.
If one wants the signal coming out of the speaker to look like the sound pressure measured from the original microphone… then time and phase and amplitude all need to be considered,
One can design something and trade off group delay for cost or something else. And someone else can use an EQ to correct for it, or ameliorate it. Just they cannot do it with a PEQ.
Most simple natural non linearities are minimum phase... including crossovers, RIAA EQ, speakers, phono cartridge cantilever and electrical resonances, and many others.
These can be corrected by simple PEQ - the simplest filters / EQ circuits are all natural Minimum phase - and because of the simple reciprocal nature of the time/amplitude/frequency relationship, if you use one of these standard filters (or old style graphic equalisers) to correct/linearlise the frequency response, the end result will return time, amplitude and frequency to their correct original alignment.
Sometimes simple is the right thing!
But some things are mixed phase.... and then it gets really complicated - and there is no simple way to correct it.
Phono cartridges are a good example of a relatively simple 99% minimum phase environment - the frequency response of a typical MM cartridge, is driven at the high end by a combination of the filter created by the cartridge inductance and the phono stage (and cable) capacitance and resistance - the result is a minimum phase filter.
All cartridges have a cantilever, and a suspended cantilever will have a natural resonance - depending on its effective mass this will be somewhere in the 8khz to 20khz range (some exotica have it even higher) - this will cause a resonant boost around it - which cartridge designers use to balance the natural drop off caused by the previously discussed electrical filter... - thereby achieving nice flat and extended frequency response by balancing several different minimum phase systems...
If Neutrality is wanted with a phono cartridge (a very rare thing!) - simply using a test record, and equalising with minimum phase filters, will achieve the goal without messing in any way with phase - because the other properties we are trying to correct for are all also minimum phase.
(Note: the one exception that I am aware of in the phono cartridge is the cantilever twisting influence on the sound, which is mixed phase - on most quality cantilevers however, this effect is very very small by comparison to the influences discussed above... and cannot be corrected for by filtering - only improved engineering of the cantilever system can help)
Holmz
Major Contributor
Thanks @dlaloum I was thinking more of passive-XOs and higher order slopes and phase shifts.
One can do the slopes in DSP with a symmetrical FIR and incurring no phase shifts.
Or maybe restore a correct phase in speaker that has altered the phase from its passive-XO.
And naturally one a HT one could use a big 6th order sub and remove the group delay with a FIR DSP, but it will not have the transcient response of an IB… but it will be better than it was.
One can do the slopes in DSP with a symmetrical FIR and incurring no phase shifts.
Or maybe restore a correct phase in speaker that has altered the phase from its passive-XO.
And naturally one a HT one could use a big 6th order sub and remove the group delay with a FIR DSP, but it will not have the transcient response of an IB… but it will be better than it was.
When I was experimenting a few years back, with improving phono cartridge performance - I started out with FIR linear phase filters - because they don't mess with phase...
But then as I got to know more about phase relationships - I worked out that the things that were altering the frequency response - were ALL minimum phase - and therefore if I used minimum phase filters to correct the frequency response, due to the reciprocal relationships, I would be correcting phase as well as frequency.... whereas if I used FIR linear phase filters (as I initially did), I would be correcting the frequency response, but leaving the altered phase untouched.
The other moving coil / moving magnet device that has identical properties to phono cartridge is of course speakers....
A Speaker/Driver, and it's crossover (non digital!), are all minimum phase - if you EQ the driver using minimum phase (typically IIR) filters, then you will correct both frequency and time/phase. - Same thing with headphones of course (as they are a simple speaker!)
I believe that once you have multiple drivers, and various other effects of a complete loudspeaker, some of the resulting interactions are no longer minimum phase.... I have not delved into that area, and therefore am not in a position to comment on that, other than to say - I don't think mixed phase behaviour is fully correctable - you would need a way to analyse it, and be able to mathematically break it down and reverse the effects....
