• 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?

KTN46

Active Member
Joined
Aug 16, 2020
Messages
104
Likes
192
I've done a lot of Googling and searching... And I still don't understand what it is or what it does!

Can anyone ELI5 what it is, assuming minimal technical knowledge?
 
It's the delay from the signal being inputted to sound being outputted.

For speakers it's usually only subwoofers where it can be a concern (as the delay is not constant but increases as frequency decreases), but even then it has to be very bad to be audible by most.

EDIT: Ah, for headphones it’s different, a messy measured group delay is what you get from multiple sound sources interacting, which means reflections within the ear cup as well as resonances of the unit.
 
Last edited:
I've been seeing a lot of it pop up in headphone reviews, but even the most sporadic graphs still show values in the order of miliseconds. Is this enough to make an audible difference?

And also to clarify; is this graph:

index.php


Saying that bass frequencies take up to... 5 miliseconds longer to produce?

Thankyou for helping clarify.
 
I believe that Amir’s focus on group delay in headphones reviews is not to say the delay in the treble is audible, rather it is to show where the driver is not well behaved.

He is saying dont EQ away the notch there because it looks like the designers put that notch in to prevent a driver issue at that frequency. If you EQ it flat again you will get distortion or ringing.

Bass always has more group delay because a 1 kHz tone takes one millisecond while a 10 hz tone takes 100 so even when a quarter wavelength delay is a huge spike on the bass chart.
 
Bass always has more group delay because a 1 kHz tone takes one millisecond while a 10 hz tone takes 100 so even when a quarter wavelength delay is a huge spike on the bass chart.

That's not what group delay is. Group delay is literally just that. Delay. It's the absolute amount of time for a signal to pass through a process independent of frequency. When group delay is constant, i.e. digital delay, some digital filters, all pass analog filters in the pass band, then it is normal to discuss group delay as phase relationships input to output are maintained. When group delay is not constant it is more normal to discuss phase as the relative timing of frequencies change.
 
Wikipedia seems to have a perfectly accurate description of group and phase delay
 
It is difficult to grasp

2 signals of different frequency superimposed/combined, like a modulated signal or a multi tone like a square wave.

The 2 signals each have 2 components:
Magnitude (an envelope)
Phase, the phase relationship between the two signals

When the signal is put thru a device it 'distorts' it, typically the device is a filter.

If the input signal vs output signal magnitudes shift in time, group delay

If the input signal components are in phase (could be any phase angle) and the output phase relationship between the 2 signals has shifted, phase delay
1B46FC8D-9F6C-4C30-8A4D-95E965F7E195.png
30E45883-094B-4097-A092-CB70B0C55E18.png
 
also worth keeping in mind that in natural objects (including speakers/headphones/cartridges) - most (but not all) frequency/phase relationships are minimum phase...

Many (most) frequency response variations both electrical and mechanical have a reciprocal behaviour in the amplitude (size/level) and phase (timing) domaing - whereby by using the relevant mathematical formulae, a knowledge of the frequency response can be used to calculate the phase response and vice versa. - This form of phase variation is called "Minimum" phase.

This means that if you use a minimum phase filter (standard "analogue" filter behaviour) to correct frequency response non linearities, you will also correct phase response non linearities. Your end result will be fully corrected in time, amplitude, frequency.

In digital EQ however, we can apply "linear phase" filters - which adjust ONLY the frequency/amplitude and leave the time domain (phase) alone.

If what we are trying to achieve is corrective EQ to compensate for natural system non-linearities, using linear phase filters is a mistake, as you won't be correcting the phase (time) relationships.

Note: I said most things are minimum phase, there is reason to believe that some of the complex twisting behaviour of phono cantilever styli, are mixed phase - and partially behave as minimum as well as partially linear phase - luckily that type of impact on audio is very minor, and completely overwhelmed by other aspects of a cartridges behaviour, all of which are minimum phase.... and therefore correctable by minimum phase EQ.
 
That's not what group delay is. Group delay is literally just that. Delay. It's the absolute amount of time for a signal to pass through a process independent of frequency. When group delay is constant, i.e. digital delay, some digital filters, all pass analog filters in the pass band, then it is normal to discuss group delay as phase relationships input to output are maintained. When group delay is not constant it is more normal to discuss phase as the relative timing of frequencies change.

Ah no.
The delay through the equipment is just a ”time delay”.

Group delay is when the delay varies across frequency, so that the entire “group”, if you will, is arriving at different times relative to each other.
It is the derivative of phase (or maybe the intergral).

In a subwoofer it is manifested as the lowest notes usually arriving way later than the 50-60 Hz notes, particularly in a high order box.
Low order ”enclosures” like IB, have a good transient response and low group delay.

In fact transcient response, impulse response and group delay are all related.
Impulse response is a time domain thing… With the X-Axis being time, and Y-axis amplitude.
Groups delay is usually plotted as Time on the Y-Axis versus freq on the X-Axis
 
Ah no.
The delay through the equipment is just a ”time delay”.

Group delay is when the delay varies across frequency, so that the entire “group”, if you will, is arriving at different times relative to each other.
It is the derivative of phase (or maybe the intergral).

No. Simply no.

Group delay is time delay. Group delay is often described as the delay of the amplitude envelope ... which is a fancy word for time delay.
 
Still don’t get it . I apologize for being so dense
No need, neither do I
But then I'm not real bright.
(Hint Hint) John

I never had a problem with delay during group sex though. LOL
 
  • Like
Reactions: CMB
If what we are trying to achieve is corrective EQ to compensate for natural system non-linearities, using linear phase filters is a mistake, as you won't be correcting the phase (time) relationships.

I think one can apply a linear phase filter to correct a magnitude issue of a minimum phase system. This will also correct the phase, at the cost of adding a constant (group) delay.
 
Exactly as @Holmz said, the group delay is the [time] difference in propagation [across some medium, channel, or component] between the lowest and highest frequency of a signal of a non-zero bandwidth.

Yes, [in time domain], it can be measured as the difference between arrival of the wave ‘fronts‘ at those different frequencies, or the change in the shape of the corresponding ‘envelope’. It can be the time difference, or phase difference (assuming the two - min vs max - frequencies started with the zero-phase difference).

Again, the term “group [of sinusoids of different frequencies]” implies the signal of a certain bandwidth, and does not make sense for a “single [frequency] tone”. If the [propagation or processing] speeds of all the waves at all frequencies in a certain medium/system are exactly the same, the group delay there is zero. While a non-zero group delay leads to certain nonlinearities thus signal distortions.
 
Last edited:
Back
Top Bottom