2-way vs 3-way speakers

[fluid]

how can you time reverse the impulse in an all analog design?

You can't really but that was not the point. You wanted to know when crossover phase change/group delay became a problem. If you take the impulse response and listen to it forwards or backwards and can't hear a difference then any phase or group delay change is inaudible. In the study they were able to equate the level of group delay at different frequencies to whether it could be heard very often, never or sometimes. If you stick to the threshold of inaudible then group delay will not be a problem. A time reversed impulse is a very hard bar to pass, fortunately it is not that hard to stay under that limit with many designs.

 

[dasdoing]

You can't really but that was not the point. You wanted to know when crossover phase change/group delay became a problem. If you take the impulse response and listen to it forwards or backwards and can't hear a difference then any phase or group delay change is inaudible. In the study they were able to equate the level of group delay at different frequencies to whether it could be heard very often, never or sometimes. If you stick to the threshold of inaudible then group delay will not be a problem. A time reversed impulse is a very hard bar to pass, fortunately it is not that hard to stay under that limit with many designs.

judging the graph you posted you kind of misenterpet it here. 10ms in the bass region is easily audible:

 

[Chromatischism]

If you add a sub to a 2-way speaker can it become as good or better than a 3-way speaker?

Huh. I guess I have 3-ways then.

2x15" Rythmiks: 0-80
SB Acoustics 6.5" : 80-1800
SB Acoustics 19mm: 1800-40000 (yes I know, it's silly extension)

 

[Newman]

I assumed the thread was about 2-way (+ sub if desired) vs 3-way (+ sub if desired).

 

[Pearljam5000]

I assumed the thread was about 2-way (+ sub if desired) vs 3-way (+ sub if desired).

It is

 

[mhardy6647]

 

[headshake]

I've got a paper from this year about group delay.

The listening test contained four different signals: the unit impulse, the pink impulse, a castanet recording, and a synthetic hi-hat cymbal sound. The two former signals are known to be the most critical for group-delay audibility, whereas the latter two resemble real-life musical signals and thus help to generalize the results better. Group-delay audibility was tested at the frequencies 500 Hz, 1 kHz, 2 kHz, 3 kHz, and 4 kHz. The results indicate that the audibility thresholds for local group-delay variation are less than ±1 ms for the most critical signals, and approximately 1.5 ms to 4.5 ms for a local positive group-delay peak and between −1.0 ms and −2.3 ms for a local negative group-delay peak for real-life signals.

Audibility of Group-Delay Equalization​

Audibility of Group-Delay Equalization​

This paper discusses the audibility of group-delay variations. Previous research has found limits of audibility as a function of frequency for different test signals, but extracting the tolerance for group delay to help audio reproduction system designers is hard. This study considers four...
ieeexplore.ieee.org

 

[fluid]

judging the graph you posted you kind of misenterpet it here. 10ms in the bass region is easily audible

Did you read the paper before deciding that I misinterpreted it based on a youtube video?

I will quote the authors instead of paraphrasing their findings.

"Figure 11(a) contains the group-delay variations of the signals that were mostly judged “same”, when the signals were played back forwards and backwards. It is seen that the group delay can exceed 10 ms below 200 Hz when compared to the group delay at high frequencies without the difference being audible."

The rapid rise in the low frequency group delay is due to the natural rolloff of the cabinet in the speaker or simulated speaker. The simple solution to this is to use a speaker or subwoofer that does not rolloff until below 20Hz or at least lower than shown in the research graphs.

 

[dasdoing]

Did you read the paper before deciding that I misinterpreted it based on a youtube video?

as said: I judged by the graph

It is seen that the group delay can exceed 10 ms below 200 Hz when compared to the group delay at high frequencies without the difference being audible

than they used wrong material. audiophile music for example is not "punchy" enough.
have you seen the video? the group delay is around 10ms and is easily audible. now you can trust a paper more than your ears. I prefer my ears

 

[Newman]

Big mistake

 

[sarumbear]

Not silly just unusual, mine have no crossover and cover 20Hz to 20KHz

And if you make a nice cabinet that considers diffraction you can get a pretty good horizontal directivity too (then stack 25 on top of each other to give you enough volume displacement)

How are you going to handle the loping that will occur as multiple non-coincident sources will be generating sound?

 

[fluid]

than they used wrong material. audiophile music for example is not "punchy" enough.
have you seen the video? the group delay is around 10ms and is easily audible. now you can trust a paper more than your ears. I prefer my ears

I have no issue with you trusting your ears and I like to test things for myself to see if I can hear them. That gives me a sample of 1 which is unlikely to convince anyone else of the validity of my experiment.

The video you posted has no explanation that I can see and is using a synthetic kickdrum where something is changed. I have no idea what, so I can't take it at face value without some further confirmation of what is happening. If you have more information about it that might help.

Reading the paper would give you the opportunity to see what they did and how they did it.

There was no music involved as explained in the paper
"The test signals for the conducted listening test were loudspeaker impulse responses obtained either from loudspeaker models or measurements".
"The loudspeaker impulse responses themselves sound like clicks. Since their spectrum is approximately flat, apart from the highpass cutoff at low frequencies, they tend to sound very bright. As the phase differences appear mainly at low and mid frequencies, one idea is to filter the impulse response to attenuate the high frequency content and to emphasize the lower frequencies. There is a similarity here to the relation between white and pink noise, as white noise sounds excessively bright whereas pink noise sounds approximately flat, since it contains a constant amount of sound energy per octave".

This research was carried out by Genelec's head of R&D, a post doctoral researcher in Signal Processing and Acoustics and a Professor in Acoustics, so you will have to forgive me if I find their words worthy of reading and sharing.

 

[fluid]

How are you going to handle the loping that will occur as multiple non-coincident sources will be generating sound?

Comb filtering occurs beyond the centre to centre distance. Because the drivers are quite small this occurs fairly high in frequency. I use FIR filters to do some "rearranging" of the impulse response at the listening position. When listening at 2m or more in room the comb filtering is not audible. You can sit on the ground, or jump up and down and the tonality does not change very much. The high vertical directivity all but eliminates floor and ceiling reflections and because of the line of drivers the vertical window moves up and down as you always have a driver at virtually the same height as your ears.

 

[Chromatischism]

The high vertical directivity all but eliminates floor and ceiling reflections

Except, unfortunately, the floor and ceiling modal bounce.

 

[fluid]

Except, unfortunately, the floor and ceiling modal bounce.

I'm not quite sure what you mean by modal bounce. The first order floor and ceiling reflections are absent apart from a 5K splash on the ceiling where there is loss of vertical directivity. They behave somewhat between a finite line source and an infinite line source due to the closeness of them to the floor and ceiling planes acting as mirror sources to some extent.

I'm not sure we should go into this too deeply here as my comments were only meant to show that there are other more unusual alternatives that not everyone is aware of.

 

[dasdoing]

The video you posted has no explanation that I can see and is using a synthetic kickdrum where something is changed. I have no idea what, so I can't take it at face value without some further confirmation of what is happening. If you have more information about it that might help.

you are right. In the video I should have explained more. but I did in the original post which I made the video for.
It's an all-pass filter. It changes nothing but the phase.
This is the group delay:

BTW:

Big mistake

I totaly get the concept of not trusting ears in ABX situations, but when there is an obvious diference not trusting your ears is kind of strange

Reading the paper would give you the opportunity to see what they did and how they did it

I can't read it without paying, right?

"The loudspeaker impulse responses themselves sound like clicks. Since their spectrum is approximately flat, apart from the highpass cutoff at low frequencies, they tend to sound very bright. As the phase differences appear mainly at low and mid frequencies, one idea is to filter the impulse response to attenuate the high frequency content and to emphasize the lower frequencies. There is a similarity here to the relation between white and pink noise, as white noise sounds excessively bright whereas pink noise sounds approximately flat, since it contains a constant amount of sound energy per octave"

so they simply played the impulses? that is kind of strange.
Never realy heard the bass with a dirac impulse for example. Never tried a pink impulse though

 

[fluid]

you are right. In the video I should have explained more. but I did in the original post which I made the video for.
It's an all-pass filter. It changes nothing but the phase.
This is the group delay:

If you look carefully your filter exceeds their threshold of audibility in the test by 300Hz so it is not necessarily the larger amount of group delay at 100Hz that you are hearing.

I can't read it without paying, right?

No open access I have attached the pdf, no excuse not to read it now

so they simply played the impulses? that is kind of strange.
Never realy heard the bass with a dirac impulse for example. Never tried a pink impulse though

It's different but they put some controls in to make sure the answers were accurate.

 

[Newman]

… “obvious”… you mean like A is obviously better than B sighted, and obviously worse in controlled conditions?

 

[abdo123]

I'm not quite sure what you mean by modal bounce.

i think he just meant Z-axis mode, and the tangential modes involving the Z-axis. Which will be there due to the prescence of the sound in a room, regardless of the absolute amount of output/directivity.

 

[abdo123]

If you look carefully your filter exceeds their threshold of audibility in the test by 300Hz

Which begs the question, how does the Neumann KH310 is getting away with analog 24db/oct filters at 650Hz and 2000Hz?