Dr. Klaus Heinz of HEDD Audio (ex ADAM Audio) - measuring speakers, in particular speaker dynamics

[andreasmaaan]

I agree. But on the other hand, we probably all agree that once you make FR linear (and room will always make it non-linear down from Schroeder frequency unless you live in anechoic chamber) you will be able to hear the improvement. Contrary to all these timing-related stuff that really helps..

Completely agree, but IIUC, the speaker already achieves linear FR without the external phase linearisation DSP, so that problem is already "solved" prior to the correction.

 

[Daverz]

You think it is because of looking at discrete signal with analog glasses?

AKA scallop loss. I think you can smooth it out by zero-padding.

The scipy source has Python code for generating a log cosine sweep (or chirp):

https://github.com/scipy/scipy/blob/v1.3.0/scipy/signal/waveforms.py#L265-L430

 

[Krunok]

Completely agree, but IIUC, the speaker already achieves linear FR without the external phase linearisation DSP, so that problem is already "solved" prior to the correction.

I'm not sure I?m following.. Are you saying that phase shouldn't be linearised? When I was doing filters mannually with rePhase I first corrected FR and after that I made phase correction in 3 steps:

- I corrected the phase of the filters to set it as close to 0 as possible
- I corrected phase of the passive crossover by entering the data about it
- I corrected phase of left speaker to avoid cancellation of low frequency response when both speakers are playing

Are you saying some of those phase corrections made more bad than good?

 

[andreasmaaan]

I'm not sure I?m following.. Are you saying that phase shouldn't be linearised? When I was doing filters mannually with rePhase I first corrected FR and after that I made phase correction in 3 steps:

- I corrected the phase of the filters to set it as close to 0 as possible
- I corrected phase of the passive crossover by entering the data about it
- I corrected phase of left speaker to avoid cancellation of low frequency response when both speakers are playing

Are you saying some of those phase corrections made more bad than good?

No, I’m just saying we don’t know as much about it.

On one hand, there’s been relatively (for this field) extensive research into the audibility of the kind of group delay caused by typical minimum phase filters. The research is quite consistent in suggesting that this is innocuous in the doses typically administered by 2nd-4th order crossover filters.

But much less research has been done into the audibility of pre-ringing in crossover filters, so it’s more of an unknown. Some recent comments of JJ’s suggest there may be more to it than I’d assumed, but he hasn’t elaborated much and I can’t find whatever research he’s referring to.

In any case, whether audible or (more likely) not, and quite aside from it being unnecessarily complex and convoluted (excuse pun), the HEDD approach doesn’t seem to me to be technically optimal, as it produces the maximum ringing (including pre-ringing) vs the alternative.

 

[Krunok]

On one hand, there’s been relatively (for this field) extensive research into the audibility of the kind of group delay caused by typical minimum phase filters. The research is quite consistent in suggesting that this is innocuous in the doses typically administered by 2nd-4th order crossover filters.

But much less research has been done into the audibility of pre-ringing in crossover filters, so it’s more of an unknown. Some recent comments of JJ’s suggest there may be more to it than I’d assumed, but he hasn’t elaborated much and I can’t find whatever research he’s referring to.

The filters you are mentioning here would be "normal" LR passive filters or DSP filters?

In any case, whether audible or (more likely) not, and quite aside from it being unnecessarily complex and convoluted (excuse pun), the HEDD approach doesn’t seem to me to be technically optimal, as it produces the maximum ringing (including pre-ringing) vs the alternative.

Excuse my ignorance but I didn't quite catch up what is the HEDD approach and what is the alternative, so can you please elaborate?

 

[Krunok]

Speaking of GD, here it is how it looks with my right speaker. As with everything else timing related refelction completely screw it - blue one is taken from LP (4m from the speaker) and green one is pseudoanechoic (30cm).

As I explained, my situation is quite different than Mictho's - not only one my woofers in each speaker is firing upward all the way up to 1800Hz but the wals and the ceiling is probably much more reflective than in Mitcho's room.

Here is the FR, to complete the picture, measured with sweep from LP. All corections are done manually with rePhase, based on REW measurements.

 

[andreasmaaan]

The filters you are mentioning here would be "normal" LR passive filters or DSP filters?

Both. I'm talking about minimum phase filters with relatively gentle (up to 4th- or arguably 8th-order) slopes. These include passive, analogue active and digital filters - the key thing is minimum phase.

Excuse my ignorance but I didn't quite catch up what is the HEDD approacch and what is the alternative, so can you please elaborate?

I'm not sure I've got it 100% right either, but the gist seems to be that they are using active analogue EQ and (of course minimum-phase) filters in the speaker to produce the desired summed amplitude response, then using DSP prior to the speaker to linearise the phase response.

The more optimal alternative would be to use DSP-based linear-phase crossover filters in the first place. This would ensure that - on-axis at least - there would be no ringing (except as a result of the subwoofer's low-frequency roll-off and any attempts made to linearise its phase response).

 

[Krunok]

Both. I'm talking about minimum phase filters with relatively gentle (up to 4th- or arguably 8th-order) slopes. These include passive, analogue active and digital filters - the key thing is minimum phase.



I'm not sure I've got it 100% right either, but the gist seems to be that they are using active analogue EQ and (of course minimum-phase) filters in the speaker to produce the desired summed amplitude response, then using DSP prior to the speaker to linearise the phase response.

The more optimal alternative would be to use DSP-based linear-phase crossover filters in the first place. This would ensure that - on-axis at least - there would be no ringing (except as a result of the subwoofer's low-frequency roll-off).

Oh, that sounds unusual.. Frankly, I thought all digital crossovers are built with linear phase filters and that minimum phase filters are used only with room correction filters.

What is this hybrid approach supposed to do better than linear phase filters?

 

[JohnPM]

Not sure what I'm seeing there in the Spectral Analysis...

Measurement artifact or stepped tone?

Windowing artefact in the Audacity spectral analysis. This is the spectral content of the sweep:

 

[andreasmaaan]

Actually, @KSTR describes the lineariser in some detail here, calling it essentially a "free lunch". Hmm... Perhaps I'm missing something.

 

[Krunok]

Actually, @KSTR describes the lineariser in some detail here, calling it essentially a "free lunch". Hmm... Perhaps I'm missing something.

Well, here is what he said, together with my first thoughts..

"The main point of it is phase unwrapping because that makes no sense to do in analog as it compromises the design right from the start in most any other regards. If you want to design true phase-coherent crossovers they cannot be transient-perfect at the same time, this is a mathematical impossibilty, systems theory thing (unless you resort to a huge peaking allpass cascade which is PITA to design and implement, PSI Audio from Switzerland is one of the few companies who do this). "

Very true, phase can only be fixed using DSP, not analog.

"However, a linear phase crossover with 100% phase coherency is simply established from the minimum phase analog XO by applying the time-inverse of the system allpass function with zero side effects (yes, on of the few instances of a free lunch, almost, some digital headroom is lost but that isn't an problem with 24bit). "

True again, but you can do it easier and better with DSP, and you already have it on hand as you used it for phase linearisation.
Or he thinks, that for some reason, it is better to do it with analog components? Am I missing something?

 

[andreasmaaan]

True again, but you can do it easier and better with DSP, and you already have it on hand as you used it for phase linearisation.
Or he thinks, that for some reason, it is better to do it with analog components? Am I missing something?

I believe he’s actually suggesting it be done with DSP, but that doing it the way the lineariser does (ie by placing a time-inverse all-pass filter upstream) is functionally equivalent to doing it with DSP-based linear-phase crossover filters in the first place.

The view I’ve been putting forward here is that this is not the case, because the former approach, unlike the latter - exhibits pre- and post-ringing on the design axis.

 

[Krunok]

The view I’ve been putting forward here is that this is not the case, because the former approach, unlike the latter - exhibits pre- and post-ringing on the design axis.

I see. Well, your logic seems sound to me.

Btw, with room correction you can only use minimum phase filters, right? So should I be worried with that ringing in IR I posted earlier? I haven't really found any articles on that subject, not even a claims that it can be heard. I came on only a few folks stated that they never really heard it, like Mitch said as well.

What is your take?

 

[DDF]

So should I be worried with that ringing in IR I posted earlier?

@j_j, the most experienced pshycho-acoustician (is that even a word?) here weighed in on it earlier:
https://www.audiosciencereview.com/...n-audio-does-it-matter.11/page-22#post-104672

"Note: There are potential filter issues with pre-ringing, MAYBE. One can propose a plausible mechanism, but nobody, EVER, yet, has actually demonstrated it cleanly. The issues are due to the nonlinearity of the EAR and the minimum-phase issues of the ear."

 

[svart-hvitt]

The new speaker is the F105 - the small one. The F2 (your link) is an old design. None of those are listed for sale at this moment. The new small F105 is not really based on anything from the F2.

The F105 is a concept design I have used to test out some new solutions for cabinet design and improved radiation pattern for small speakers. They are not expensive enough to generate any interest, but for their purpose they are nice - easy to move around, good enough for evaluation of the technical solutions.

What can not be achieved is exactly what the HEDD designer discuss in the movie - dynamics. The improved radiation at lower frequencies improves clarity and imaging, gives less coloring, but the impact and slam of the larger speakers simply is not there. And they can not play loud enough for what I call "full-scale" music reproduction. Some of this can be improved with better drivers, but it is not possible to generate the wavefront that a larger speaker creates. There is a difference even at moderate volume. Soundstage and presentation of the instruments are also different, but not necessarily better or worse than the larger speaker - this changes dramatically though, in a room without the acoustic treatment in Room2, the radiation pattern is quite wide and you will not get the deep reflection free gap I have here in Room2 in a room with reflective surfaces close to the speakers.

("Dynamics" here is a term I believe most of us understand the meaning of - it would be better to use the term "transient reproduction." The definition of dynamics is the difference in loudness between quiet and loud parts, and that is not what we mean here. Here "dynamics" means sense of power and realism and immediacy of attack on drums and all other sorts of percussive instruments.)

What I am doing in this exercise with the F105 (very small) and the F2 (slightly larger) is to find how to make a small speaker with the sound of a large speaker. The presenter in the HEDD video talks about how to make a speaker with the dynamics of a large speaker of the old studio monitor type combined with the resolution, smoothness and imaging of newer hifi-type designs. That is not the same. Because he (HEDD) - as I understand it - does not mind if the speaker is big.

I already have a speaker with both dynamics and resolution and imaging, one that is more like slightly larger in size. But this speaker does not sound like a big hifi-speaker, because the radiation pattern is very different. The HEDD speaker looks like the presentation would be more in the traditional hifi-direction. This does not mean he (HEDD) is wrong an I am right, it is a matter of choice.

I have several articles on the blog-page about this now, one about "Can a small speaker perform like a large", one where the F105 is presented with both measurements and sound impressions. I will not spam this post with links, those interested can easily find it on my web-page in the blog-section.

@Kvalsvoll , you define «dynamics» as «transient reproduction», «sense of power and realism and immediacy of attack on drums and all other sorts of percussive instruments».

So I wondered, what are your experiences with compression horn (F2) vs other tweeter designs, say dome tweeter? And sensitivity; any thoughts on sensitivity (of speaker) vs «transient reproduction»?

 

[Krunok]

@j_j, the most experienced pshycho-acoustician (is that even a word?) here weighed in on it earlier:
https://www.audiosciencereview.com/...n-audio-does-it-matter.11/page-22#post-104672

"Note: There are potential filter issues with pre-ringing, MAYBE. One can propose a plausible mechanism, but nobody, EVER, yet, has actually demonstrated it cleanly. The issues are due to the nonlinearity of the EAR and the minimum-phase issues of the ear."

Are you sure he is not talking about DAC filters there?

"This is why I would prefer 64 kHz sampling. With that sampling rate, there is no such mechanism available. Even 50khz would probably suffice.

None the less, we are stuck at 44.1, see above, and that's how it is. I admit that for most sources, 44.1 should not be an issue, either. "

 

[mitchco]

For those of us who peeked over the edge into the very deep hole, but don't fancy the long trip to the bottom, what's the easy solution called?

Audiolense XO

 

[Snarfie]

Probably an open door but listning to a monitor distance around 1 meter Adam explaind at minute 21;20 (if i understand him correctly) that that is the most objective distance to have a balanced sound. For sure at this distance room reflection are minimal. Does that mean that this distance is the distance where you can judge speakers optimal?. For me i listen the last several years to music from this (near field/monitor) distance an found the music incredible accurate. Esspecialy switching between roomcorrection and/or bypass mode revield the quality of roomcorrection because i guess a more flat frequence respons is than obtaind which give a much more balanced experience.

 

[svart-hvitt]

Probably an open door but listning to a monitor distance around 1 meter Adam explaind at minute 21;20 (if i understand him correctly) that that is the most objective distance to have a balanced sound. For sure at this distance room reflection are minimal. Does that mean that this distance is the distance where you can judge speakers optimal?. For me i listen the last several years to music from this (near field/monitor) distance an found the music incredible accurate. Esspecialy switching between roomcorrection and/or bypass mode revield the quality of roomcorrection because i guess a more flat frequence respons is than obtaind which give a much more balanced experience.

For an example of how a science-driven company guides its clients on direct sound vs reverberation, see the clip below (source: https://www.genelec.com/sites/default/files/immersive_audio_brochure_180806_web_0.pdf).

Please note that the recommended listening distances depend on speaker type and model.

 

[Snarfie]

For an example of how a science-driven company guides its clients on direct sound vs reverberation, see the clip below (source: https://www.genelec.com/sites/default/files/immersive_audio_brochure_180806_web_0.pdf).

Please note that the recommended listening distances depend on speaker type and model.

View attachment 27650

Thanx explains a lot regarding distance. Keep it with my current speakers at 1 meter