LinFIR – DSP Software for FIR/IIR Filter Design and Speaker Correction

[somebodyelse]

I read the Myth paper and this is what I read in section 3:

1. A 3 person BBC Study - 2 famous Violinists and a dealer asked to identify specific instruments - 3 different strads and a modern violin. Participants identified 2 out of the 4 instruments correctly. 50% success.
2. German study - 3 acoustic experts and a group of arborists (113 participants) had to 5 violins to asses - 1 strad and 4 violins made from Norwegian spuce. 39 of them correctly identified the strad. 39/113 = 34.5% correct.

So this would seem to contradict your general statement “ the violinists and other experts can’t identity the Stradivarius by sound”

The standard is to be better by a statistically significant margin than the null hypothesis - in this case that there's no audible difference. Neither look convincing by that standard, although I don't think we have enough information about exactly what was asked to run the stats. With that sort of error rate I think we can safely say that any difference is subtle at best rather than obvious though.

 

[Salt]

Back on topic, please...

 

[landco1]

Hello! I have checked out your software.
Unfortunately, it falls short compared to similar (free) programs, such as Cavern EQ or DecayCore.
The functionality is worse, and there is no support for multi-channel modes with a subwoofer.

 

[john61ct]

Stop looking for things you want in software not designed for those functions.

Rather than complaining climb the learning curve just a bit, so you can ask informed questions.

 

[Arnwald]

Yes understood and agreed, but if you are seeking minimum waveform distortion, then flat gain and phase is the criteria

Assuming the audibility of phase distortion, seeking a perfectly flat phase above 5kHz at the listening position is a utopian. It will depend on air temperature...

Plus, you won't be able to spot the difference on the waveform itself. Needless to say that you will never be able to hear the difference with a few degrees of phase deviation.

 

[Waveform Fidelity]

Hi Aranald

This is my thinking

-air is not a dispersive medium for audio frequencies, so all frequencies travel at the same velocity
- temperature affects the velocity of the wave, but at any short time interval, the air temperature is not changing and it remains the same along the propagation path.

So, if the source is emitting a CW, a microphone placed at a certain, fixed distance away will receive the wave at a certain phase wrt the source. If we monitored this phase over time, and assumed the temperature and distance did not change wouldn’t the phase remain constant? I am also assuming a reflection-free environment.

Hearing any difference is another topic - I just wanted to check we are in agreement on the fundamental physics.

 

[Arnwald]

In fact, air IS dispersive:

The acoustic source and environment by Adam Weisser

Non-idealized acoustics of real-world sound sources, sound propagation in space; interaction with the environment; the ubiquity of naturally occurring sound modulation

hearingtheory.org

Fig. 3.3: between 100 Hz and 10 kHz there is a difference of about 0.06 m/s, which is roughly equal to a phase rotation of 4° at 20 °C, 50% humidity and at 2 meters from the source.

But the temperature in a room is not homogeneous. In fact you can have strong gradients between opposite surfaces (e.g. between ad external and internal wall), and we saw that a difference of only 1 degree can lead to a phase rotation of the magnitude of the phase distorsion of your amp. And here the problem is interesting: low frequencies do not see the gradient but high frequencies do.

And I did not mention the air reactance near the source of the walls.

In my opinion, speaking a perfectly flat phase is rather vain.

 

[Julf]

In my opinion, speaking a perfectly flat phase is rather vain.

Indeed. Not that it matters anyway, as the phase differences aren't audible.

 

[john61ct]

I think even discussion of inaudible factors is a waste

much less actually spending time / energy / money chasing them

 

[Waveform Fidelity]

In fact, air IS dispersive:

The acoustic source and environment by Adam Weisser

Non-idealized acoustics of real-world sound sources, sound propagation in space; interaction with the environment; the ubiquity of naturally occurring sound modulation

hearingtheory.org

Fig. 3.3: between 100 Hz and 10 kHz there is a difference of about 0.06 m/s, which is roughly equal to a phase rotation of 4° at 20 °C, 50% humidity and at 2 meters from the source.

But the temperature in a room is not homogeneous. In fact you can have strong gradients between opposite surfaces (e.g. between ad external and internal wall), and we saw that a difference of only 1 degree can lead to a phase rotation of the magnitude of the phase distorsion of your amp. And here the problem is interesting: low frequencies do not see the gradient but high frequencies do.

And I did not mention the air reactance near the source of the walls.

In my opinion, speaking a perfectly flat phase is rather vain.

Thanks for the references which I will study.

I certainly would not worry about a 4 degrees phase shift between 100 and 10,000 Hz, but a 20 degree phase shift between 1000 Hz and 10,000Hz is worth making compensation.

 

[Waveform Fidelity]

Indeed. Not that it matters anyway, as the phase differences aren't audible.

ah! the same on-going recital with audiophiles in general which treats phase shift independent from amplitude. I invite you to think of audible waveform distortion caused by phase shifts. You know about the Hilbert-Bode transform right?

 

[Keith_W]

Indeed. Not that it matters anyway, as the phase differences aren't audible.

Sorry, that is incorrect. See JJ's post here. The real question is whether additional phase distortion introduced by the loudspeaker is audible, given that many acoustic recordings already have phase distortion, and the room distorts phase even further. But what if you listen to music that has no phase distortion (i.e. electronic music), and you have set up your speakers so that there is less phase distortion (i.e. you have pushed all early reflections in your ETC to below 20dB)? Is it worthwhile linearising the phase of your loudspeakers then? If your software is capable of doing it (and it is trivial with linear-phase DSP) you can generate a set of minphase vs. linphase filters and listen for yourself. IMO the biggest difference is in transparency and transient attack. I think it is easily audible, but unfortunately that's a personal anecdote and not a controlled study.

 

[Julf]

Sorry, that is incorrect. See JJ's post here.

Thank you - I am quite familiar with my friend JJ's work. When I wrote "the phase differences aren't audible" I should have been more clear in stating I was talking about the phase differences we are talking about here. Of course some phase differences are audible, especially ones that differ between the ears.

The real question is whether additional phase distortion introduced by the loudspeaker is audible

Indeed. There was a huge "phase linear" speaker movement in the 1970s/1980s (remember the B&W "pregnant penguin"?) but it all died off as people realized it didn't matter.

If your software is capable of doing it (and it is trivial with linear-phase DSP) you can generate a set of minphase vs. linphase filters and listen for yourself.

Indeed, making sure you only listen with your ears - so in a proper double-blind test.

I think it is easily audible, but unfortunately that's a personal anecdote and not a controlled study.

Indeed. Your subjective perception is noted.

 

[Julf]

ah! the same on-going recital with audiophiles in general which treats phase shift independent from amplitude. I invite you to think of audible waveform distortion caused by phase shifts. You know about the Hilbert-Bode transform right?

Of course - no way I could avoid my basic electrical engineering 101 courses.

I invite you to demonstrate (with evidence) audible waveform distortion caused by non-differential phase shifts (so ones that affect both ears equally).

 

[Waveform Fidelity]

Well, that’s a tough request, as it involves gaining both the opinion of an un-biased audience, in a double blind listening test with special attention to be given to the acoustic environment, both of which are not available to me, no likely other forum members.

Do you know of any research where others have performed the type of test you are proposing?

 

[john61ct]

un-biased audience

no such thing

Only a significant sample size makes it asymptotically approach objectivity.

I reco Dr Toole

Sound Reproduction | The Acoustics and Psychoacoustics of Loudspeakers

This fourth edition of Sound Reproduction: The Acoustics and Psychoacoustics of Loudspeakers, Rooms and Headphones explains the physical and perceptual

www.taylorfrancis.com

 

[Julf]

Well, that’s a tough request, as it involves gaining both the opinion of an un-biased audience,

No, they can be as biased as they want. That is the nice thing about double-bind ABX. I assume you are familiar with ABX methodology? Either they can tell there is a difference (at a statistically significant way) or they get roughly half wrong.

 

[Waveform Fidelity]

Actually not that familiar with double blind ABX, but back to your belief about phase shift not causing audible effects, you must have arrived at that conclusion somehow? Have you done experiments yourself, read scientific papers, etc?

On non differential phase shifts, I am wondering what you mean? Aa phase shift by its definition is a change of phase with respect to an point on a waveform, so there is always a “difference” involved.

 

[Julf]

Actually not that familiar with double blind ABX, but back to your belief about phase shift not causing audible effects, you must have arrived at that conclusion somehow? Have you done experiments yourself, read scientific papers, etc?

Read a bunch of papers and textbooks, yes.

On non differential phase shifts, I am wondering what you mean? Aa phase shift by its definition is a change of phase with respect to an point on a waveform, so there is always a “difference” involved.

I did clarify that by "differential", I meant a difference between the ears (so difference between left and right channels), as the human hearing system is good at detecting that difference to tell the direction the sound is coming from.

 

[Waveform Fidelity]

Oh! do you mean time difference of arrival (TDOA) or phase shift? I understand that it is the time difference of a signal arriving at left and right ear that the brain uses to deduce the direction from which the sound originates.