Jump to content

so, i created a 3ms time delay convolution (did this at 400hz for the scala)


tofu

Recommended Posts

dr who got me thinking. i did some research and found out how to create my own convolutions. there are many convolution supporting applications, but the one most of us probably use is foobar2000. you would load this file into the convolver.

now, i was just fooling around here, and i definitely don't have this thing optimized correctly. i'd much appreciate if dr who could give me more accurate numbers for this convolution.

what i did was create a highpass filter at 400hz (6db butterworth) - FILE A

then create a lowpass filter at 400hz (6db butterworth) - FILE B

add a 3ms silence at the end of file A and paste file B into it, thus creating a 3ms delay for everything above 400hz at a 6db slope.

now to create this properly, i'd need to know where the woofer crosses into the squawker (i assumed 400hz.. it may vary from network to network), what slope i should use, and what the actual delay should be.

the convolution can be acquired here: http://phile.net/delay.wav

Link to comment
Share on other sites

  • Replies 136
  • Created
  • Last Reply

Top Posters In This Topic

Hey, TBrennan! Did you see I got my KHorns? I'm proud of 'em. There is definitely a noticeable difference in the base. Not so noticeable in the MF and HF, though - except on sudden stops w/quiet - then, crash! The bass sounds like it's coming from everywhere between the two corners. Really neat!

Sorry the thing got crazy enough on the Rating thread to make you mad. My advice? Take whatever I say in jest. It was a strong debate, but I didn't get personal. In a good debate, my style is very sarcastic, but never hateful.

Wonder if Tofu's really going to follow through on this delay issue and confirm or refute Dr.Who's uploads?

Link to comment
Share on other sites

dr who got me thinking. i did some research

and found out how to create my own convolutions. there are many

convolution supporting applications, but the one most of us probably

use is foobar2000. you would load this file into the convolver.

now,

i was just fooling around here, and i definitely don't have this thing

optimized correctly. i'd much appreciate if dr who could give me

more accurate numbers for this convolution.

what i did was create a highpass filter at 400hz (6db butterworth) - FILE A

then create a lowpass filter at 400hz (6db butterworth) - FILE B

add

a 3ms silence at the end of file A and paste file B into it, thus

creating a 3ms delay for everything above 400hz at a 6db slope.

now

to create this properly, i'd need to know where the woofer crosses into

the squawker (i assumed 400hz.. it may vary from network to network),

what slope i should use, and what the actual delay should be.

the convolution can be acquired here: http://phile.net/delay.wav

I just downloaded foobar and the convaluter thingy and have no clue how you're doing what you're doing...

Link to comment
Share on other sites

Jeff,

Ok.. but Butterworth filters can't compensate for time delay erros like that! If there's a joke here I'm missing, It's on me!

Al K.

There's a joke behind it. We had a hilarious forum the other night. It got pretty "passionate," so lawsuits were threatened, and the thread got killed and pulled by management. See "Ratings Thread..... Dead."

P. S. Work w/Tofu on helping him out. I think he's serious (really). We have a debate about what's noticeable by the human ear.

Link to comment
Share on other sites

The ear/brain mechanism is complex -- who knows what the heck is really going on in there! We know people hear differently, but I think I've also decided that we aren't all sensitive to the same types of distortion or "artifacts". Not only that, but we all listen differently too. I tend to zone out on drum and guitar work, and this is where my critical listening takes place. Someone else might be big into bass and vocals -- so naturally we are going to come to different conclusions about the overall sound. My emphasis is always on perceived distortion and transients/dyanmics. I don't seem to care so much about things like ambient retrieval, space between the instruments, and other assorted things often talked about. However, I do seem to want some atmosphere -- I like the sound projected out into the room and not hanging two feet in front of the baffle.

Link to comment
Share on other sites

Tofu & Dean:

If you guy's want to create outstanding state of the art filters check with Bell/Labs-Western Electric/Lucent-Ma Bell with their telephone modem stuff. I worked with filters in the lower baseband that would drop 12khz levels down 70db at the 3HZ points with just a few 10th's thru loss.

JJK

Link to comment
Share on other sites

Folks, this debate is a moot point. We can debate it from the point of view of many not familiar with the extensive work that has been done in the area that has been accepted as valid and dynamically evolving. But nevertheless, our 'feelings' don't change the facts. But we would all benefit by learning enough about the material, methods and language - the new paradigm so that a meaningful dialogue can be conducted.

Welcome to the realm of the time domain. And I find it humorous how many stand at the edge of the great abyss and speculate regarding how deep it is! Especially as so many are still stuck on the idea of trying to determine it is is really valid!!!!

J. Blauert and P. Laws, "Group Delay Distortions in Electroacoustical

Systems," J. Acoust. Soc. vol. 63, pp. 1478-1483, May 1978. documents the audibility of group delay. Although as John Murphy points out, more research is necessary to explore increased detail in audible limits, especially in the LF range where group delay is typically greater..

Frequency Threshold

500 Hz 3.2 ms

1 kHz 2 ms

2 kHz 1 ms

4 kHz 1.5 ms

8 kHz 2 ms

"If a network is minimum phase, there exists a unique relationship between

amplitude and phase which allows a complete determination of phase from amplitude... and the amplitude and phase are Hilbert transforms of each other". (Dick Heyser "Loudspeaker Phase Characteristics and Time Delay Distortion: Part 1",p.22, JAES, Jan 1969 published in Time Delay Spectroscopy):

"A minimum phase system is one that has an orderly predictable transform between its magnitude and its phase (i.e.: one predicts the other). The transform that accomplishes this is a local to local transform is called the Hilbert transform. The observation that the phase response is the Hilbert transform of the magnitude response is sufficient proof that the system is a minimum phase system.

Electronic minimum phase systems can be treated with either FFT's or TEFs. Non-minimum phase, non-linear, time dispersion, noisy systems (i.e., real sound systems) require TEF analysis.

A non-minimum phase system is one that exhibits an excess delay of the signal over that termed the phase delay. This excess delay prohibits using the Hilbert transform to predict the phase from the magnitude." Davis, Sound System Engineering, p486)

...but in order to understand this, it requires a further understanding of the Nyquist/Heyser response and its component real impulse response and imaginary doublet response in the time domain, and the corresponding real coincident response and imaginary quadrature response in the frequency domain...Additionally, the relationship of the complex reactance to potential (real) and kinetic (imaginary/reactance) system energy is fundamental.

"The Differential Time-Delay Distortion and Differential Phase-Shift Distortion as Measures of Phase Linearity" by Marshall Leach, JAES, Vol. 37, No.9, September 1989 also explores additional relationships withing the time domain.

And John Murphy also amplifies these results on his TrueAudio website.http://www.trueaudio.com/post_010.htm

I think its great that a few (a very few!) are actually thinking about this. It is also a shame that so many who know so much (yeah...;-) choose to declare it a non-issue and live in the 1930's.

But this is addressed at length in Heyser's Anthology, and in Davis' Sound System Engineering, and rather than be considered an unusual position, it is acknowledged, presented, developed and it remains an area bearing further scrutiny and understanding.

Unfortunately, as so few are aware of this here, and even fewer are open minded to explore it, one wonders what is to be gained trying to discuss it on the site except to encounter another Luddite slander fest....And that is a shame.

Also, just a quick comment as a description does not replace the fundamental work required to become familiar with and to learn the new paradigm!- as the tools heretofor which folks are using are inadequate. Both the language and the traditional measurements are inadequate to grasp and to describe the relationship between perception and the means by which to provide meaningful measurements that correlate to that perception. This is a fundamental position of Heyser, and exactly what he tries to address! Only instead of adopting the subjectivist position encountered here of simply dismissing science and wallowing in their 'feelings', he moves beyond the traditional paradigms to create a new model that will more accurately map perception to science, and measurements to audible experience. And in doeing so, it renders both camps, if you remain in the old paradigm, as superfluous...And with it the debate between both of the inadequate old school camps as simply noise.

Just please note as well - you will not simply jump into this debate after reading a few paragraphs! It is akin to quantum electrodynamics! It is not a simple world where only one or two rules have changed!!!! And there are allot of fundamentals that need to be addressed first! As the fundamental rules have changed! This new world is a blend of physics and psychoacoustics, of phenomena and perception! And that is the problem with almost all of these discussions! People seem to have an idea that they just need the new bit of info and they can can proceed as they have before, albeit with the new fact or value!!!! Nope!!!!

And without a strong basic understanding, a strong foundation - this will all sound Martian! As the basic rules have changed! It quite litereally is akin to trying to address quantum with the classical rules of physics! It does not work! Audio has indeed travelled through the looking glass. And its time to learn the new paradigm.

Link to comment
Share on other sites

I think we need abit of a reality check here. First what is it you are trying to fix, and more importantly does it need to be fixed.

In applying the convolution there will be a time delay in itself. This, of course, is proportional to the number of delay-slide-multiplies that one is performing (utimately a function of the number of taps or the steepness of the filter). Is the entire signal (across the entire bandwidth) being delayed followed by an additional delay in the band of interest?

Second, is this being done because of the additional delay during the transduction with the frequency components through the bass bin (and its various folds). If this is the case, then you may want a steeper filter, since a simple, low order Butterworth will have appreciable overlap with what is going through the mid-range horn (I will ignore the fact that the crossover itself creates relative phase delays - which are probably incosequential - except where the spectra overlap between the 2 drivers).

Third, be careful with the zero padding (the 3 ms of silence). This deserves attention regarding what windowing or gating is applied. If you do not know what I am referring to, you need to do some homework.

I hope what I am conveying is that this DSP is tricky business and not as straight forward as you might think. However, I am interested in what you come up with.

In aswer to my other question - does it need to be fixed? Take this as a grain of salt, since there will possibly be sonic consequences when you start putting the signal through extra stages (op amps , ADC/DAC, reconstruction filters etc).

Good luck,

-Tom

Link to comment
Share on other sites

Guys,

Just so a few points are clear:

1-A loudspeaker is NOT a minimum phase system. It has excess phase shift out the ying-yang!

2-Time or "propogation delay" is NOT the same a "group dealy". Group dealy distorts a complex waveform by scattering the phase relationships of the harmonics. You ear/brain can't determine those phase relationships in real time. It just sin't that fast a computer! It is NOT and oscilloscope!

3-Group delay is simply a measure of phase linearity over a range of frequency. The human ear is deaf to phase, so group delay is a moot point.

4-The B&L threashorld numbers were dtermined by listeners TRAINED to know what to listen for and were only percieved with test tones through headphones!

I have actually measured the time delay and group delay of filters and they are not anywhere equal. Not even close!

Al K.

Link to comment
Share on other sites

add a 3ms silence at the end of file A and paste

file B into it, thus creating a 3ms delay for everything above 400hz at

a 6db slope.

You mean at the beggining right?

yes, sorry. got my a's and b's mixed up. cut me some slack, i've just been using my convolutions to equalize [:P]

as for a steeper slope, i figured that, but i still get the damn echo even at 180db (which is the highest my software goes). now should i be doing this time delay where the woofer is no longer crossing over into the squawker's area? perhaps the filter is adding a delay of its own, thus making 3ms far too much. hell if i know...

if you can explain how you edited those nightwish files i might be able to do this correctly. until then, i'll get back to gluing this subwoofer cabinet together.

Link to comment
Share on other sites

Many loudpeakers are minimum phase!

Interesting, as you cannot EQ a non-minimum phase signal. - rather makes you wonder what many are using them for!

And an interesting definition of group delay!
Group delay is the rate of change of phase with respect to frequency! - In other words, the derivative of phase. (the tangent slope of the rate of change of phase with respect to requency)

And with speakers such as the LaScala and the KHorn, the rate of change of phase with the radically offset acoustical centers, are No where near a smooth phase transition! Thus the phase change and the resultant group delay is important! Focusing on one and ignoring the other is like saying that a rain drop is important, but the rain from a hurricane is not!

As I said, this is a pointless discussion if everyone isn't aware of the foundations as we simply yell at each other, each preaching their own philosophy. And if you would bother checking the sources from which I presented the information, you just might discover that not only is there more, but that this is not simply my off the wall interpretation of the facts.

So, by all means, persist in doing whatever you like....
As far as those who think this is incorrect, please consider publishing a book, as I am sure AES is just waiting to publish it! Isn't it ironic that with all of the significant audio pioneers, that the only author to have his collected works published is Dick Heyser? Or at the very least, try reading the works of someone who has made a significant contribution to the industry!

Have fun....

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...