Jump to content

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


tofu

Recommended Posts

The bigger question that will remain unanswered here is, although our ears do not process the signals as discrete signals, radically separate from each other, does that mean that the smearing is meaningless. In other words, does complex behavior within this time differential not matter simply because we do not hear the discreet events separately - while we do hear the gross changes resulting form changes in the particular components? And because we do not process them as discrete events, does that mean that the events within that interval are dismissible as meaningless as so many are so quick to do? **

Ah, young grasshopper.... I mean Dragonfly! Reading your comment carefully, you assume "we do hear gross changes." That's the whole argument, and instead of proving it, you assume it and say everyone else is too easily dismissive. YOU are too easily dismissive.

Here's the test. Get 2 tracks - 1 original and 1 time-delayed. Play these tracks through the exact same speakers with 10 people in the room - all motionless and with their eyes closed. Tell them nothing about which track is which. Play the tracks back to back a few times if they need it. Then, ask them, separately, and not in the presence of each other or within the hearing of each other, which track was original and which was delayed. I bet you get a screwball set of answers all over the place because on average, nobody can tell - and that's what's important. These comparative tracks will leave people guessing like monkies (audiophiles[:P]) and some will guess correctly out of shear chance only. That's my hypothesis. Care to test your assumption that they are audible differences?

Link to comment
Share on other sites

  • Replies 136
  • Created
  • Last Reply

Top Posters In This Topic

I could do your exact same test comparing different pairs of speakers and nobody would pass the test either...especially if the test subjects had never heard of the speakers before.

So just because it is difficult for an untrained ear to identify in words which is which, doesn't mean the the untrained ear still doesn't notice the differences and can distinguish between the two. To get around this variable, try conducting an ABX test. The test subjects listen to sample A and then sample B (being told which is which). Then they listen to a 3rd sample, X and have to write down if they think X is A or B. X is of course randomly generated each time the test is performed (say 10 trials?). I believe a success rate of 70% indicates that there is a sufficient audible difference to make it worthwhile.

Link to comment
Share on other sites

The bigger question that will remain unanswered here is, although our ears do not process the signals as discrete signals, radically separate from each other, does that mean that the smearing is meaningless. In other words, does complex behavior within this time differential not matter simply because we do not hear the discreet events separately - while we do hear the gross changes resulting form changes in the particular components? And because we do not process them as discrete events, does that mean that the events within that interval are dismissible as meaningless as so many are so quick to do? **

Ah, young grasshopper.... I mean Dragonfly! Reading your comment carefully, you assume "we do hear gross changes." That's the whole argument, and instead of proving it, you assume it and say everyone else is too easily dismissive. YOU are too easily dismissive.

Here's the test. Get 2 tracks - 1 original and 1 time-delayed. Play these tracks through the exact same speakers with 10 people in the room - all motionless and with their eyes closed. Tell them nothing about which track is which. Play the tracks back to back a few times if they need it. Then, ask them, separately, and not in the presence of each other or within the hearing of each other, which track was original and which was delayed. I bet you get a screwball set of answers all over the place because on average, nobody can tell - and that's what's important. These comparative tracks will leave people guessing like monkies (audiophiles[:P]) and some will guess correctly out of shear chance only. That's my hypothesis. Care to test your assumption that they are audible differences?

Untrained ear????

You can't train someone to increase the resolution of their hearing! It is called the Haas effect! It is fundamental to psychoacoustics!

You do not hear "gross changes" in the sense of 'large' changes as you seem to interpret the words.You hear gross changes in the overall quality changes in the larger sense of the experience! In other words, you do not hear the distinct signals as discrete events, but the overall character of the sound will change - the gross experience - the initial signal takes precedence and the delayed signals within this period are 'fused' - smeared with an apparent increase in gain and the sonic character is altered due to the smear.

This is not a hypothesis! And I do not "assume"! This is a synopsis of well established research!

Hypothesize all you like! This is an established fact!

This is allot of fun, but all of this is already well established!

You missed the entire point! I am glad youare asking a few questions which are new to you, but this has already been well researched!

Link to comment
Share on other sites

"I believe a success rate of 70% indicates that there is a sufficient audible difference to make it worthwhile. "

Typically to be considered to 'pass' an ABX test you have to hit the 95% confidence level. What that means is the probability that you actually heard differences, and weren't just guessing, is 95% or greater.

IOW you need to hit the number of correct answers such that the possibility of you guessing is 5% or lower.

A table of this is here:

http://www.pcavtech.com/abx/abx_bino.htm

For example on 10 trials 8 correct responses equates to a probablility of guessing about 5.5% of the time.

Shawn

Link to comment
Share on other sites

Folks, you are chasing the wrong rabbit here!

The Haas effect is not subject to ABX testing nor to 70 or 80 or 85% of the people listening!

And no! Reading up on the Haas effect will NOT explain what is happening during this period! It only describes the net results of fusion and precedence!

Have fun!

When you are ready, read Heyser and discover what is happening during the smear - and how the non-linear and non-minimum phase environment might be addressed - and more...

(Oh, and to reference the point made by AL K earlier - most speakers ARE non-minimum phase - but not ALL! - and the non-minimum phase environment renders our linear systems at distinct disadvantage! This is the realm where Heyser shines! Read him! As I am not going to try to repeat what he presents much more completely and elegantly!)

This forum reminds me of a college. Imagine that they have dug big holes all over the campus and camouflaged them (rather like the legendary tiger pits!) Only this time, if you fall into them, they are going to offer you complete assistance and tutoring to solve your questions.

Just like on too many campuses, too many here spend all of their time crawling around on their hands and knees lest they fall into one of these 'hazards'! Such are the writings of Heyser and the text of Davis. They address the fundamental issues so many have, and they go far beyond them! So, whatever you do!!! DO NOT fall into the pit and actually increase your knowledge and understanding! This is where you stand with your hands over your ears chanting "LaLaLaLaLaLa......" lest you hear about this![:P]

Link to comment
Share on other sites

To address the combined responses of a few of you in your prior posts, you are right in that it is not important that a listener can tell which is which. It is more important as to whether they can tell a difference. I thought about that while I was out the door on my way to lunch. I knew one of you would raise the issue, so here I respond.

Try the same test, but this time give them 3 samples, A=B and C is delayed. Ask them whether A=B, A=C or A=B=C. I still bet you get a answers about what you'd expect if the room was full of monkies. BTW I admit my response in the now infamous Ratings thread indicated I thought I heard a difference. It could be, and probably was, a monkey-like response.

As to whether some dude with a name wrote on this before or not (in terms of your trying to legitimize there is an audible difference), that's fine if you are inclined to believe it. But do not necessarily contend that it makes you KNOW there is a difference. All you can claim to KNOW is some guy wrote that there was. If you want to prove it to me show me the test results from the simple test format I suggested.

Also, in my "monkey-like" response in the Ratrings thread, you'll recall I thought the delayed version sounded better to me. So, using the same suggested test, ask each person which one(s) sounded better. You'll get all kinds of non-sense! Or........ sense, as I would hypothesize - the sense being there's no difference. That would leave all this talk about 7ms delays between mid and woofer to be just a bunch of rubbish. I am SURE you'd have responses from the test equivalent to the randomness you'd get out of monkies - nowhere close to 70% in terms of a CORRECT response (which would be A=B and C is different, and A and B are better).

I'll bet you'd also get monkey-like responses from a room full of so-called audiophiles. Dr. Who, you've got the equipment. Try it. And then, post the HONEST results. It'd be very interesting to know.

In the end, even if there was a strong correlation showing people can discern, we will have the back-up question on which to place our focus. Can somebody tell what "better" is. If they fail this test, then, who cares whether there is a difference?

I will maintain that what sounds "better" is not the same as which sample exudes a more accurate reproduction. Hell, why do we care about accurate? We want "more pleasing." Think about it. That's why Aerosmith "reproduced" the Beatles' "Come Together." Some people like the "inaccuracies" of the Aerosmith version.

At least one thing's for sure, this started out with somebody pushing an idea that an echo would be noticeable from a 7ms delay. Not a difference, but a real echo (two taps instead of one). It seems everybody has come to grips that this was just plain false.

Do you want to go further to prove there's a difference? Try my simple test lay-out. You've got the equipment, Dr. Who. Post honest results, and let's see what happens.

Link to comment
Share on other sites

To address the combined responses of a few of you in your prior posts, you are right in that it is not important that a listener can tell which is which. It is more important as to whether they can tell a difference. I thought about that while I was out the door on my way to lunch. I knew one of you would raise the issue, so here I respond.

Sorry, but you have missed the point of my references. And you have misunderstood the acoustical points I alluded to.

As to whether some dude with a name wrote on this before or not (in terms of your trying to legitimize there is an audible difference), that's fine if you are inclined to believe it. But do not necessarily contend that it makes you KNOW there is a difference. All you can claim to KNOW is some guy wrote that there was. If you want to prove it to me show me the test results from the simple test format I suggested.

Amazing arrogance and ignore-ance.

We do it all the time in tuning rooms! The smearing due to early reflections arriving within the Haas Initial Signal Delay Gap are removed as they interfere with the intelligibility of the direct (initial) signal.

Funny, "some dude" wrote about quantum mechanics too. Prove it to me!

Also, in my "monkey-like" response in the Ratrings thread, you'll recall I thought the delayed version sounded better to me. So, using the same suggested test, ask each person which one(s) sounded better. You'll get all kinds of non-sense! Or........ sense, as I would hypothesize - the sense being there's no difference. That would leave all this talk about 7ms delays between mid and woofer to be just a bunch of rubbish. I am SURE you'd have responses from the test equivalent to the randomness you'd get out of monkies - nowhere close to 70% in terms of a CORRECT response (which would be A=B and C is different, and A and B are better).

I'll bet you'd also get monkey-like responses from a room full of so-called audiophiles. Dr. Who, you've got the equipment. Try it. And then, post the HONEST results. It'd be very interesting to know.

In the end, even if there was a strong correlation showing people can discern, we will have the back-up question on which to place our focus. Can somebody tell what "better" is. If they fail this test, then, who cares whether there is a difference?

I will maintain that what sounds "better" is not the same as which sample exudes a more accurate reproduction. Hell, why do we care about accurate? We want "more pleasing." Think about it. That's why Aerosmith "reproduced" the Beatles' "Come Together." Some people like the "inaccuracies" of the Aerosmith version.

At least one thing's for sure, this started out with somebody pushing an idea that an echo would be noticeable from a 7ms delay. Not a difference, but a real echo (two taps instead of one). It seems everybody has come to grips that this was just plain false.

Do you want to go further to prove there's a difference? Try my simple test lay-out. You've got the equipment, Dr. Who. Post honest results, and let's see what happens.

"All you can claim to KNOW is some guy wrote that there was. If you want to prove it to me show me the test results from the simple test format I suggested."

We do it all the time in tuning rooms! The smearing due to early reflections arriving within the Haas Initial Signal Delay Gap are removed as they interfere with the intelligibility of the direct (initial) signal.

"I'll bet you'd also get monkey-like responses from a room full of so-called audiophiles. Dr. Who, you've got the equipment. Try it. And then, post the HONEST results. It'd be very interesting to know."

Really Doc? You have equipment to measure in the time domain?[:P] Have you gotten a TEF(TDS/MLS), SMAART or MLSSA??

Thus far I am reading "monkey-like" postulating here!

"As to whether some dude with a name wrote on this before or not (in terms of your trying to legitimize there is an audible difference), that's fine if you are inclined to believe it. But do not necessarily contend that it makes you KNOW there is a difference. All you can claim to KNOW is some guy wrote that there was. If you want to prove it to me show me the test results from the simple test format I suggested."

You happen to be wrong. I have been involved in such experiments. And we address the results all the time in tuning rooms! The smearing due to early reflections arriving within the Haas Initial Signal Delay Gap are removed as they interfere with the intelligibility of the direct (initial) signal.

Funny, "some dude" wrote about quantum mechanics too. Prove it to me!

Let's see, in audio, Faraday, Haas, Heyser, Davis, Keele, Patronis, DiAntonio and many more - Yup!! - some dudes!!!!

Look, I don't mean to be rude, but you started out by saying you weren't current with all of your math and whatever. Well, I don't feel the need to go back and reinvent acoustics for you. If you want to doubt the world is round or that the sun is hot without me bringing you proof, be my guest. I have no problem reviewing appropriate current material germain to the immediate present and the future, but I am not a surrogate for your reading and research.

And your issue with time and pitch mixes two domains. You are committing a fundamental category mistake. And the two spaces do not even map in a linear fashion to each other unless they are minimum phase. So your using characteristics of one to quality the other are fundamentally flawed. And no, I do not have the time nor the inclination to 'invent' the mapping of topological spaces to another for you as you watch.

As is too often the case, instead of these explorations presenting issues that rise to explore more mature treatments to further illuminate concepts, this, as is so often the case, simply degrades into drivel.

If anyone wants to refer systematically to Heyser or to Davis' applied text (as I was present for a good portion of the applied concepts - at least those since ~1988 in the book) I would love to discuss and explore the concepts with anyone so interested. Using these works as focal points i don't mind digressing in order to try to clarify or explain concepts, but I am not interested in listening to someone show up and start debating the premise of the work - you know - such folks as those who feel calculus is fundamentally flawed, and the other drivel that we have heard before here. And if anyone even wants to go 'off site' and create a Yahoo conference chat room we can do that and hopefully have more access to graphics.

But I guess I become bored really easily with the simple dismissal of established issues which have not only been reproduced, but which have withstood the challenge of some exceptional peers. And while the implications of ideas are fun to explore, simply assuming a Humean (like Hume) air of doubting everything just to be cute is rather childish. And I also get bored simply proving first principals again simply because someone is too lazy to research the topic themselves. Helping is one thing, doing it for them because they are too lazy is another. I already know it, I don't feel the compulsion to reinvent the wheel simply as a character building exercise! Besides, character isn't marketable anymore anyway!

But personally, I would love to discuss the relevance and the potential for Heyser's applied ideas as well as his theory. And with enough interest, we can play with the TEF and also get some of the ETF folks up to speed so that they can carry the discussion!!!!.As it would be wonderful to engage in an expansive discussion for a change.

I will excuse myself at this point and leave you folks to the monkies - or are you the monkies? He wasn't very clear at different points...Hmmm,...Oh well.

Besides, I still have to finish mowing the yard before it gets dark!

I will leave the rest of you to the monkey house![:P]

Link to comment
Share on other sites

dragnfyr,

I'm confused...you do think time-alignment is an issue, right? Or

perhaps I should ask what does Heiser say on the matter? [;)] Is this

specific topic discussed in that book you referenced in the PM?

I know I've been all over the map struggling with terminology trying to

explain some of what I think happens, so perhaps you could in a

nutshell explain the core concept behind the importance (or

uninportance) of the time-alignment. I must confess that I was just

spitting out concepts in hopes that someone would correct them if they

were wrong, but alas everyone interpreted them in completely different

ways, lol. The way I see it, you have 3 seperate sounds happening over

a period of time that together are percieved as a single instant in

time, but the overall timbre changes compared to a time-aligned system

because our ears detect the "average" over the time period.

Btw, as far as equipment I think he meant that I have the ability to

time delay an audio sample to create some comparison listening tests.

For the few songs I've editied I have been able to distinctly tell the

difference...I would even create a playlist of the two audio samples,

click next a million times so that it lands on a random track and I

could always tell what was playing without peaking. Heck, when at

colter's place on saturday we could tell the difference right away on

his crappy computer speakers.

I also figured out how to make a smaller time delay. Apparently if

you zoom in all the way you can have increments down to .01 ms. Silly

program...now to figure out how to get a lower crossover. I would

really love to provide an audio sample to demonstrate the differences

between a khorn and a time-aligned khorn. [;)]

Link to comment
Share on other sites

dragnfyr,

I must confess that I was just spitting out concepts in hopes that someone would correct them if they were wrong, but alas everyone interpreted them in completely different ways, lol.

Dr Who, me, too! You appeared to have tried the "no-peek" test. Did you try w/A,B and C sample where 2 are original and 1 is modified? Mixing them up in different orders to see if you're still capable of determing the difference?

Link to comment
Share on other sites

Oh, Tofu....... I have noticed you've been REAL quiet about putting in your 2 cents. Looks like your quietly working on the test model. Tell us whether we should sell a kidney or not.... not.... not...

send me a second amp and i'll tell you in a couple of minutes

Link to comment
Share on other sites

if all your files do is exaggerate the time delay

rather than correct it, so a person can get a general idea, i can

create a convolution for that real quick at 400hz so they can use it

with any music file they want.

let me know.

I wouldn't mind if you explained how to do the convolution thing....I

don't have the menus in your screenshot in my recent download of

foobar....are there plugins that I need to download?

I do have the thing to implement the convalution thing....could you

make one with a 400Hz crossover and 7ms delay on the low end? I don't

have khorns here so I wanted to listen to something similar.

Link to comment
Share on other sites

basically, as i understand it from my lacking research, convolvers "mold" whatever media you're playing around an impulse. the impulse is what i'd be creating in this case. it does this all via frequency.

you can create lowpass and highpass filter convolutions with a bunch of softwares, such as scopefir. what i did was take a "neutral" impulse file (all frequency decibel levels the same and there is no delay or effects added) and made my highpass/lowpass from that, since i do not posses a license for the other specialized software.

convolutions can do many things. act as an equalizer (fully customizable of course), use it as a crossover (not with foobar's convolver though.. it's a bit limited), make accurate reverb effects, etc... the possibilities really are endless. and it does all these things extremely efficiently, and much cheaper than buying the actual hardware to do it standalone. what i'd really like to try is some DRC. there is software that creates impulses catered to what it analyzes from your room, but i don't have a microphone.

i uploaded the impulse i start with here: http://phile.net/impulse.wav

i'll work on that k-horn simulator after i grab a tasty beverage.

Link to comment
Share on other sites

Let's go back to square one. Where did the 7mS come from? A Khorn bass bin is 6' (1'up, 2'back and 3'forward). The K-400 is 2' long. The net therefore is 4'. 4'/1100'/sec=.003634sec. From there one would have to calculate the difference in inertia between the K-33 and K-77 diaphrams less the time loss in the autoformer v the woofer inductor. Totaled, it is hard to believe in my little pea brain that 7mS is the correct figure.

A Belle would be 1.36mS plus inertia, etc.

In modern "time aligned" loudspeakers, the rake between the woofer and mid is what 4". That is 0.0003mS a totally insignificant change except for styling and marketing purposes.

Rick

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...