I'm not filtering R.A.C.E.
Is that what you mean?
I'm high passing the drivers themselves using minidsp.
I'm letting R.A.C.E. high pass at 6 db along with the top of the 2 way, to blend in with the bottom of the 2 way 1st order. Smooth transition between the xtc filter and the original stereophonic signal. R.A.C.E. will be 6 db down at 200 hz, 12 db down at 100 hz, 18 db down at 50 hz.
The stereophonic signal will be 6 db down at 200 hz, 12 db down at 400 hz and so on.
If the concern is channel matching, well I'm using the minidsp not passive filters.
Is that what you mean?
I'm high passing the drivers themselves using minidsp.
I'm letting R.A.C.E. high pass at 6 db along with the top of the 2 way, to blend in with the bottom of the 2 way 1st order. Smooth transition between the xtc filter and the original stereophonic signal. R.A.C.E. will be 6 db down at 200 hz, 12 db down at 100 hz, 18 db down at 50 hz.
The stereophonic signal will be 6 db down at 200 hz, 12 db down at 400 hz and so on.
If the concern is channel matching, well I'm using the minidsp not passive filters.
But you can!
You could obtain a RACE impulse response. I have a program that can generate them that you can have if you don't want to extact one. You could use one of the other public domain XTC impulses. Then you can butcher them to your heart's content with an audio editor
and use a convolver.I use Convolver — a convolution plug-in
I have used it with windows media player, and the VST version with Winamp, AudioMulch, Minihost, Savihost, Synthedit etc.
It can take a little effort to get going.
Probably fine as you are doing it to both channels, but it is very slow..
For a source panned hard left or right you are going from the drivers been driven with essentially opposite phase with say 3db difference, to just one been driven. I'm not sure I can get my head round that without a measurement or simulation.. I think you will have some non-constant head-related attenuation, which may, or may not be a good thing. That would have some effect on Interaural Time Delay.
Well, I think you should right a tutorial on this subject.
I know I would like to learn how to manipulate algorithms.
I'm sure everybody on this thread would. I'm just not versed in it.
I've spent my time studying acoustics and building loudspeakers. I started before I knew of any digital hoopla
I would like this in a piece of hardware as per miniambio.
You should contribute some info on the thread I started "BACCH vs. R.A.C.E"
It could use some action, there are 3 posts and two of them were me!
hahaha!
I know I would like to learn how to manipulate algorithms.
I'm sure everybody on this thread would. I'm just not versed in it.
I've spent my time studying acoustics and building loudspeakers. I started before I knew of any digital hoopla
I would like this in a piece of hardware as per miniambio.
You should contribute some info on the thread I started "BACCH vs. R.A.C.E"
It could use some action, there are 3 posts and two of them were me!
hahaha!
Last edited:
Well, I think you should right a tutorial on this subject.
I know I would like to learn how to manipulate algorithms.
I'm sure everybody on this thread would. I'm just not versed in it.
I've spent my time studying acoustics and building loudspeakers. I started before I knew of any digital hoopla
I would like this in a piece of hardware as per miniambio.
You should contribute some info on the thread I started "BACCH vs. R.A.C.E"
It could use some action, there are 3 posts and two of them were me!
hahaha!
Best for me not to say what I think about all the hype, given the amount of money involved.
Princeton Alumni Weekly: BREAKING GROUND: Unlocking the key to 3D sound
He did start off involved with Ralph and all that entailed..
Ambisonic Info | Ambiophonics
Reading from David Wareing- who I understand to be one of the earlier developers of crosstalk cancellation software- in this thread, just made my day. Thanks for chiming in, and more so, for all your work and contribution.
Understanding all those concepts and their influence on the sound so that one can even begin to try different configurations without falling in unending, chaotic, hit-or-miss trial and error would take me years, even if I had an expert and a laboratory to teach me. (I wish I did)
However, after having listened to the barrier for some weeks now, I hooked the miniambio back on for comparative purposes. The very short barrier that remains,(check photo on post 562 in this thread) with its reflective foam by the tweeters, makes the miniambio more satisfying than it used to be without it. Short delay (20n)always gave better separation but deadened the highs too much. Now the highs are present again.
It is quite enlightening to alternate listening to both barrier and electronic cancellation for a period of time, the former being a natural effect and a good reference.
Understanding all those concepts and their influence on the sound so that one can even begin to try different configurations without falling in unending, chaotic, hit-or-miss trial and error would take me years, even if I had an expert and a laboratory to teach me. (I wish I did)
However, after having listened to the barrier for some weeks now, I hooked the miniambio back on for comparative purposes. The very short barrier that remains,(check photo on post 562 in this thread) with its reflective foam by the tweeters, makes the miniambio more satisfying than it used to be without it. Short delay (20n)always gave better separation but deadened the highs too much. Now the highs are present again.
It is quite enlightening to alternate listening to both barrier and electronic cancellation for a period of time, the former being a natural effect and a good reference.
Last edited:
I hope I helped wean Ralph off the 'mattress' approach, and can do the same for you
The whole thing goes back to Atal and Schroeder but I learned of the Stereo-dipole (stole the idea) )from Kirkby et al.I would be more than happy to 'teach' you all I know, but I don't want to educate corporations.
From the picture and posts you are obviously a dedicated Ambiophile. I combination of XTC and a barrier might well be a good system. I found that when I was playing with the Jambox recently and also many years ago that a smalll barrier improved the sound. I the case of the Jambox a computer mouse did the trick! This gave me the idea of having small speakers spaced at ear distance with a head, sphere or cylinder in front of it. This would, I think, give twice the normal head attenuation. The idea seemed to work although It occurred to me that nobody would want a TV with a head in front of it staring at them.
I am hoping to find a few experienced Ambiophiles here willing to trade my latest VST for free for some beta-testing. It does not have controls for delay or attenuation but it does have ten knobs to play with.
By the way David.......welcome to the forum!
Sorry I didn't say that earlier, it was very late here and I was dozing off.
We will see how it goes with my crossover design. I think it will work.
I plan on taking measurements with a dummy head at the appropriate distance (12 feet away from the array where the vectors cross)
I plan on taking measurements with a dummy head with the pinna removed, then with the pinna attached, to see what's happening at higher frequencies.
Then eq'ing appropriately. Of course I won't eq deep notches at high frequencies, just an average sum, I plan on cutting R.A.C.E around 6 k, just where fundamentals stop and only harmonics are left to stereo.
Is your program going to include the solution to the over-under estimates that R.A.C.E is faulty at?
And are you even going to add the attenuation and delay to the two bottom knobs that say (knob)?
This is the kind of feedback I was looking for in the BACCH filter vs. R.A.C.E. Thread I started, not just decoding BACCH.
Again, welcome to the thread and will be looking forward to more that you have to say.
Sorry I didn't say that earlier, it was very late here and I was dozing off.
We will see how it goes with my crossover design. I think it will work.
I plan on taking measurements with a dummy head at the appropriate distance (12 feet away from the array where the vectors cross)
I plan on taking measurements with a dummy head with the pinna removed, then with the pinna attached, to see what's happening at higher frequencies.
Then eq'ing appropriately. Of course I won't eq deep notches at high frequencies, just an average sum, I plan on cutting R.A.C.E around 6 k, just where fundamentals stop and only harmonics are left to stereo.
Is your program going to include the solution to the over-under estimates that R.A.C.E is faulty at?
And are you even going to add the attenuation and delay to the two bottom knobs that say (knob)?
This is the kind of feedback I was looking for in the BACCH filter vs. R.A.C.E. Thread I started, not just decoding BACCH.
Again, welcome to the thread and will be looking forward to more that you have to say.
I would feel honored to try your software. Ten knobs sound like a lot of fun.
Right now I have a virus-sick overheating laptop with a noisy fan contraption that I strapped to it plus a jiggly dc supply jack-engineered to malfunction after a year of normal use so one has to change the whole motherboard or better yet buy a new computer. Add to that the two usb audio adapters and dc supplies plus the cable mess, and the idea of a relaxed listening session has turned into a nightmare.
I will try to fix these problems best I can. Having a go at your software is tempting enough to forget these issues.
Thanks for your kind words.
I hope someone will make dummy head recordings of a barrier in action to find out what is actually going on there. I just bought a shop dummy head, binaural mics and a pro soundcard myself. I used to stick panasonic capules in my own ears. Angelo Farina provided me will MLS software years ago for Cooledit, but that has long gone.
RACE sounds much better than I would have expected but I can certainly hear the effect of low and high frequencies not moving much with pink noise.
It may be 'non-optimum' but so is everything else, except perhaps in the inventor's own mind.
Computer generated impulses are capable of much more precise control of frequency and phase than barriers or RACE algorithms, but they take time to perfect and are fixed. I don't see the point of providing controls so large that enable you to generate unsuitable combinations of attenuation and delay.
My VST does eight convolutions and mixes them together in various proportions to satisfy the demand for knobs!
"Is your program going to include the solution to the over-under estimates that R.A.C.E is faulty at?"
My filters always have as I am not restricted to IIRs or 'recursion', although after inverse filtering to create a Dirac impulse at the ear, they look similar to RACE as they must in effect 'cancel the cancellations' to obtain a single pulse. This is probably not strictlty necessary to do, but 'looks right'. A flat frequency response would probably do. I'm not sure why RACE does not attempt some filtering within the feedback, perhaps the original software used would not allow it.
My filters and VST do many other things than just XTC which is just a part of what I think is needed for good stereo reproduction. Happily it sounds very good with binaural recordings, although it was meant for ordinary stereo recordings and ordinary rooms.
I have not labeled the knobs at the bottom but I may well call them Hugo and Zuccarelli as they create interference patterns at the ears, I think.. They certainly enhance the extraction of distance cues and spatial resolution.
You are on!
Ralph had a nightmare getting the previous 2 channel version to work. There are some peculiarities to ConvolverVST involving library files, and peculiarities with AudioMulch 1 involving reloading VSTs which meant that he loaded a wrong config file that Windows kept remembering it and could not be replaced. I also takes a very long time to load in Win7 which I think is down to SynthEdit and the fact that it is Complicated(tm). I could not get him to use MiniHost, which was much more straightforward, although ASIO4ALL is more prone to glitching than the AudioMulch drivers.
What VST hosts do you have, if any?
It will work with just two front speakers but to get the full benefit you need a second rear stereo-dipole. There is a third pair of outputs that can be connected to a normal stereo setup for comparison purposes.
Ralph said that with my VST on the front pair and RACE on the rear, it was the best he had ever heard. I don't think he can use this version without taking his system apart.
You would be the first person to hear this outside of my living room!
"David Wareing seems to have disappeared from the known universe"
What's the matter David? You don't like our universe?
Hahahaha!
Welcome back buddy!
RACE can look very good indeed on paper, for certain settings and head models. The frequency response, both at the ear and speakers is very smooth compared to prof.. Choueiri's "freebee" EYCv2L-44.wav, or even my own filters.
This for a RACE impulse I made for 2dB attenuation and 2 samples .
An externally hosted image should be here but it was not working when we last tested it.
@44100 delay.
I think it was Savihost that I used first. I think it had problems crashing and not easy to recover. Minihost was more trouble-free and had low latency with ASIO4ALL drivers so that I could just about get lipsync with the TV.
VST host from the same author as Savihost has the ability to output many channels which I don't think the other two do.
VSTHost
AudioMulch has the most flexibility and reliablity. The free trial lasts a good while BUT is not good at "reload VST" if you need to change the config file . I use it to output to HDMI to an AV receiver.
It is probably best get used to using Convolver, then ConvolverVST and the VST hosts to avoid the problems Ralph had....
In particular you need to pay attention to this bit:
"2. convolverVST requires libsndfile and libfftw DLLs to run. For many host applications, these must be in the execution path. To add the them to the execution path, right click on My Computer in Windows Explorer, select Properties, the Advanced tab, and the Environment Variables button and append ;C:\Program Files\Convolver\Convolver (or the path to which you installed Convolver in the first step) to either the User or System Path variable using the corresponding Edit button:"
ConvolverVST
To use convolver with the raw filer IRs outside of the my VST you would create a config txt file like this:
44100 2 2 0
0 0
0 0
C:\Filters\David_Wareing3_Left.wav
0
0.0
0.0
C:\Filters\David_Wareing3_Left.wav
1
0.0
1.0
C:\Filters\David_Wareing3_Left.wav
1
1.0
0.0
C:\Filters\David_Wareing3_Left.wav
0
1.0
1.0
Index of /Filters
Those filters are almost the same as within my VST but they roll off to mono at 200 Hz instead of 100 Hz and are not quite as good at group delay.
My VST does two stereo-dipoles and much more.
I hope the process will be worth it in the end!
Hi all,
How one is taking the unsymmetry of HRTF into account for cross talk cancellation ?
Here's an example of unsymmetry for +/-30 deg angles of stereo triangle. Ok, the level might be smaller for 10 deg stereo dipole but maybe not negligible.
Because, in my experience less than 6 dB of ILD is enough to shift the image totally to the side. So level differences of a few dB ccould create huge spatial errors.
As far as I can see, only the physical barrier or individualised HRTF (both sides of the head measured) can eliminate this unsymmetry effect.
Of course unsymmetry of HRTF is individual, some have it more than others. It's just something you have to live with
- Elias
How one is taking the unsymmetry of HRTF into account for cross talk cancellation ?
Here's an example of unsymmetry for +/-30 deg angles of stereo triangle. Ok, the level might be smaller for 10 deg stereo dipole but maybe not negligible.
Because, in my experience less than 6 dB of ILD is enough to shift the image totally to the side. So level differences of a few dB ccould create huge spatial errors.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
As far as I can see, only the physical barrier or individualised HRTF (both sides of the head measured) can eliminate this unsymmetry effect.
Of course unsymmetry of HRTF is individual, some have it more than others. It's just something you have to live with
- Elias
Why alternating ? Why not combine them !?
Barrier is ideal for high freqs, but it cannot attenuate at low freqs. Electronic cancellation works at low-mid freqs but becomes too challenging at high freqs. It would be intuitive to combine these two methods, to pick the best of both of them: Using barrier for high and electronic for low freqs. I havent seen it been done ? Why not ?
- Elias
Is it possible that your hearing is quite different from the average
:Interchannellevel differences and interchannel time differences
Is it possible that your hearing is quite different from the average
Interchannellevel differences and interchannel time differences
While everything is possible, why would you think so ?
I was talking about ILD (interaural level difference which is measured typically at the entrance or inside of the ear canals), the document you linked shows interchannel differences of the stereo media or the loudspeaker signals.
Those interchannel differences are transferred to ILD via HRTF, so they are not the same.
- Elias