Bob, I was talking with BYRTT about this project of yours and he mentioned something I can only agree with. I know you're an all analog man for lack of a better description. But the things you propose could most probably be very easy to implement with JRiver. Not for keeps, but to find your recipe. It's fast, very flexible and cheap to experiment that way so you actually know what to expect when you build it all in analog.
It could be a great "Lab tool" for you to find what you're looking for. Just a friendly suggestion
.
Yep, but the setup is weird and unacceptably unconfortable, just worth the shairing an experiment...
But, and most important for me, as usual with any experiment of this kind, is that it always half works, hence never really worth the hassle, except for the fun of setting up weird unusefull stuff.
Btw, if Mr Geddes really needs a 3rd channel to make his speakers image better with a stereo program, if find it a really disappointing info regarding his designs...
I never found that horns image any good, but had some faith that i was never given the opportunity to listen to the good ones...
Btw, if Mr Geddes really needs a 3rd channel to make his speakers image better with a stereo program, if find it a really disappointing info regarding his designs...
I don't use the center for stereo, don't need it, the imaging is perfect without it. Don't make up things that I didn't say.
Hence, you should not criticize that which you don't know anything about.
Last edited:
I didn't, simply thought, in the context of this file about weird stereo shufflers, extenders and stuff like this, that this point needed to be clarified before people begin to make up things like "in the end will have to do the same as Geddes: put a damned center channel!".
Last edited:
GDO,
maybe i misunderstand but did discussion over time take a turn and get more and more a subjective opinion mine speakers is better than yours debate, it looks most speakers happen is unacceptably uncomfortable / half works / not worth the hassle and weird including also thread filter exercises is also just weird. Personal have none of the speakers M2 or horns you criticize myself so you don't stand on my sore toes, but if you have all the keys to perfect acoustic speaker please share in diy language. Also some measurements data would be nice to study where actual gedlee's speakers and also wesayso's diy build have some fine data we can compare to.
maybe i misunderstand but did discussion over time take a turn and get more and more a subjective opinion mine speakers is better than yours debate, it looks most speakers happen is unacceptably uncomfortable / half works / not worth the hassle and weird including also thread filter exercises is also just weird. Personal have none of the speakers M2 or horns you criticize myself so you don't stand on my sore toes, but if you have all the keys to perfect acoustic speaker please share in diy language. Also some measurements data would be nice to study where actual gedlee's speakers and also wesayso's diy build have some fine data we can compare to.
GDO,
I admit tone is bitter and sarcastic, but intentions are not bad at all: see 2 posts above...
I have simply reported here that i share the main issue commented by Pano, and explained my way of dealing with them. Only my 2 cents, and probably do not deserve more.
Btw, i don't like to overly complicate setups with additonal treatments and speakers, all the more so as no miracle seems to be expected, since all experiments i have read or heard about ( as reported by the experimenter in person like S.L. on Watson above) seem to lead to that conclusion owing to flaws belonging the recording techniques involved.
Obviously i also understand that guys might want to have fun experimenting too, and it's true i have not much more to add on this topic, nor want to steal more bandwidth...
Last edited:
I've also been down the path of more and more complexity and have always found it to be fascinating at first, but unsatisfying in the long run. Then rinse and repeat.
The dulling of the center image is an effect that has interested me for years, so out of a desire to tinker I looked for a fix and stumbled across the shuffler. It worked well on my large horn system, but created the new problem of a slightly too bright overall tonal balance. Not surprising when you consider the effect it has on center EQ, but that's just another thing to fix. Like speaker placement, toe in, room treatment, etc.
FWIW, the Gedlee speakers image just fine as a stereo pair. No problem in that department. Never had a chance to test them for phantom center dullness, but I suppose that could be room dependent.
The dulling of the center image is an effect that has interested me for years, so out of a desire to tinker I looked for a fix and stumbled across the shuffler. It worked well on my large horn system, but created the new problem of a slightly too bright overall tonal balance. Not surprising when you consider the effect it has on center EQ, but that's just another thing to fix. Like speaker placement, toe in, room treatment, etc.
FWIW, the Gedlee speakers image just fine as a stereo pair. No problem in that department. Never had a chance to test them for phantom center dullness, but I suppose that could be room dependent.
Pano,
I've been watching and reading the thread for awhile now and I don't see how it matters much if you are using a line array as Wesayso is or are using a horn system like Earl does? Since we are talking about comb filtering caused by the two separate channels having the same information that is supposed to create a center panned recorded sound how does it matter which speaker type you are using as long as the sum of the two speakers creates that null on the center listening axis? Even if you toe the speakers in and cross in front of the listener it seems you would still have a very similar end result perhaps at a slightly different frequency due to off axis response. I'm not sure how little phase change is actually required in the shuffler to do what your after but it does seem like about the minimum requirement at a couple of frequencies to get the results your looking for.
I keep going back in my thoughts to how the majority of speakers are tested and even when they are tested in a stereo arrangement and not just doing mono response and phase curves the majority of the time it is done with a single microphone and not with two mics set up like a bi-n aural recording setup so I take it this is why this isn't a common discussion about the tonal balance.
I've been watching and reading the thread for awhile now and I don't see how it matters much if you are using a line array as Wesayso is or are using a horn system like Earl does? Since we are talking about comb filtering caused by the two separate channels having the same information that is supposed to create a center panned recorded sound how does it matter which speaker type you are using as long as the sum of the two speakers creates that null on the center listening axis? Even if you toe the speakers in and cross in front of the listener it seems you would still have a very similar end result perhaps at a slightly different frequency due to off axis response. I'm not sure how little phase change is actually required in the shuffler to do what your after but it does seem like about the minimum requirement at a couple of frequencies to get the results your looking for.
I keep going back in my thoughts to how the majority of speakers are tested and even when they are tested in a stereo arrangement and not just doing mono response and phase curves the majority of the time it is done with a single microphone and not with two mics set up like a bi-n aural recording setup so I take it this is why this isn't a common discussion about the tonal balance.
Comb filtering? Null? This sounds anechoic thinking to me. Real untreated rooms should prevent this to be an issue, or to some extent only...
I mean comb filtering does not matter at all ( if it did Geddes stuff would suck as well, as far as i can imagine...) the problem must be else where.
Last edited:
So you don't experience this problem GDO, we get it...
To me, my speakers are a test bed to learn a thing or two about audio reproduction so I experiment. I don't have anechoic like behavior but do have suppressed early reflections if you compare to more common in room results. I actually wanted that, strange huh?
I'd have to agree (and think I've stated as much) with Kindhornman here that under the circumstances of suppressed or absent early reflections it would most probably be a Stereo triangle problem that plagues more than one speaker design. It might be a bit different from one speaker to the next and I wouldn't even call it a big problem. Though if we can improve upon it with this super simple trick, why not! I guess you'd first want to know/check how your phase is actually doing at the listening spot to see if you need it at all.
To me, my speakers are a test bed to learn a thing or two about audio reproduction so I experiment. I don't have anechoic like behavior but do have suppressed early reflections if you compare to more common in room results. I actually wanted that, strange huh?
I'd have to agree (and think I've stated as much) with Kindhornman here that under the circumstances of suppressed or absent early reflections it would most probably be a Stereo triangle problem that plagues more than one speaker design. It might be a bit different from one speaker to the next and I wouldn't even call it a big problem. Though if we can improve upon it with this super simple trick, why not! I guess you'd first want to know/check how your phase is actually doing at the listening spot to see if you need it at all.
The reason there can be a difference with different speakers is directivity or power response. At least above 900Hz. If the sound reaching your ears in the first 1/2 millisecond or so has a high reverberant content, it could cancel or significantly destroy the comb filtering. That's what the original, pulse based shuffler does.
We haven't experimented (right?) with the amount of delay on the reflection arrival that will subjectively fix the comb filter dip.
I don't think the frequency would change, that would depend on the angle of the stereo triangle. It's the delay between the arrival of left and right sounds that causes the comb filter dips.
Agree. There is a good deal of merit in testing with a single omni-directional mic, but it may miss some important phenomena. Our ears are not omni-directional, most of us have two ears, and they are separated by a skull. That's different enough from a point source mic to be important. The single mic certainly will see comb filtering with two speakers playing, but it won't be the comb filtering you hear with two ears and a head.
FWIW, I've done in ear measurements of my speakers and the combined on and off axis response looks suspiciously like the B&K room curve. Don't know if that's significant, but it's interesting.
I don't know if you remember or read, a few years back, the test a number of us did to look at direct vs reflected sound in our own rooms and systems. It's buried in a long thread somewhere that I can never find.
My Altec system and room had the lowest amount of reflected vs direct sound of the tests posted. Not anechoic by any means, but fairly dry. It was also pretty consistent over a wide bandwidth. No surprise that the dull center combing effect would be noticeable on that system.
Yep, might not be Rome, but all roads seem to lead to this conclusion, which btw is not based on any theory, but simple consumer preference inquiry, the Harmann/ Sean Olive way, or similar, without needing to enter into anecdotic details...
Dummy heads are useful for some things, but not so much for developing loudspeakers IMO. We aren't aiming for a (flat/target) response at the eardrum, just at the point in space where the sound field begins to interact with your head. Any head related functions should not be present in the measurements, unless you are doing something specifically related to the physiology of hearing, for development of binaural audio for instance.
This was just my knee jerk reaction... maybe there is more that can be learned from binaural measurements... they are just really chaotic to look at though, and what we see in those measurements is so "dumbed down" compared to the information our brain extracts. It seems more useful to focus on the sound in space, and take the listener out of the equation, when it comes to measuring speakers and rooms.
It seems like a binaural head mic would be the best way to actually measure the high frequency comb filter effects of center images. You could move the head from side to side a bit, and/or rotate it some, and see what happens with the cancellations. Then using a phase shuffler with two delays, so the cancellations caused by a single shuffler phase delay could be largely filled in my the second shuffler delay, that would have cancellations at different frequencies. A strategic ratio of the two delays might be 1:1.4 or 1.62. That way, the integral multiple cancellations might not double up as much, if at all (I'm too lazy to do the math, but those ratios work with harmonics). Am I getting this right, or did someone say this screws up the in-between images?
Dynomike,
I know when I do normal speaker development that I only use one mic, perhaps that truly is a mistake as we do have two ears and that is what we are really listening to, we don't cover one of our ears. I'm not sure how you would do that without the headform as just placing two bi-n aural mics in an array won't show you what is truly going on here. I actually find this discussion very interesting. Today there are so many multi-channel HT systems that we can't really compare that with a two channel system, the center channel in a multi-channel discrete recording would of course solve that problem but that would discount the majority of recorded music.
I know when I do normal speaker development that I only use one mic, perhaps that truly is a mistake as we do have two ears and that is what we are really listening to, we don't cover one of our ears. I'm not sure how you would do that without the headform as just placing two bi-n aural mics in an array won't show you what is truly going on here. I actually find this discussion very interesting. Today there are so many multi-channel HT systems that we can't really compare that with a two channel system, the center channel in a multi-channel discrete recording would of course solve that problem but that would discount the majority of recorded music.
No, I think the mistake would be to use a dummy head for anything other than investigating head-related phenomena. This phantom center issue is a good candidate for the dummy head though! (I find this discussion interesting too
)A better alternative to an omnidirectional mic for speakers-in-rooms measurements is a tetrahedral ambisonic mic ala Gerzon. Trinnov's system uses this - it can extract a lot more useful information with which to perform corrections.
There is some crossover between these related goals:
1. 2 channel in -> 2 channel out
Improving reproduction of stereo content which appears for whatever reason to have a "dull spot" in between the stereo speakers (on a two-channel setup)
2. discrete center -> phantom center
Reproducing surround content which was mixed with a discrete center speaker on a two-channel setup, aiming to replicate the sound of the physical center speaker. Reproducing mono mixes with a phantom center falls into this category as well.
3. phantom center > discrete center
Improving reproduction of stereo content which appears for whatever reason to have a "dull spot" in between the stereo speakers by upmixing for a physical center speaker
Mostly the thread has revolved around 1, a little of 2... these two projects are closely related but the methods to best achieve each may differ. The 3rd option, upmixing stereo content to a physical center speaker, may well be the most effective but also the least likely to be adopted by two-channel enthusiasts!
I think we all agree that having a >2 channel recording reproduced on >2 speakers (including a center channel) is less problematic than stereo. The techniques discussed here are workarounds for the above 3 situations.
Last edited: