I recently passed on a Sony soundbar in my local thrift, which had plain, two pin right / left / center connections. I didnt like that the grille wouldnt come off so I could see the drivers... (I usually wont buy a used speaker that I cant see the driver's cones - help keep the junk pile minimized)
In retrospect, it would have been a cheap vehicle for experimenting with these connections...not that I dont have others sets of 3's to try.
I'm pretty sure there's more tricks than just that in the recording engineer's toolbox...
Cool thread. Tough subject. Little conclusive agreement. I moved to 3.1 after researching Paul Klipsch's center channel notions, Gerzon's work on Ambisonics and prior, Dave Greisinger's surrounds for Lexicon/HK, Michael Barton's Trifield implementations for Meridian, and my own experience with Dolby stuff. I use Dolby PLIIx (from an old $100 Dolby DP564) too in a 3.1 system and often prefer that... so I largely agree with ASCTim. One has to be willing to accept digital signal processing in your music to use this, and I do.
At least for now, I've done the same.
Folks may have noticed that I haven't been posting much lately, and a big part of that has been that 3.1 just works.
Some background:
1) Twenty years ago, I spent fifteen hours a week commuting to work, and I was quite active in car stereo. I became interested in doing three channel, based on an obscure PhD paper that I found, where the author proposed using a single tweeter in the center of the soundstage, and omitting the left and the right tweeters entirely. This was particularly compelling for car stereo, because in car stereo we're often putting speakers up on the dash. So cutting that down from two to one speaker has a lot of benefits. Most importantly, it's very difficult to get a good soundstage in a car without DSP or chopping up the car. And car DSPs in 2003 weren't affordable or very good.
2) That experience note above led me to Opsodis, which is basically just an evolution of the same ideas. https://www.diyaudio.com/community/threads/cbt-with-crosstalk-cancellation.405434/
3) When re-evaluating Opsodis this year, for the first time in ages, all of the simulations looked atrocious. It looked like you basically had to keep your head in a vise, and moving your head even 10cm would signficantly screw up the high frequencies. Another issue with Opsodis is that it's a lot like taking a very very tall multiway array and flipping it on it's side; you wind up with extremely narrow horizontal beamwidth and wide vertical beamwidth... the last thing we could possibly want.
But I'd had good experiences with Opsodis in the past, so I figured "might as well give this a go."
What I ended up with is basically a crude 3.1 setup... that somehow works pretty great.
If you look at the matrix settings for Dolby Prologic II, all of the processing can be done with a $80 MiniDSP:
https://www.diyaudio.com/community/threads/diy-prologic-ii.308266/
There's absolutely nothing "fancy" here. The center channel is mono. The left and the right channel are still the left and the right channel, but their level is attenuated. The only thing that's particularly novel is that the surrounds are reversed and out-of-phase and attenuated.
I did make one significant change, and I think this change is The Secret Sauce:
In a movie theater, up until the last few years, the speakers were mostly behind the screen. And I think this is how the vast majority of people set up their home theaters too. (In the last 10-15 years, theaters have begun to put speakers all around the room, I'm about to get to that...)
In the Opsodis literature, the speakers form a semi-circle around you.
From everything I can find, Opsodis seems to be kaput, and I'm willing to bet that part of the issue is that they never seemed to figure out if they were in the business of selling sound bars or selling processing software.
Perhaps I'm missing something, but for me, the big breakthrough with my funky 3.1 channel setup was bringing the left and the right speakers much much closer. I started out with the left and right speakers about one foot forward of the center channel, and what I found was that I could bring the left and right closer and closer and closer. And you'd expect the sound stage would get really weird or too close, but that wasn't the case at all. As of today, I have the left and right speakers immediately to my left and to my right, and the center directly in front of me.
This next part is super important, and it's another 'trick' I learned from car audio. Basically, you MUST use DSP to get all three speakers to arrive at the same time. In car audio, you can use DSP to move the soundstage to left and to the right, and even a time delay of a fraction of a millisecond is audible. In MiniDSP, you can literally increase and decrease the delay to one of two speakers, and hear how the stage shifts accordingly.
With this 3.1 setup, this is absolutely necessary. Unless, of course, you have the freedom to place the speakers exactly equidistant. And I'm not exaggerating here, a difference of even 5-10cm is audible.
This next idea might be controversial, but here goes:
Obviously, I've long been fascinated by the Danley speakers. I once rented an SH-50, and one of the really peculiar aspects of that speaker is that it's not easy to tell where the soundstage begins and ends. I've told this one a hundred times: I had the SH-50s playing at a fairly high volume, and our daughter was seated on the couche, literally less than a meter away. At one point she asked "is it on?" IE, the origin of the sound in an SH-50 is so nebulous, she was getting blasted by one less than a meter away and she couldn't figure out if the sound was coming from the speaker or somewhere else. Remarkable.
The reason that the SH-50 can do this is because it's possible to produce a wavefront that is IN PHASE even if the loudspeakers aren't physically located at the same point in space. Ideally you'd perform this trick inside of a big horn, like the SH-50. But you can get "in the ball park" using conventional speakers and delay.
End fire and gradient subwoofer arrays work off of this concept. But the ideas work at all frequencies; there's nothing stopping you from making an end fire midrange array.
In the current implementation of my modest 3.1 setup, I'm leveraging these ideas, by placing the left and right speakers closer than the center and then "fixing" the pathlength difference with DSP delay. YMMV but I find the effect to be pleasant; it makes the depth of the stage more nebulous. Because DSP can "fix" pathlength issues, but the sound is still coming from three speakers that AREN'T equidistant. At some point I might move the speakers so that they ARE equidistant, but if memory serves, when I did that, the sound stage depth wasn't as good. Obviously this is a subjective thing, because you're trading pinpoint imaging for a deeper soundstage.
A few additional observations and things on my "todo" list:
1) Now that I have one speaker in front of me, one to the left, and one to the right, the idea of having five channels across the front starts to look compelling. The math for that is a lot trickier, I don't think a four channel miniDSP can do it. It might be possible with two. It's been a long time, but I vaguely recall that the folks at MiniDSP created the plugin I'm using in response to my car audio experiements, and I have no idea if that plugin will work on an eight channel MiniDSP. I don't think so, so one would have to find TWO four channel MiniDSPs to pull this off (and they're no longer for sale.)
2) I have a friend that does mastering for movies, and he's a big fan of Penteo: https://www.perfectsurround.com/ IIRC, this is what he uses in his car, and listening to mixes that he made in both stereo and multichannel were a big inspiration for me
3) The more that I mess around with multichannel, the more that I get curious about using speakers with a narrow beamwidth. This is subjective, of course, but I think that in a stereo setup we tend to select speakers which produce a lot of reflections off of the sidewalls, because it makes the stage sound bigger. And this concept is perfectly reasonable if it's 1980 or even 2000, when amplification and processing were relatively expensive and simple. But it's 2023 now and amplifiers are practically free, they're so cheap, and DSP has come a loooong way. I think in a lot of ways we can create a soundstage electronically and once we do that, narrow beamwidth speakers might be better (since we don't want these speakers interacting with each other as much.)
4) Another frontier that's worth a look is using AI to split the tracks. This is kinda advanced stuff, but it's fairly trivial to use AI to split a stereo track into "stems" and then send the stems to discrete speakers. Spleeter is free and works great under Linux. If one is splitting stereo tracks, then you have to figure out a way to encode them. This would be the opportunity to take an entire music library and re-encode it from wav or flac into a surround format. There are dozens, and some support something like a hundred channels. One cool thing about going this route is that you can make your library completely portable. IE, the "old school" way of doing music upsampling was to plug your stereo source into a home theater receiver, and then configure Dolby Prologic II to produce 3.1 or 5.1. I'm not a big fan of this AT ALL. First off, I like tinkering with amplifiers, and home theater receivers aren't something I get excited about. Even worse, I find that home theater receivers just LOVE to randomly lose all of their settings. At my house I have two identical home theater receivers in two different rooms, and I'm constantly finding that some setting got borked or someone clicked a button they weren't supposed to, and due to that, all of my careful measurement and adjustments just went right out the window. I HATE THAT. Sometimes it takes 2-4 hours to "dial in" all the settings, and losing them because someone hit some random button is infuriating. I very much prefer to make all the changes in the audio file itself, and making any last minute tweaks in MiniDSP, basically removing the home theater receiver from the equation entirely.
Folks may have noticed that I haven't been posting much lately, and a big part of that has been that 3.1 just works.
Some background:
1) Twenty years ago, I spent fifteen hours a week commuting to work, and I was quite active in car stereo. I became interested in doing three channel, based on an obscure PhD paper that I found, where the author proposed using a single tweeter in the center of the soundstage, and omitting the left and the right tweeters entirely. This was particularly compelling for car stereo, because in car stereo we're often putting speakers up on the dash. So cutting that down from two to one speaker has a lot of benefits. Most importantly, it's very difficult to get a good soundstage in a car without DSP or chopping up the car. And car DSPs in 2003 weren't affordable or very good.
2) That experience note above led me to Opsodis, which is basically just an evolution of the same ideas. https://www.diyaudio.com/community/threads/cbt-with-crosstalk-cancellation.405434/
3) When re-evaluating Opsodis this year, for the first time in ages, all of the simulations looked atrocious. It looked like you basically had to keep your head in a vise, and moving your head even 10cm would signficantly screw up the high frequencies. Another issue with Opsodis is that it's a lot like taking a very very tall multiway array and flipping it on it's side; you wind up with extremely narrow horizontal beamwidth and wide vertical beamwidth... the last thing we could possibly want.
But I'd had good experiences with Opsodis in the past, so I figured "might as well give this a go."
What I ended up with is basically a crude 3.1 setup... that somehow works pretty great.
If you look at the matrix settings for Dolby Prologic II, all of the processing can be done with a $80 MiniDSP:
https://www.diyaudio.com/community/threads/diy-prologic-ii.308266/
There's absolutely nothing "fancy" here. The center channel is mono. The left and the right channel are still the left and the right channel, but their level is attenuated. The only thing that's particularly novel is that the surrounds are reversed and out-of-phase and attenuated.
I did make one significant change, and I think this change is The Secret Sauce:
In a movie theater, up until the last few years, the speakers were mostly behind the screen. And I think this is how the vast majority of people set up their home theaters too. (In the last 10-15 years, theaters have begun to put speakers all around the room, I'm about to get to that...)
In the Opsodis literature, the speakers form a semi-circle around you.
From everything I can find, Opsodis seems to be kaput, and I'm willing to bet that part of the issue is that they never seemed to figure out if they were in the business of selling sound bars or selling processing software.
Perhaps I'm missing something, but for me, the big breakthrough with my funky 3.1 channel setup was bringing the left and the right speakers much much closer. I started out with the left and right speakers about one foot forward of the center channel, and what I found was that I could bring the left and right closer and closer and closer. And you'd expect the sound stage would get really weird or too close, but that wasn't the case at all. As of today, I have the left and right speakers immediately to my left and to my right, and the center directly in front of me.
This next part is super important, and it's another 'trick' I learned from car audio. Basically, you MUST use DSP to get all three speakers to arrive at the same time. In car audio, you can use DSP to move the soundstage to left and to the right, and even a time delay of a fraction of a millisecond is audible. In MiniDSP, you can literally increase and decrease the delay to one of two speakers, and hear how the stage shifts accordingly.
With this 3.1 setup, this is absolutely necessary. Unless, of course, you have the freedom to place the speakers exactly equidistant. And I'm not exaggerating here, a difference of even 5-10cm is audible.
This next idea might be controversial, but here goes:
Obviously, I've long been fascinated by the Danley speakers. I once rented an SH-50, and one of the really peculiar aspects of that speaker is that it's not easy to tell where the soundstage begins and ends. I've told this one a hundred times: I had the SH-50s playing at a fairly high volume, and our daughter was seated on the couche, literally less than a meter away. At one point she asked "is it on?" IE, the origin of the sound in an SH-50 is so nebulous, she was getting blasted by one less than a meter away and she couldn't figure out if the sound was coming from the speaker or somewhere else. Remarkable.
The reason that the SH-50 can do this is because it's possible to produce a wavefront that is IN PHASE even if the loudspeakers aren't physically located at the same point in space. Ideally you'd perform this trick inside of a big horn, like the SH-50. But you can get "in the ball park" using conventional speakers and delay.
End fire and gradient subwoofer arrays work off of this concept. But the ideas work at all frequencies; there's nothing stopping you from making an end fire midrange array.
In the current implementation of my modest 3.1 setup, I'm leveraging these ideas, by placing the left and right speakers closer than the center and then "fixing" the pathlength difference with DSP delay. YMMV but I find the effect to be pleasant; it makes the depth of the stage more nebulous. Because DSP can "fix" pathlength issues, but the sound is still coming from three speakers that AREN'T equidistant. At some point I might move the speakers so that they ARE equidistant, but if memory serves, when I did that, the sound stage depth wasn't as good. Obviously this is a subjective thing, because you're trading pinpoint imaging for a deeper soundstage.
A few additional observations and things on my "todo" list:
1) Now that I have one speaker in front of me, one to the left, and one to the right, the idea of having five channels across the front starts to look compelling. The math for that is a lot trickier, I don't think a four channel miniDSP can do it. It might be possible with two. It's been a long time, but I vaguely recall that the folks at MiniDSP created the plugin I'm using in response to my car audio experiements, and I have no idea if that plugin will work on an eight channel MiniDSP. I don't think so, so one would have to find TWO four channel MiniDSPs to pull this off (and they're no longer for sale.)
2) I have a friend that does mastering for movies, and he's a big fan of Penteo: https://www.perfectsurround.com/ IIRC, this is what he uses in his car, and listening to mixes that he made in both stereo and multichannel were a big inspiration for me
3) The more that I mess around with multichannel, the more that I get curious about using speakers with a narrow beamwidth. This is subjective, of course, but I think that in a stereo setup we tend to select speakers which produce a lot of reflections off of the sidewalls, because it makes the stage sound bigger. And this concept is perfectly reasonable if it's 1980 or even 2000, when amplification and processing were relatively expensive and simple. But it's 2023 now and amplifiers are practically free, they're so cheap, and DSP has come a loooong way. I think in a lot of ways we can create a soundstage electronically and once we do that, narrow beamwidth speakers might be better (since we don't want these speakers interacting with each other as much.)
4) Another frontier that's worth a look is using AI to split the tracks. This is kinda advanced stuff, but it's fairly trivial to use AI to split a stereo track into "stems" and then send the stems to discrete speakers. Spleeter is free and works great under Linux. If one is splitting stereo tracks, then you have to figure out a way to encode them. This would be the opportunity to take an entire music library and re-encode it from wav or flac into a surround format. There are dozens, and some support something like a hundred channels. One cool thing about going this route is that you can make your library completely portable. IE, the "old school" way of doing music upsampling was to plug your stereo source into a home theater receiver, and then configure Dolby Prologic II to produce 3.1 or 5.1. I'm not a big fan of this AT ALL. First off, I like tinkering with amplifiers, and home theater receivers aren't something I get excited about. Even worse, I find that home theater receivers just LOVE to randomly lose all of their settings. At my house I have two identical home theater receivers in two different rooms, and I'm constantly finding that some setting got borked or someone clicked a button they weren't supposed to, and due to that, all of my careful measurement and adjustments just went right out the window. I HATE THAT. Sometimes it takes 2-4 hours to "dial in" all the settings, and losing them because someone hit some random button is infuriating. I very much prefer to make all the changes in the audio file itself, and making any last minute tweaks in MiniDSP, basically removing the home theater receiver from the equation entirely.
Attachments
One would think this to be the darling of the home decor, where you can put speakers anywhere (within reason) and "tune out" pathlength differences, relative to a preferred listening position, using DSP. Perhaps one day in a multichannel amp you'll see a delay control with a center detent, where you can adjust + or - in feet of distance, next to the level control, for each channel.
I'm pretty sure my Zoudio amps have delay, but it works out to be something like a generous 6". When they included that in the DSP part of the chip's design, I assume they were thinking in baffle board dimensions. Since there's plenty of room for transistors on a die these days, one would think the idea could be extended to feet (meters) worth of delay easily. "But why would anyone want that capability?"
Re. "3D localization" > There is more than one kind.
The highest quality of 2 Ch. stereo stipulates: Height, Width & Depth. However, all that comes from in front of you
and nothing from behind you. In stereo, the closest thing resembling 'phantom space' comes from signal/content
that is out of phase left to right. (I guess a major part of this whole thread)
The other kind of "3D localization" comes from discrete digital channels that can have sound coming from behind you,
and even to the sides of you.
CAN YOU SEE THE DIFFERENCE ???
Just as so many people comment on > It is the mixing/production of the music that ultimately determines what you hear.
There are near perfect standards for Movie production, but that doesn't apply for Music in 5.1 or 7.2
So, searching for a single perfect solution for the Music of your surround system is essentially futile.
That is why I say, buy the very best AV amplifier you can afford, and use the numerous 'presets' to your advantage.
A lot of research goes into some of todays products
Last edited:
Opsodis! I've been looking for that for years now. I remember seeing pictures of their setups with the high frequency drivers closer and the lower frequencies out wider, but I didn't know what it was called or who was promoting it. Thanks! I experimented with making my tweeters closer together and my midwoofers further apart because the cancellation looked like it'd work out better. For whatever reason I didn't like the way it sounded. I also tried it with a 5 speaker array, with the outer speakers playing lower frequencies, inside speakers playing higher, and center speaker playing full range using the little Sony bookshelf speakers. There were some good things about that but overall I couldn't beat the purity and balance of just sticking to 3 speakers for all the crosstalk reduction. Below 220 Hz I revert to regular stereo with the corner horns, so I guess it really is a 5 speaker array I'm using.
Some people might find something of interest in this Google 'thread'.
https://www.google.com/search?sca_e...EA_4Q5t4CegQIGRAB&biw=1318&bih=654&dpr=1#ip=1
https://www.google.com/search?sca_e...EA_4Q5t4CegQIGRAB&biw=1318&bih=654&dpr=1#ip=1
I recently made a big improvement to this 3 speaker array.
The side channels need to be EQd!
I always knew and had noticed it in measurements, but I didn't think EQing the side speakers without the center was legit because they wouldn't all match perfectly for optimal crosstalk reduction. I finally got smart and tried it again. I took side panned and center panned measurements at the left and right ear position. On a left panned signal there are thee speakers playing, and they sum to louder than they should at some frequencies. Specifically, it's a broad peak around 4000 Hz. A -5 PEQ on the side channels at 4000 and about 1.5 octave width did the trick.
The stereo width remains, and the tone is now even across the soundstage. That brightness to the sides could be airy and scintillating at times, but overall it's much better this way.
So simple, and yet it took me so long.
That's now one less objection I can raise to the sound of this setup. The soundstage remains wide and detailed and now is enhanced with beauty of better tonal balance. It's really lovely. The better balance brings out even more detail. The smootheness combined with the width and clarity is sweet and captivating.
The total system still has some noticeable tonal flaws, but I have now a better view of how to get closer to ideal.
The side channels need to be EQd!
I always knew and had noticed it in measurements, but I didn't think EQing the side speakers without the center was legit because they wouldn't all match perfectly for optimal crosstalk reduction. I finally got smart and tried it again. I took side panned and center panned measurements at the left and right ear position. On a left panned signal there are thee speakers playing, and they sum to louder than they should at some frequencies. Specifically, it's a broad peak around 4000 Hz. A -5 PEQ on the side channels at 4000 and about 1.5 octave width did the trick.
The stereo width remains, and the tone is now even across the soundstage. That brightness to the sides could be airy and scintillating at times, but overall it's much better this way.
So simple, and yet it took me so long.
That's now one less objection I can raise to the sound of this setup. The soundstage remains wide and detailed and now is enhanced with beauty of better tonal balance. It's really lovely. The better balance brings out even more detail. The smootheness combined with the width and clarity is sweet and captivating.
The total system still has some noticeable tonal flaws, but I have now a better view of how to get closer to ideal.