• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

What makes speakers "disappear " and can it be measured?

I don’t need to use imagination. Sounds 1 and 9 clearly come from outside the speaker boundaries. I’ve treated my room heavily and have experimented with left right delay until it was perfect.


In fact most mixes I listen to now it seems like the speakers are merely decorative pieces in the room.

What I mean is that ilusion works as long as I don't concentrate on the speaker. I have all first reflection point treated. 2-8 are all very precisly positioned here, 1 and 9 are not that precise
 
What I mean is that ilusion works as long as I don't concentrate on the speaker. I have all first reflection point treated. 2-8 are all very precisly positioned here, 1 and 9 are not that precise
It’s kind of weird. I like to face the center but use my eyes only to track it. I had some difficulty on the left side but ended up toeing in a little better and fixed it. The major adjustment was time of flight delay on the right monitor. While also using another test track to ensure left and right has the same level at the listening position.

Most mixes don’t have anything coming directly from the speakers. They seem to be decorative pieces in the room.

Also while white is nice looking color for them black helps in the disappearing act. Also having them 12 feet apart and out of my peripheral vision helps a lot.

My setup is an isosceles triangle but with a reduced height. Keeps me closer to the speakers but with a wide soundstage. Helps with SPL and direct/reflected sound.

I still get flutter echo which will be fixed soon with ceiling panels. A floor rug did not help much.
 
Stahl-Sackkarre_pdplarge-mrd--000066891395_PRD_f_all.jpg
 
It’s kind of weird. I like to face the center but use my eyes only to track it. I had some difficulty on the left side but ended up toeing in a little better and fixed it. The major adjustment was time of flight delay on the right monitor. While also using another test track to ensure left and right has the same level at the listening position.

Most mixes don’t have anything coming directly from the speakers. They seem to be decorative pieces in the room.

Also while white is nice looking color for them black helps in the disappearing act. Also having them 12 feet apart and out of my peripheral vision helps a lot.

My setup is an isosceles triangle but with a reduced height. Keeps me closer to the speakers but with a wide soundstage. Helps with SPL and direct/reflected sound.

I still get flutter echo which will be fixed soon with ceiling panels. A floor rug did not help much.

you know this track?
 
I just tried the Pano shuffle for the first time. I had no idea it could be so easily used with rePhase. Wow. I used Ambiophonics before and developed my own mid-side eq based on the ITU measurements in Floyds book. But the way the Pano stuff open up the center between the speakers is unbelievable.
 
I just tried the Pano shuffle for the first time. I had no idea it could be so easily used with rePhase. Wow. I used Ambiophonics before and developed my own mid-side eq based on the ITU measurements in Floyds book. But the way the Pano stuff open up the center between the speakers is unbelievable.
What is the pano shuffle?
 
Code:
 - added "Pano Phase Shuffler" presets in the Paragraphic Phase EQ tab
        * source: http://tinyurl.com/pano-shuffler
        * settings: http://tinyurl.com/pano-shuffler-preset

1665789230732.png


You can create convolution files via rePhase. Try it. If you haven't also try Ambiophonics.

This came out of this thread. https://www.diyaudio.com/community/threads/fixing-the-stereo-phantom-center.277519/

I had neither the time nor the patience to read completely through it.

e: If anyone wants to try this, I attach some files I created. Rephase default settings with exensive optimization, merged left and right files into stereo via SoX. Simpy throw them into your convolver of choice.
 

Attachments

  • pano.zip
    5.2 KB · Views: 82
Last edited:
I tried a method of crosstalk elimination with my four Sony SSCS-5 speakers last night and it really did a great job of making the speakers "disappear." See crude diagram below. I like this arrangement because you can do it with just four bookshelf speakers and a simple 2 channel setup. No DSP or EQ is necessary for good results, although I think a subwoofer is best otherwise you might cancel out the bass too much. I was delighted at the effectiveness of this arrangement in terms of the apparent purity and cleanness of the sound, and the very good width and depth of the sound stage, the ease of keeping my head in the right location by resting it up against the divider wall, and the repositioning of the divider baffle behind my head instead of right smack in my face. Everything seemed to be coming from around and behind the speakers, so even center panned vocals coming straight from the position of the speakers seemed to be singing from behind the speakers, as if the speakers had nothing to do with it. This is even more surprising because the speakers were quite close to my face as I didn't have the means to make this setup very large in this quick and dirty experiment. I could easily reach out and touch the speakers. The sound stage with my eyes closed was wonderfully deep, wide, spacious. I'm reluctant to use the word enveloping but it at least seemed I was on the edge of a very wide open space right ahead of me. One interesting thing I discovered on accident while getting the arrangement dialed in was that even when I was set up wrong so that the speakers behind me were too close, I still perceived the sound as coming from the front. On initial tests using left and right channel test tones the channels sounded reversed! I was hearing the rear channels louder and sooner than the front channels but my brain still perceived the sound as coming from ahead of me. If I got too close to the rear channels it would then become apparent that the sound was coming from behind me, but it seems the precedent effect with forward to back sounds has a strong preference for forward as the perceived direction, at least in this setup. Once I moved the front speakers close enough to me, the left right channels developed nice, wide separation in the correct directions, and the music was really wonderful. I especially like the effect for big orchestral recordings.

So why do this instead of the Polk style arrangement or recursive DSP? To my ears, it's much cleaner sounding. Because of the divider wall it's a pretty clean elimination at each ear with very little residual side effects and colorations. With the Polk system I can get a very wide sound stage but then the crosstalk signal also reaches the ear it isn't supposed to. Fortunately the ear is hearing only the correct channel signal but it's out of phase and time delayed so now there's new comb filtering. With the DSP it loads up the system with signal and costs headroom. It can sound quite good but not as clean and pure, at least not the dsp methods I've tried so far.

I'm sure that with some EQ applied to the rear channels to compensate for HRT and possibly effects from the divider baffle the effect could be enhanced further. But it really sounds good without the DSP! Using front divider baffle arrangements with the baffle right in my face I have succeeded in getting enough separation that hard panned test sounds seemed to be coming from a full 90 degrees to the sides. This rear divider baffle setup only got me to about 70 degrees or so, but that's enough to really sound good.

It is interesting to me that the speakers seem to "disappear" once the direct crosstalk is eliminated. All the cabinet diffraction issues, the grills, early reflections from the walls behind the speakers, the ceiling and floor are still there but don't seem to matter much. The speakers reveal their location when the crosstalk is not eliminated. They disappear when it is eliminated, or at least attenuated to a substantial degree. So it could be that the crosstalk gives our ears a clue to the true source of the sound even though we are also noticing stereo effects. Or it's possible that the rear channels are adding enough confusion in some other way to force my brain in to perceiving the sound as coming from other than the direction of the speaker.

One other thing I'll say about this setup is that it never gave me any perception that the sound sources were actually happening in the room with me, but instead it gave me a sense of hearing into another space. So this was definitely more of a "I am there" experience, or perhaps more like my listening space has developed an opening into that recorded space. I can still hear my space, but I can hear that space really well too.
CrosstalkEliminationSetup.jpg
 
Last edited:
I tried a method of crosstalk elimination with my four Sony SSCS-5 speakers last night and it really did a great job of making the speakers "disappear." See crude diagram below. I like this arrangement because you can do it with just four bookshelf speakers and a simple 2 channel setup. No DSP or EQ is necessary for good results, although I think a subwoofer is best otherwise you might cancel out the bass too much. I was delighted at the effectiveness of this arrangement in terms of the apparent purity and cleanness of the sound, and the very good width and depth of the sound stage, the ease of keeping my head in the right location by resting it up against the divider wall, and the repositioning of the divider baffle behind my head instead of right smack in my face. Everything seemed to be coming from around and behind the speakers, so even center panned vocals coming straight from the position of the speakers seemed to be singing from behind the speakers, as if the speakers had nothing to do with it. This is even more surprising because the speakers were quite close to my face as I didn't have the means to make this setup very large in this quick and dirty experiment. I could easily reach out and touch the speakers. The sound stage with my eyes closed was wonderfully deep, wide, spacious. I'm reluctant to use the word enveloping but it at least seemed I was on the edge of a very wide open space right ahead of me. One interesting thing I discovered on accident while getting the arrangement dialed in was that even when I was set up wrong so that the speakers behind me were too close, I still perceived the sound as coming from the front. On initial tests using left and right channel test tones the channels sounded reversed! I was hearing the rear channels louder and sooner than the front channels but my brain still perceived the sound as coming from ahead of me. If I got too close to the rear channels it would then become apparent that the sound was coming from behind me, but it seems the precedent effect with forward to back sounds has a strong preference for forward as the perceived direction, at least in this setup. Once I moved the front speakers close enough to me, the left right channels developed nice, wide separation in the correct directions, and the music was really wonderful. I especially like the effect for big orchestral recordings.

So why do this instead of the Polk style arrangement or recursive DSP? To my ears, it's much cleaner sounding. Because of the divider wall it's a pretty clean elimination at each ear with very little residual side effects and colorations. With the Polk system I can get a very wide sound stage but then the crosstalk signal also reaches the ear it isn't supposed to. Fortunately the ear is hearing only the correct channel signal but it's out of phase and time delayed so now there's new comb filtering. With the DSP it loads up the system with signal and costs headroom. It can sound quite good but not as clean and pure, at least not the dsp methods I've tried so far.

I'm sure that with some EQ applied to the rear channels to compensate for HRT and possibly effects from the divider baffle the effect could be enhanced further. But it really sounds good without the DSP! Using front divider baffle arrangements with the baffle right in my face I have succeeded in getting enough separation that hard panned test sounds seemed to be coming from a full 90 degrees to the sides. This rear divider baffle setup only got me to about 70 degrees or so, but that's enough to really sound good.

It is interesting to me that the speakers seem to "disappear" once the direct crosstalk is eliminated. All the cabinet diffraction issues, the grills, early reflections from the walls behind the speakers, the ceiling and floor are still there but don't seem to matter much. The speakers reveal their location when the crosstalk is not eliminated. They disappear when it is eliminated, or at least attenuated to a substantial degree. So it could be that the crosstalk gives our ears a clue to the true source of the sound even though we are also noticing stereo effects. Or it's possible that the rear channels are adding enough confusion in some other way to force my brain in to perceiving the sound as coming from other than the direction of the speaker.

One other thing I'll say about this setup is that it never gave me any perception that the sound sources were actually happening in the room with me, but instead it gave me a sense of hearing into another space. So this was definitely more of a "I am there" experience, or perhaps more like my listening space has developed an opening into that recorded space. I can still hear my space, but I can hear that space really well too.
View attachment 239954
I spent some time this weekend listening to this arrangement. While it does a great job of making the speakers disappear and creating a big, deep soundstage, it's extremely sensitive to head movement and speaker placement if you want precise imaging. Playing test tones reveals some weirdness in apparent locations of sounds that move up and down and in and out with pretty subtle head motion. I couldn't get the same exact mirror imaging of left and right when playing a tone through just one channel and then the other. With most actual music and soundtracks it seems far more forgiving but I can't see this catching on in a big way. Something fun to play with though. On the whole, I think the solution I'd most prefer is some kind of high quality up mixing of stereo recordings to 3 or 5 or even 7 channels across the front. That would allow excellent freedom of motion while maintaining a wide soundstage with stable sound sources.
 
I'm trying another setup. This is a really simple 2 to 3 channel up mix. Center channel is just left+right. Left channel is left-right. Right channel is right-left. This works way better than I expected it to. It's incredibly stable moving around the room with the center image always seeming to come from behind the center channel. It doesn't make the speakers seem to disappear in a super magical way, but it's very pleasant. It's like the sound is coming from a large sound field behind the speakers but somehow it's obvious the speakers are involved in creating it. The stereo spread is nice and wide but with very solid center image and good depth. It's a bit thicker sounding but I think I like that for easy listening. Center vocals are really solid. Nothing comes out the side speakers except reverb effects for the center image vocals, so the comb filtering is gone! I think I want to find a way to permanently keep this available. What kind of a digital channel mixer box can do this? Maybe a miniDSP box? (Answering my own question: yes! Channel mixing with inversion is totally a known thing.) Right now I'm making it work by re-mixing stereo tracks in Audacity and listening to them using VLC.
 
Last edited:
You do realize that using "left+right" for the center means that the center output will be twice as loud as either the L or the R. Perhaps you should try (L+R)/2.
 
I'm trying another setup. This is a really simple 2 to 3 channel up mix. Center channel is just left+right. Left channel is left-right. Right channel is right-left. This works way better than I expected it to. It's incredibly stable moving around the room with the center image always seeming to come from behind the center channel. It doesn't make the speakers seem to disappear in a super magical way, but it's very pleasant. It's like the sound is coming from a large sound field behind the speakers but somehow it's obvious the speakers are involved in creating it. The stereo spread is nice and wide but with very solid center image and good depth. It's a bit thicker sounding but I think I like that for easy listening. Center vocals are really solid. Nothing comes out the side speakers except reverb effects for the center image vocals, so the comb filtering is gone! I think I want to find a way to permanently keep this available. What kind of a digital channel mixer box can do this? Maybe a miniDSP box? (Answering my own question: yes! Channel mixing with inversion is totally a known thing.) Right now I'm making it work by re-mixing stereo tracks in Audacity and listening to them using VLC.
This is very interesting. It’s reminds me of the old days we would derive a L-R signal by simply connecting the non ground wires to a speaker. That gives you the difference signal.

It’s not as precise as what you are doing in software but sounds similar.

L+R is simply connecting both L and R together with ground on the other terminal.

Now watch out I think this can blow up some amps.


 
This is very interesting. It’s reminds me of the old days we would derive a L-R signal by simply connecting the non ground wires to a speaker. That gives you the difference signal.
That was the Dynaco/Hafler setup for a pseudo-surround effect.
 
You do realize that using "left+right" for the center means that the center output will be twice as loud as either the L or the R. Perhaps you should try (L+R)/2.
Yes I realized the summing issue, and it caused some clipping on some tracks so I resorted to a minus 3dB attenuation before mixing. I actually do the minus 3dB on all the tracks before mixing. Interestingly the center image is not too strong. I actually turned down the side channels by 3dB in the receiver settings to make the center a little louder. I also tried various delays on the center channel, and setting the speakers up all in a row vs in a U-shaped arc so they were equal distance to my ears. I think it is optimal at equal distance to each ear, in which case you get some cross talk cancelation benefit. But it's not critical for enjoyable sound. This is a robust setup that's very tolerant of speaker position and listening position. It may not be accurate in terms of stereo spread but it's pleasant and clear enough on all kinds of music, and anything with a center imaged vocal or instrument is really quite a relief to have stabilized. I'm using a subwoofer too. Another thing this setup does is remove all the mono bass from both side channels.
 
This is very interesting. It’s reminds me of the old days we would derive a L-R signal by simply connecting the non ground wires to a speaker. That gives you the difference signal.

It’s not as precise as what you are doing in software but sounds similar.

L+R is simply connecting both L and R together with ground on the other terminal.

Now watch out I think this can blow up some amps.


Yes, I was thinking about trying just hooking up the speaker wires but I worried the amps might start fighting each other. Even mixing at line level seems potentially problematic if the op-amps are outputting into each other. Maybe they're not too affected by that. It'd be nice if it could be done with simple passive line level connectors. Doing it digitally seems like the cleanest approach. I had read about these kinds of arrangements before but my thoughts were that they'd just muddle up the sound with all the reverse phase added "noise". Now I understand it more as a compromise with very real benefits. Depending on the listening situation the pros can outweigh the cons. I really would like to hear a more sophisticated up mixing system that actually pulls the center channel out of each side channel without adding all the reverse phase information from the opposite channel at the same time. It seems there's no simple way to do that.
 
Last edited:
Hello @Tim Link,

I have been following/reading your recent posts with great interests since I have thought about, but not actually done, the similar multichannel experiments a few years ago with my rather flexible PC-DSP-based multichannel multi-amplifier multi-SP-driver setup.

One of the reasons for not going into "your approach/experiments" in my case is/was that I could well achieve the wonderful "disappearance" of SPs by perfect L&R SPL balance with 0.1 msec scale precision time alignment among all the SP drivers using the group delay functionalities of DSP EKIO, and I could develop my own primitive but reliable reproducible accurate objective time alignment measurement methods (microphone recording only by separate PC) on actual room air sound even at my listening position. You would please refer to my summary post here.

Let me simply ask you, therefore, whether have you achieved complete (0.1 msec precision) time alignment among all the SP drivers or not, and if your answer would be yes, then what would be the favorable results and subjective impression of "the precision time alignment" in comparison with your current multichannel reverse polarity experiments?
 
Hello @Tim Link,

I have been following/reading your recent posts with great interests since I have thought about, but not actually done, the similar multichannel experiments a few years ago with my rather flexible PC-DSP-based multichannel multi-amplifier multi-SP-driver setup.

One of the reasons for not going into "your approach/experiments" in my case is/was that I could well achieve the wonderful "disappearance" of SPs by perfect L&R SPL balance with 0.1 msec scale precision time alignment among all the SP drivers using the group delay functionalities of DSP EKIO, and I could develop my own primitive but reliable reproducible accurate objective time alignment measurement methods (microphone recording only by separate PC) on actual room air sound even at my listening position. You would please refer to my summary post here.

Let me simply ask you, therefore, whether have you achieved complete (0.1 msec precision) time alignment among all the SP drivers or not, and if your answer would be yes, then what would be the favorable results and subjective impression of "the precision time alignment" in comparison with your current multichannel reverse polarity experiments?
.1 millisecond precision between drivers corresponds to about 1" of distance at the speed of sound. I have horns that work from 600Hz up to 20kHz so that entire range can come from a point source. But there are still diffraction effects around the mouth of the horn so it's not a perfectly clean time alignment. Once you have multiple drivers you can't get the perfect time alignment at every location so it ends up being a head in a vice situation. Cross talk cancellation is very finicky about precise listening location and speaker setup too. If you aren't doing crosstalk cancellation, you're hearing crosstalk no matter how accurate everything is. If you like that or can hear through it you're good to go. I prefer not hearing it even though the sound can still be very enjoyable with it and just two channel stereo. What causes speakers to disappear is the big question. I have not found that it requires excellent time alignment between drivers or extreme balance precision and speaker placement precision to make it happen, although balance is very important for a balanced sound field. It doesn't seem to require low diffraction cabinets either. And it doesn't even mean the sound is great or that the imaging is great. It just means that for whatever reason the impression becomes that the speakers have nothing to do with generating the sound you are hearing.

So even though I can't say I've achieved perfect phase response and time alignment even at one single location, the potential benefit of these reverse phase multiple speaker arrangements are minimizing of crosstalk at your ears, which reduces comb filtering and thereby improves tonality and imaging. The downside to this is that you have to sit in the exact right spot for the cancellation to work, just like you'd have to sit in the exact right spot for all the drivers to time align at your ears.

If you're not going for crosstalk cancellation but instead going for 3 channel up mix, the benefit is that the center image can remain stable and free of the two channel comb filtering as you move around the room because the center image signal is removed from the left and right speakers and played solely through the center speaker.

No matter how accurately you do it, two channel playback has some inherent issues, so that's the reason for trying out all these methods. It's to find out what they sound like and see if I prefer it or not. Ever since I heard two channel playback with crosstalk eliminated I've not been fully satisfied with traditional two channel playback that allows the sound from both speakers to hit both ears simultaneously with slight time delay when trying to create a center image. A real center image should send one stream of sound to each ear, not two signals with slight time delay and different HRTF, neither of which correspond to a sound coming from straight ahead.

With standard 2 channel playback the only time you won't hear direct comb filtering between the speakers is when the signal is hard panned to the left or right speaker. Those will be the situations where the clarity and coherence will be the highest. With the three channel setup it's the opposite. The greatest imaging clarity is in the center of the sound stage. I think this sounds better and makes more sense. Imagine a camera lens that is blurry in the middle and sharp around the edges of the picture vs a lens that is sharp in the middle and blurry around the edges. Some may prefer the sharper around the edges but for most pictures I don't.
 
Last edited:
Back
Top Bottom