Stereo Crosstalk Elimination (reduction) Par Excellence!

[LIΟN]

For envelopment, the suggested angle is 110. I am not sure of the delay, you are using but with various multiple IRs that ranges from 0.5 to 2s, I find any delay of more than 5 ms is audible as echo. Try using clicks to time align them. Depending on the hardware a 20ms setting may not be 20ms. You have measure at your ears to see the exact delay. I find it easier to time align using clicks rather than measurements.

Yes. Since I manually align all the peaks to zero in loopback and REW and I align arbitrarily (final crosstalk channels to original ear delay based on angle and distance of my files), there is no issue with impulse alignment. Thank you for your concern.

 

[tjcinnamon]

Here is a good read for those care to run all of your sources through your computer https://archimago.blogspot.com/2023/10/stereo-crosstalk-cancellation-xtc.html

There is a DAC with an XTC plugin, the Weiss DAC 5xx series. See section 2.3.10 in the manual https://weiss.ch/wp-content/uploads/2022/07/dac50x-userman.pd

Here is my email reply to the archimago blog: "I had perused the manual when I first got the unit. I had no interest in most of those plug-ins, and it was my ignorance that caused me to ignore the others. Your article was the incentive that I needed. The XTC system was so very easy to use. It did give a better defined soundstage width and depth. Movie dialog is now much more legible. This one preset is active on all inputs. I do avoid computer based applications, because the assumption is that the computer is the only source. Many people complain that the Weiss DACs are overpriced. I say not true. Their DACs include the computer, with many plug-ins already available and in a very accessible fashion. I never stop learning. Thanks!!"

I do continue to read threads like this and other threads, but I no longer have the energy to do all of that work.

@STC I just read through that article. Very cool. I'm looking for a way to have an external box that can use those plugins with consistent delay. I want to run it through my AVR/AppleTV and not just through a PC. Is it possible to have an external box (not that super expensive DAC)? Perhaps running RCA through RaspberryPi to an amp?

@Tim Link how far outside the plane of the speakers is the soundstage? I'm looking at this for a home theater application. As TV's get larger it's difficult to get the effect. My Polk L800's provide it but with my 83" TV it's a stretch and not fully optimal. I figured I could run your config as a "center" channel with the summed LR and basically run a 6.4.4 instead of a 7.4.4

I was thinking of the KEF Q150's because I can symmetrically lay on their side.

 

[tjcinnamon]

Yes. Since I manually align all the peaks to zero in loopback and REW and I align arbitrarily (final crosstalk channels to original ear delay based on angle and distance of my files), there is no issue with impulse alignment. Thank you for your concern.

what is your setup? Are you using 3 speakers and what plugins/signal chain are you using? Is it similar to Tim's? L-R L+R R-L?

 

[Tim Link]

@STC I just read through that article. Very cool. I'm looking for a way to have an external box that can use those plugins with consistent delay. I want to run it through my AVR/AppleTV and not just through a PC. Is it possible to have an external box (not that super expensive DAC)? Perhaps running RCA through RaspberryPi to an amp?

@Tim Link how far outside the plane of the speakers is the soundstage? I'm looking at this for a home theater application. As TV's get larger it's difficult to get the effect. My Polk L800's provide it but with my 83" TV it's a stretch and not fully optimal. I figured I could run your config as a "center" channel with the summed LR and basically run a 6.4.4 instead of a 7.4.4

I was thinking of the KEF Q150's because I can symmetrically lay on their side.

My speaker array is 3 feet wide. I'm about 7 feet back from them. The stereo sound stage is about 12 feet wide, which I just now determined by playing correlated pink noise and volume panning it hard left and right. I've heard some music get a bit wider than that, more like 20 feet, but I think that requires some phase differences between channels baked into the recording. Some special effects will once in a while give me impressions of stuff completely off to the sides or even behind me, but not very often.

 

[tjcinnamon]

My speaker array is 3 feet wide. I'm about 7 feet back from them. The stereo sound stage is about 12 feet wide, which I just now determined by playing correlated pink noise and panning it hard left and right. I've heard some music get a bit wider than that, more like 20 feet, but I think that requires some phase differences baked into the recording. Some special effects will once in a while give me impressions of stuff completely off to the sides or even behind me, but not very often.

Give Radiohead's Kid-A (song not the album) a shot. My L800's are 7' apart but the sound is about 6' to 7' out the side of each speaker. So it's about a 21' sound stage but I know there's some distortion where if my head is not perfect one of the speakers will be cancelled over the other.

Another song is Yosi Horikawa's - Kingdom of Frogs (Diasuke remix).

It's sub optimal but I can't imagine going back to normal speakers. So I'm trying to find a solution that offers crosstalk within the physical constraints of a 80+ inch TV.

Those songs are extremely effective examples.

Crazy XTC plus Dirac ART is what I'm really after. A cancellation fest!

edit: any ideas how I could do this with hardware so I could make something work with my AVR? I was thinking a MOTU ultra-light.

 

[STC]

I just read through that article. Very cool. I'm looking for a way to have an external box that can use those plugins with consistent delay. I want to run it through my AVR/AppleTV and not just through a PC. Is it possible to have an external box (not that super expensive DAC)? Perhaps running RCA through RaspberryPi to an amp?

Thanks for the kind words.

MiniDSP used to sell MiniAmbio that can use as a one box solution. They also incorporated AmbiophonicsDSP via MiniDSP but due to sampling rate difference the plug-in no longer works for the HD version.

You can take analogue output and connect to a Motu interface and do the crosstalk cancellation. I use Motu8A . The other solution is to use Polk SDA speakers like their Legend series.

 

[tjcinnamon]

Thanks for the kind words.

MiniDSP used to sell MiniAmbio that can use as a one box solution. They also incorporated AmbiophonicsDSP via MiniDSP but due to sampling rate difference the plug-in no longer works for the HD version.

You can take analogue output and connect to a Motu interface and do the crosstalk cancellation. I use Motu8A . The other solution is to use Polk SDA speakers like their Legend series.

Of course! I respect your work!

I have the original MiniDSP 2x4 but I’m concerned about the poor sampling rate. I did buy the plugin though.

I also, have the Polk L800’s I’ve got to experiment by combining to the miniDSP and the L800’s

So great minds think alike! I’ll try that Motu. My L800’s are too far apart but still “work”. My concern is when I switch to a 98” TV and they have to be even further apart.

Wonderful speakers though.

 

[STC]

Of course! I respect your work!

I have the original MiniDSP 2x4 but I’m concerned about the poor sampling rate. I did buy the plugin though.

I also, have the Polk L800’s I’ve got to experiment by combining to the miniDSP and the L800’s

So great minds think alike! I’ll try that Motu. My L800’s are too far apart but still “work”. My concern is when I switch to a 98” TV and they have to be even further apart.

Wonderful speakers though.

No No No….I just implemented what Ralph did. He guided me and taught everything that is needed about human hearing, preference and crosstalk. He wanted it to be free for all and hoped someday a developer could come up a one box solution for the Domestic Concert Hall.

Best XTC is with speakers in Ambiopole position. When speakers are too far apart crosstalk cancellation can only be achieved with phase manipulation. I used MiniAmbio with HT but my TV is small at 65 inch. Having said that, XTC for TV or HT is not practical if it involves more than one person.

 

[tjcinnamon]

No No No….I just implemented what Ralph did. He guided me and taught everything that is needed about human hearing, preference and crosstalk. He wanted it to be free for all and hoped someday a developer could come up a one box solution for the Domestic Concert Hall.

Best XTC is with speakers in Ambiopole position. When speakers are too far apart crosstalk cancellation can only be achieved with phase manipulation. I used MiniAmbio with HT but my TV is small at 65 inch. Having said that, XTC for TV or HT is not practical if it involves more than one person.

Isn’t there a configuration which can be done with 4 speakers? If so, what is that called? Perhaps that would work in a wider configuration?

 

[STC]

Isn’t there a configuration which can be done with 4 speakers? If so, what is that called? Perhaps that would work in a wider configuration?

For stereo, you can use the same signal and do the rear XTC. It doesn’t increase the width but adds envelopment. If you have quad or 4.0 format, then the rear XTC can create 360 sound. IME, listeners still prefer the rear speakers turned on even with standard 2.0 format. Ralph doesn’t know the reason and as far as he knew only two of us had it implemented. you can find AES papers how this can be corporates with 5.1 or other multichannel system.

This is Farina’s paper.

http://www.angelofarina.it/Public/Papers/204-AES118.pdf

 

[LIΟN]

what is your setup? Are you using 3 speakers and what plugins/signal chain are you using? Is it similar to Tim's? L-R L+R R-L?

Nope.
Since I already have respective binaural responses to L-L, L-R, R-L, and R-R, I'm manually creating an XTC filter and listening to it as an ideal XTC at the desired angle.
Recently, my most satisfying and effective filter generation method is applying arithmetic to AlignSum and FL-L (both with data up to approximately 5ms of the direct shaded area), and based on that, I've been listening to Crosstalk channels such as FL-R and FR-L.
It has almost the same tone balance as the original, and only unnecessary crosstalk (damaging stereo images) is removed.
The advantage of this filter generation is that you get the best, most personalized XTC for that angle.

 

[tjcinnamon]

All three speakers in the center array run 220 to 20,000. The 10" woofers do 220 to 1100Hz, the tweeters 1100Hz and up. I've found it's best to keep them all the same. I've tried a lot of things, including changing the volume of the center compared to the sides, and EQing the sides different than the center. I think it's fundamentally more correct to run them all the same. That's what my ears are telling me.

What runs the 220Hz and below?

If I were to try this with three KEF150 bookshelves would this work?


L-R
220Hz -> 10000Hz

L+R
Full range

R-L
220Hz -> 10000Hz

Then I'd use the AVR to crossover to my subs at whatever measures best (likely between 80 and 100Hz)

 

[Tim Link]

What runs the 220Hz and below?

If I were to try this with three KEF150 bookshelves would this work?


L-R
220Hz -> 10000Hz

L+R
Full range

R-L
220Hz -> 10000Hz

Then I'd use the AVR to crossover to my subs at whatever measures best (likely between 80 and 100Hz)

I have found that it is best to run the three speakers at the same bandwidth, so I'd cross the center at 220Hz also. The three speakers should be set up identically, including crossover settings. That's what sounds most natural and coherent to me. But you should experiment and see how it sounds to you. For now I'm using a Denon 7.2 reciever to run everything. The stereo side channels at 220Hz and below are run from two of the channels on the receiver. Everything is sent from the comptuer to the receiver via HDMI. It's really convenient, getting rid of a lot of cables and separate dacs and amps, although it's not the state of the art in sound quality. A used receiver from craigslist is very affordable and I'm happy with it.

 

[tjcinnamon]

I have found that it is best to run the three speakers at the same bandwidth, so I'd cross the center at 220Hz also. The three speakers should be set up identically, including crossover settings. That's what sounds most natural and coherent to me. But you should experiment and see how it sounds to you. For now I'm using a Denon 7.2 reciever to run everything. The stereo side channels at 220Hz and below are run from two of the channels on the receiver. Everything is sent from the comptuer to the receiver via HDMI. It's really convenient, getting rid of a lot of cables and separate dacs and amps, although it's not the state of the art in sound quality. A used receiver from craigslist is very affordable and I'm happy with it.

Thanks for all of the help!

What runs 220Hz and below?

My weirdness, is setting it up so it will integrate with Video content (AppleTV and video games)

 

[Tim Link]

Thanks for all of the help!

What runs 220Hz and below?

My weirdness, is setting it up so it will integrate with Video content (AppleTV and video games)

You mean what speakers am I running below 220Hz? It's a pair of big corner horn woofer arrays.

 

[tjcinnamon]

@Tim Link @STC @Lion

I’m prepping to make an All-In-one box which means it goes at the end of the signal chain. The big issue this presents is that the EQing from the AVR (Dirac or Audyssey) can only be applied equally meaning if the left signal has a 2dB boost then the right signal has a 2dB boost. Otherwise, they won’t fully cancel or center.

This was unanticipated. With Audyssey MQX, I’m able to achieve that. However, with Dirac, I won’t be able to apply any correction on LR. All the EQ will have to be on the post summing signals (i.e. the output channels of the MOTU Ultralight MK5).

Signal Chain:

HTPC/AppleTV -[HDMI]> AVR -> room correction -[Analog]> MOTU -> summing/subtracting -[analog]> amp -> speaker

 

[tjcinnamon]

When speakers are too far apart crosstalk cancellation can only be achieved with phase manipulation. I used MiniAmbio with HT but my TV is small at 65 inch. Having said that, XTC for TV or HT is not practical if it involves more than one person.

How does the phase manipulation work?

With regards to "only one person". I think its "only one person who cares/notices"

 

[Tim Link]

@Tim Link @STC @Lion

I’m prepping to make an All-In-one box which means it goes at the end of the signal chain. The big issue this presents is that the EQing from the AVR (Dirac or Audyssey) can only be applied equally meaning if the left signal has a 2dB boost then the right signal has a 2dB boost. Otherwise, they won’t fully cancel or center.

This was unanticipated. With Audyssey MQX, I’m able to achieve that. However, with Dirac, I won’t be able to apply any correction on LR. All the EQ will have to be on the post summing signals (i.e. the output channels of the MOTU Ultralight MK5).

Signal Chain:

HTPC/AppleTV -[HDMI]> AVR -> room correction -[Analog]> MOTU -> summing/subtracting -[analog]> amp -> speaker

If each speaker is adjusted to sound identical at the listening position, it should still work. I'd try it and see how it sounds.

 

[tjcinnamon]

If each speaker is adjusted to sound identical at the listening position, it should still work. I'd try it and see how it sounds.

But if I EQ before I sum the signals won’t the L-R and R-L not work properly?

Basically, something that should only end up in the center would end up slightly left or right as well. Because LR would no longer be the same after EQing only one of them

 

[STC]

@Tim Link @STC @Lion

I’m prepping to make an All-In-one box which means it goes at the end of the signal chain. The big issue this presents is that the EQing from the AVR (Dirac or Audyssey) can only be applied equally meaning if the left signal has a 2dB boost then the right signal has a 2dB boost. Otherwise, they won’t fully cancel or center.

This was unanticipated. With Audyssey MQX, I’m able to achieve that. However, with Dirac, I won’t be able to apply any correction on LR. All the EQ will have to be on the post summing signals (i.e. the output channels of the MOTU Ultralight MK5).

Signal Chain:

HTPC/AppleTV -[HDMI]> AVR -> room correction -[Analog]> MOTU -> summing/subtracting -[analog]> amp -> speaker

My best wishes for the one box solution. I believe this is similar to the MiniAmbio version of Minidsp. If that is the case, the box should do cancellation only if the RC taking place before the box. Otherwise, do RC and then XTC.

Having said that, I don’t quite follow the method shared by Tim. It sounds like mid side processing which is basically extra if the centre so that it doesn’t go through XTC process. This can be achieved by using mid side encoder and decoder in a DAW. The centre is routed to the end of the signal unaltered and cancellation involves only the sides and then send to the end of the signal and mixed there. That approach is rather straight forward and you only adjust the level between centre/mid and sides after XTC. Otherwise, I am clueless with the approach that being discussed here.