Stereo Crosstalk Elimination (reduction) Par Excellence!

[Tim Link]

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

It depends. If you are EQing for subtle differences between the driver's responses, then it will make it work better. If you are EQing for room response, I just consider it one speaker and EQ the center as needed, and then copy the same EQ to the two side speakers so they all match.

 

[STC]

@tjcinnamon, if you mean the centre is slightly skewed to either side then I too experience this. When I adjust for precise centre using two mono left and right signal vs centre stereo ( single channel mono converted to stereo) the centre skews towards one side. With XTC this will be more evident.

 

[STC]

How does the phase manipulation work?

Our ears are very sensitive to sound coming about 30 degrees off the centre ( especially HF) and we can localize them even with one ear. Mere cancellation cannot suppress this since the direct source will still be located. And the only way, for the sound to extend the boundaries of speakers position is to use phase to artificially create a bigger stage. It can sound fantastic with some and mediocre with others. That’s the reason Ambiopole speakers configuration is recommended for crosstalk cancellation by placing both speakers Infront of the listener.

 

[LIΟN]

@tjcinnamon I'm doing XTC a little differently than others here. So my method is personal effective, but on the contrary, I definitely need a personal ear measurement...

This is My Front some BRIR messurement. (LL , LR, RL, RR)
I've used these to cancel using reverse phase signals repeatedly, like ubacch, and I've also tried to cancel using channel mixing.

The content itself is a bit complicated. Just to give you a rough overview, we apply arithmetic to the main response in the alignment response (the combined response of both ears we hear), extract it only up to about 5ms of direct sound, and then use it as a filter on the side of the crosstalk channel.
Then it works as a personalized optimal XTC filter for that position, that angle. (I've already tested all the 1~7m distance, 8~70 degrees angle.)

And it's pretty much the same when you compare it to the original response (which we hear with the original stereo).


The different methods had their own pros and cons. They're all interesting.
But if you want a more accurate, cleaner XTC that suits you, you'll end up having to measure your ears....

 

[tjcinnamon]

Radial Mix 2:1 2-channel Passive Mixer

2-channel Passive Mono Summing Mixer, with 2 Inputs and 1 Mono Output

www.sweetwater.com

Here’s an analog derivative. It’s passive too. Curious other things I’d have to account for? Reduced gain? Would I need a preamp? I could get one for each channel.

My outputs are rca, so I’d be feeding it less. Thoughts @Tim Link @STC @Lion

 

[Tim Link]

Radial Mix 2:1 2-channel Passive Mixer

2-channel Passive Mono Summing Mixer, with 2 Inputs and 1 Mono Output

www.sweetwater.com

Here’s an analog derivative. It’s passive too. Curious other things I’d have to account for? Reduced gain? Would I need a preamp? I could get one for each channel.

My outputs are rca, so I’d be feeding it less. Thoughts @Tim Link @STC @Lion

I don't know what this thing does to gain. It seems you'd have to lose some to isolate the channels from each other with resistors. Other than that concern, two of these should do the trick.

 

[tjcinnamon]

I don't know what this thing does to gain. It seems you'd have to lose some to isolate the channels from each other with resistors. Other than that concern, two of these should do the trick.

It actually attenuates by about 6dB to account for the clipping. I think I could account for it by adjusting the gain.

Also, wouldn't I need 3 of these? L+R, L-R, R-L or I suppose I could just wire these in inverse parallel.

I was also looking at these for L-R and R-L: https://www.parts-express.com/FMOD-Crossover-Pair-200-Hz-High-Pass-266-278?quantity=1

200Hz Highpass. Then the entire thing would be analog.

 

[Tim Link]

It actually attenuates by about 6dB to account for the clipping. I think I could account for it by adjusting the gain.

Also, wouldn't I need 3 of these? L+R, L-R, R-L or I suppose I could just wire these in inverse parallel.

I was also looking at these for L-R and R-L: https://www.parts-express.com/FMOD-Crossover-Pair-200-Hz-High-Pass-266-278?quantity=1

200Hz Highpass. Then the entire thing would be analog.

Yes, it should do that because it might gain to 6 dB and cause clipping. Is 6dB attenuation on each input enough to isolate two inputs from each other, especially if they're running out of phase of each other? It seems like most outputs want a lot more resistance than that. That's why I was thinking active buffering would be necessary. The in-line passive networks look cool. I've built things like that before and had good luck with them. I followed a stereophile article, but mine were 6 dB/ Octave. At 12 dB/octave it seems you'd once again end up reducing input impedance at some frequencies so that could be a problem. I guess as long as it remains considerably lower than the output impedance of the preamp stage, and the preamp can produce the needed current without distortion, it'll be fine.

 

[tjcinnamon]

Yes, it should do that because it might gain to 6 dB and cause clipping. Is 6dB attenuation on each input enough to isolate two inputs from each other, especially if they're running out of phase of each other? It seems like most outputs want a lot more resistance than that. That's why I was thinking active buffering would be necessary. The in-line passive networks look cool. I've built things like that before and had good luck with them. I followed a stereophile article, but mine were 6 dB/ Octave. At 12 dB/octave it seems you'd once again end up reducing input impedance at some frequencies so that could be a problem. I guess as long as it remains considerably lower than the output impedance of the preamp stage, and the preamp can produce the needed current without distortion, it'll be fine.

hmmm. My preamp is going to be the pre-outs on a Denon AVR. I'm curious if I would have to move to a Marantz pre-pro? Not my first choice because it's twice as expensive as the X6800. I think my ART cleanbox will add distortion? It doesn't feel like very "audiophile" grade device.

Back of the napkin, would an AVR pre-output out enough current?

Edit: another product in the chain: https://www.radialeng.com/product/j4 this would boost the current and convert the RCA to XLR.

The digital alternative is to get a splitter with inverted cables into a MiniDSP with at least 4 inputs: https://www.minidsp.com/products/ht-series/flex-htx

 

[tjcinnamon]

hmmm. My preamp is going to be the pre-outs on a Denon AVR. I'm curious if I would have to move to a Marantz pre-pro? Not my first choice because it's twice as expensive as the X6800. I think my ART cleanbox will add distortion? It doesn't feel like very "audiophile" grade device.

Back of the napkin, would an AVR pre-output out enough current?

Edit: another product in the chain: https://www.radialeng.com/product/j4 this would boost the current and convert the RCA to XLR.

The digital alternative is to get a splitter with inverted cables into a MiniDSP with at least 4 inputs: https://www.minidsp.com/products/ht-series/flex-htx

I forgot that the reason I didn't like the miniDSP option was the latency given that this takes place after the AVR. I think it's tolerable (8ms) and has some solid options for crossing over to larger speakers. Too bad their return policy is basically non-existent. So it's a very expensive experiment.

Digital Domain: about $1000 for the miniDSP and cables
Pros: has crossover and a single box, could cross over to MBM's, more flexibility
Cons: I had it introduce phase issues when I had a Flex. It made things chaotic; 8ms or more latency; custom phase inverting cables

Analog Domain: $700 for summing boxes, $100 cables and FMOD crossovers, (potential $350 for line level booster and xlr-to-rca converter). $1200
Pros: no latency, potentially cleaner than MiniDSP
Cons: potentially more cost OR potentially less current; more complexity

Huge bonus for both: if it works, it will work with video/movies/TV/games

 

[voodooless]

Cons: I had it introduce phase issues when I had a Flex. It made things chaotic; 8ms or more latency; custom phase inverting cables

You just had to input about 2.7 meters less distance in the speaker config in the AVR for the channels that use the flex. That would have fixed the latency issue.

 

[tjcinnamon]

You just had to input about 2.7 meters less distance in the speaker config in the AVR for the channels that use the flex. That would have fixed the latency issue.

I wish I would have taken screenshots on what the MiniDSP flex did to the phase on my speakers. It made it go haywire and that was without any processing engaged.

 

[tjcinnamon]

You just had to input about 2.7 meters less distance in the speaker config in the AVR for the channels that use the flex. That would have fixed the latency issue.

I think I only have 7ft, so about 1.5ms short but not terrible.

 

[LIΟN]

It actually attenuates by about 6dB to account for the clipping. I think I could account for it by adjusting the gain.

Also, wouldn't I need 3 of these? L+R, L-R, R-L or I suppose I could just wire these in inverse parallel.

I was also looking at these for L-R and R-L: https://www.parts-express.com/FMOD-Crossover-Pair-200-Hz-High-Pass-266-278?quantity=1

200Hz Highpass. Then the entire thing would be analog.

If you don't mind, may I ask why you want to do channel mix in analog?
It's not a criticism, it's just a pure question. Don't get me wrong! (It's a translator, so it may sound rather stiff)

 

[Chrispy]

TLDR but when is crosstalk in modern gear a problem particularly?

 

[tjcinnamon]

If you don't mind, may I ask why you want to do channel mix in analog?
It's not a criticism, it's just a pure question. Don't get me wrong! (It's a translator, so it may sound rather stiff)

Great question! In know you are a good person so I don’t interpret it as stiff (nor is the translator phrasing it that way)

I plan on implementing this after the AVR, so the latency matters. Also, until right after I ordered my Flex there wasn’t a 4 input minidsp. Literally a week after I recieved my Flex they released the HTx which has 4 inputs. I’d need to make 2 of the inputs out of phase.

Due to the non-existent return policy of the MiniDSP’s generally, they wouldn’t let me return the flex I’d just received for the flex htx

Now that it’s in one box it’s doable. Before multiple boxes would yield timing shifts.

Also, when I did hook up my flex to my Left and Right channels the result was not good. The phase went nuts and there was timing differentials between REW sweeps. 4 sweeps would usually have one outlier sweep where there was a delay. Like 3ms, 3ms, 3ms, 5ms…. Given the sensitivity of the sound needing to come out at the same time, I figured that would be a huge risk with the return policy. Acceptable risk if I could use the Htx for my subs but this new world, I’d have no use for it.

Analog would have consistent timing, better return options…. That said the ability to crossover to midbass speakers 250Hz to 80Hz would be ideal.

So long story but the reason is risk of it not working as intended

 

[LIΟN]

I plan on implementing this after the AVR, so the latency matters. Also, until right after I ordered my Flex there wasn’t a 4 input minidsp. Literally a week after I recieved my Flex they released the HTx which has 4 inputs. I’d need to make 2 of the inputs out of phase.

Due to the non-existent return policy of the MiniDSP’s generally, they wouldn’t let me return the flex I’d just received for the flex htx

Now that it’s in one box it’s doable. Before multiple boxes would yield timing shifts.

Also, when I did hook up my flex to my Left and Right channels the result was not good. The phase went nuts and there was timing differentials between REW sweeps. 4 sweeps would usually have one outlier sweep where there was a delay. Like 3ms, 3ms, 3ms, 5ms…. Given the sensitivity of the sound needing to come out at the same time, I figured that would be a huge risk with the return policy. Acceptable risk if I could use the Htx for my subs but this new world, I’d have no use for it.

If that's the case, that's totally understandable...
In the case of umik microphones, the timing is sometimes off due to the USB microphone's own clock, and even minidsp doesn't guarantee a really accurate delay, but 3ms and 5ms is a big difference.

This is a capture of an old post by another user in the Korean community for a different purpose (don't bother with the right side, it's the instability of the voice meter software, the left side is minidsp)

I know you've already made up your mind to go analog, so I don't want to add anything else, but I was wondering if the problems you were having with flex were really correct.

 

[tjcinnamon]

If that's the case, that's totally understandable...
In the case of umik microphones, the timing is sometimes off due to the USB microphone's own clock, and even minidsp doesn't guarantee a really accurate delay, but 3ms and 5ms is a big difference.

View attachment 343434

This is a capture of an old post by another user in the Korean community for a different purpose (don't bother with the right side, it's the instability of the voice meter software, the left side is minidsp)

I know you've already made up your mind to go analog, so I don't want to add anything else, but I was wondering if the problems you were having with flex were really correct.

Perhaps the Flex is not a real issue?

I’m still on the fence on a direction. If Parts Express ends up selling an HTx, that’s my preferred route. I like the idea of being able to crossover at 250Hz and then midbass to 80Hz to AVR crossover to subs below.

I’ll do some experiments and post the bizzare results with my minidsp flex

 

[LIΟN]

I’ll do some experiments and post the bizzare results with my minidsp flex

I'm looking forward to your experiments.

I'm very interested in speakers and sound, imaging, and XTC, but ironically I've disposed of all the speakers for now. (These days, I don't use headphones, and I enjoy the XTC speaker sound with IEM).
So I respect not only you but also the experiences of outstanding intellectuals such as the OP of Thread and STC.

 

[tjcinnamon]

I'm looking forward to your experiments.

I'm very interested in speakers and sound, imaging, and XTC, but ironically I've disposed of all the speakers for now. (These days, I don't use headphones, and I enjoy the XTC speaker sound with IEM).
So I respect not only you but also the experiences of outstanding intellectuals such as the OP of Thread and STC.

Yeah, I love my Polk L800’s but they are getting too far apart. When I get a bigger Tv (90+ inch) this will only get worse. So I’m planning and hoping I can capture the magic of the Polks. If it was pure music, I’d be totally happy with them.

I highly recommend them or recommend auditioning them