Doing crossovers with Jriver/Dirac..possible or not..?

[dallasjustice]

I think there are several more prosaic prerequisites needed to make computer based crossover work with DIRAC as it currently exists. If DIRAC will continue to use the DAP, then there needs to be a way to measure using DLCT with the computer crossover in place. For example, if you use the Jriver room correction tab for crossover/delay, DLCT MUST be compatible with Jriver ASIO driver. Currently , it is not compatible. I opened a ticket about this issue a few months ago. DIRAC says they are working on it.

 

The second issue with using DAP along with computer crossover/delay is that DAP must come last in signal routing. There is no way to route the signal to anything other than the playback device and still benefit from the DIRAC filter.

 

There are pluses and minuses to the DAP. I really like that it works for everything. For example, I watch a lot of Amazon instant video. Those movies still get routed through DAP. However, DAP also limits channel routing flexibility as well. DAP is also limited to ASIO or kernel devices. It can't do upnp or dlna output.

 

Michael.

 

Great idea!

 

I would like to get support for more channels. Currently I use two Lynx-cards in my HTPC, which gives me a total of 12 channels.

That would let me have multiple subwoofers, all corrected independently. Today I have two, but intend to get four more.

And, a crossover function would be high on my wanted list. At-least between mans and sub(s).

[retro]

There is no way to route the signal to anything other than the playback device and still benefit from the DIRAC filter

 

Well, there is actually. With the help of Asiobridge you can route the signal back to jriver and perform crossovers after Dirac. This is how I play 2.2 at the moment, with jriver crossing over at 40hz to my uncorrected subs. Sounds so good that I still havent bothered doing correction on the subs.

Also, again with Asiobridge, you can make crossovers in jriver which work through DLCT and then perform measurements with functioning crossovers. I tried this briefly, just to make sure it did work, and played the test-tones with steep filters in action. Haven't yet gone further to a full calibration, but I will eventually do a full 5.2 correction this way.

 

Then there's the question of where to do crossovers..before or after Dirac..? We discussed this a bit on jrivers forum, and opinions differed.

Maybe our friend flak can comment..

[dallasjustice]

I used the virtual cable. It's a PITA but I got it working. It introduced a lot of latency. Maybe there's a way to reduce the latency but the real problem is DAP. You can make the crossovers but they won't work with DAP.

Well, there is actually. With the help of Asiobridge you can route the signal back to jriver and perform crossovers after Dirac. This is how I play 2.2 at the moment, with jriver crossing over at 40hz to my uncorrected subs. Sounds so good that I still havent bothered doing correction on the subs.

Also, again with Asiobridge, you can make crossovers in jriver which work through DLCT and then perform measurements with functioning crossovers. I tried this briefly, just to make sure it did work, and played the test-tones with steep filters in action. Haven't yet gone further to a full calibration, but I will eventually do a full 5.2 correction this way.

 

Then there's the question of where to do crossovers..before or after Dirac..? We discussed this a bit on jrivers forum, and opinions differed.

Maybe our friend flak can comment..

[dallasjustice]

DAP currently cannot filter more channels than the original DLCT test sweep.

[retro]

You can make the crossovers but they won't work with DAP.

 

Not sure what you mean..? They work with DAP, I play my mains, which are corrected, crossed at 40hz..

 

And later, when I do a full 5.2 correction with jrivers crossovers in place, they come before the DLCT which has no option but to correct what its fed, such as my crossovers...

 

Or am I misunderstanding you..?!?

[dallasjustice]

The channels added with your crossover are NOT being filtered with DIRAC correction filter in DAP if you are adding them BEFORE DAP in playback signal routing. DLCT will only create filters for DAP solely from the test signal channels. So if you add channels with a crossover in playback BEFORE DAP, the additional channels will not have a DIRAC correction filter applied to them.

Not sure what you mean..? They work with DAP, I play my mains, which are corrected, crossed at 40hz..

 

And later, when I do a full 5.2 correction with jrivers crossovers in place, they come before the DLCT which has no option but to correct what its fed, such as my crossovers...

 

Or am I misunderstanding you..?!?

[retro]

The channels added with your crossover are NOT being filtered with DIRAC correction filter in DAP. DLCT will only create filters for DAP solely from the test signal channels. So if you add channels with a crossover in playback BEFORE DAP, the additional channels will not have a DIRAC correction filter applied to them.

 

No, and thats exactly what I wrote a bit further up:

 

This is how I play 2.2 at the moment, with jriver crossing over at 40hz to my uncorrected subs. Sounds so good that I still havent bothered doing correction on the subs.

 

But, when my amp for the rear channels is done, I will make a full 5.2 calibration with jrivers crossover in place before DLCT. Like I wrote, I have tried DLCT with jrivers crossovers in place and it works, at-least up to the point where I can clearly hear the test-tones being filtered.

All this possible only of-course with the help of Asiobridge, which will need to be the default windows driver. Just like I use it now. DAP does not need to be default, even if it wants to, it works perfectly with Asiobridge as default.

 

I know from a previous test that you tried to get something like this to work. I was gonna comment in that thread but never did. Maybe I should have, and then saved you the cost of your dexq unit..

 

Sorry for the under-linings, but I have actually gotten crossovers to work this way.

[dallasjustice]

Retro,

If you don't believe me, it's easy to test my theory. Go ahead and add 2 subwoofer channels using Jriver room correction tab and measure using ASIO bridge like you did. In DLCT, build a correction curve with a very steep high pass filter at 50hz. Use this test filter and tell me if you hear plenty of bass below 50hz. You will hear everything below 50hz because the DIRAC filter you created in DLCT will not work on the additional channels added in Jriver BEFORE DAP. If I am wrong, you won't hear any bass below 50hz.

 

This is exactly what I did when I tested the exasound MCH DAC. I spent a lot of time trying to get it to work before I realized that the software isn't designed for this and would need to be redesigned for it to work.

 

Michael.

[retro]

You will hear everything below 50hz because the DIRAC filter you created in DLCT will not work on the additional channels added in Jriver BEFORE DAP. If I am wrong, you won't hear any bass below 50hz.

 

How can I hear bass below 50hz if its not audible in the test-tone that the mic picks up..? It's the test-tone that rules whatever is coming next in the process. If I can filter the test-tone, then DLCT will only correct what the mic is getting.

It's the same as trying to correct a pair of 3-inch book-shelf speakers who drops of at 100hz like a stone, therefore no audible bass in the test-tone which the mic is hearing, and then still getting bass down to 20-50hz when corrected..?!? Impossible. Same thing.

 

Look, I know you tried. Obviously you didn't succeed with the exasound dac. What did you pick as output device in DCLT..?

 

No offense, but you not succeeding doesn't mean its not possible.

I will try soon. If I'm wrong, you win.

If I'm right, everybody wins...

[dallasjustice]

I don't think you understand what I am saying. All you have to do to test is to create a filter with a steep roll off at 50hz and listen to your 2.2 setup and tell me if you still hear bass below 50hz.

 

It's not a matter of winning or losing. The software simply can't do what you want it to do.

 

Michael.

 

How can I hear bass below 50hz if its not audible in the test-tone that the mic picks up..? It's the test-tone that rules whatever is coming next in the process. If I can filter the test-tone, then DLCT will only correct what the mic is getting.

It's the same as trying to correct a pair of 3-inch book-shelf speakers who drops of at 100hz like a stone, therefore no audible bass in the test-tone which the mic is hearing, and then still getting bass down to 20-50hz when corrected..?!? Impossible. Same thing.

 

Look, I know you tried. Obviously you didn't succeed with the exasound dac. What did you pick as output device in DCLT..?

 

No offense, but you not succeeding doesn't mean its not possible.

I will try soon. If I'm wrong, you win.

If I'm right, everybody wins...

[retro]

I don't think you understand what I am saying. All you have to do to test is to create a filter with a steep roll off at 50hz and listen to your 2.2 setup and tell me if you still hear bass below 50hz.

 

It's not a matter of winning or losing. The software simply can't do what you want it to do.

 

And I don't think your reading what I'm writing..? No offense, I'm not trying to wind you up, not intentionally anyway..

Guess I have to drag out my mic and stand tonight if I have the energy...

 

Again, what did you use as output-device in DLCT..?

[dallasjustice]

Just do a simple stereo setup DLCT. (You can't select a 2.1 configuration because DLCT won't crossover. That's strictly for home theater LFE.) Setup crossover in Jriver the way you want. Use ASIO bridge as output in DLCT and route ASIO bridge to Jriver ASIO so test signal goes through your crossovers in Jriver.

 

Build your target curve so that the FR cuts off less than 50hz. Use this test filter in DAP. Set Jriver audio output to DIRAC, like normal. Make sure you check the room correction tab so that it's applying your crossover and adding the sub; same as before during the test sweep. Play music with bass below 50hz. Tell me if you hear plenty of bass below 50hz.

 

And I don't think your reading what I'm writing..? No offense, I'm not trying to wind you up, noto intentionally anyway..

Guess I have to drag out my mic and stand tonight if I have the energy...

 

Again, what did you use as output-device in DLCT..?

[retro]

I can't really make sense of your last post:

 

Just do a simple stereo setup DLCT. (You can't select a 2.1 configuration because DLCT won't crossover. That's strictly for home theater LFE.)

 

Of-course I can make a 2.1 in DLCT. That's how I played it for many months with my previous Thiel-sub. It obviously sweeps the sub with a full-range signal, and I then used jriver to crossover. That gave me the most perfect integration of my sub ever.. And after listening to my system, a couple of friends instantly bought Dirac because of the seamless integration of the sub. They now use Dirac the exact same way, sweeping 2.1 full-range, then making crossovers in jriver. I've heard their systems many times, sub-integration is perfect.

But then we're still talking about adding crossovers after DLCT. Not like we're discussing here, crossovers before both DLCT and DAP, right..? But even if that approach gave subjectively excellent results, I wan't to try this method out of curiousity, and if the sonics can be improved further by not sweeping the sub full-range, but limited to the FR it should really be working in. In my case that's 0-40hz and the mains 40hz-up. Hopefully, as a bonus, it will give Dirac an easier job creating filters due to the limited FR of both mains and sub.

 

Build your target curve so that the FR cuts off less than 50hz.

 

I'm not with you here..? That's what the crossovers made by jriver should do in the first place during DLCT sweeps, low-pass below 50hz at my selected slope..??

 

Make sure you check the room correction tab so that it's applying your crossover and adding the sub; same as before during the test sweep. Play music with bass below 50hz. Tell me if you hear plenty of bass below 50hz.

 

Again, not with you..? If my approach is working, which is yet unproven , then I won't need to add the crossovers in jriver. That would just be doubling up on the crossover used during the sweep.

[dallasjustice]

Okay. I've explained to you why one can't crossover using Jriver and DIRAC. PLEASE now explain your routing in the 2.1 system where you've crossed over your main R/L with a DIRAC filter applied to all 3 channels in your 2.1 system. If you've done this, would I be asking too much for you to build a test target curve that drops off like a rock below 50hz? It's just a test which will demonstrate to you that even though you may like your 2.1 system, DIRAC wasn't applying any filter to the subwoofer when you and your friends were listening to it.

 

 

I can't really make sense of your last post:

 

 

 

Of-course I can make a 2.1 in DLCT. That's how I played it for many months with my previous Thiel-sub. It obviously sweeps the sub with a full-range signal, and I then used jriver to crossover. That gave me the most perfect integration of my sub ever.. And after listening to my system, a couple of friends instantly bought Dirac because of the seamless integration of the sub. They now use Dirac the exact same way, sweeping 2.1 full-range, then making crossovers in jriver. I've heard their systems many times, sub-integration is perfect.

But then we're still talking about adding crossovers after DLCT. Not like we're discussing here, crossovers before both DLCT and DAP, right..? But even if that approach gave subjectively excellent results, I wan't to try this method out of curiousity, and if the sonics can be improved further by not sweeping the sub full-range, but limited to the FR it should really be working in. In my case that's 0-40hz and the mains 40hz-up. Hopefully, as a bonus, it will give Dirac an easier job creating filters due to the limited FR of both mains and sub.

 

 

 

I'm not with you here..? That's what the crossovers made by jriver should do in the first place during DLCT sweeps, low-pass below 50hz at my selected slope..??

 

 

 

Again, not with you..? If my approach is working, which is yet unproven , then I won't need to add the crossovers in jriver. That would just be doubling up on the crossover used during the sweep.

[retro]

Okay. I've explained to you why one can't crossover using Jriver and DIRAC. PLEASE now explain your routing in the 2.1 system where you've crossed over your main R/L with a DIRAC filter applied to all 3 channels in your 2.1 system. If you've done this, would I be asking too much for you to build a test target curve that drops off like a rock below 50hz? It's just a test which will demonstrate to you that even though you may like your 2.1 system, DIRAC wasn't applying any filter to the subwoofer when you and your friends were listening to it.

 

Dallas, I don't need to demonstrate anything to myself when it comes to my previous 2.1 system. It worked extremly well. Of-course the subwoofer was Dirac-processed..

Explain the routing..? Channel 1 (Left), Channel 2 (Right), Channel 4 (sub), standard routing.

Just to make you happy, I started DLCT and loaded one of my previous projects. I'm gonna show you every screen except the mic-screen, which is of no importance in this case.

 

 

As you can see on picture 4 and 5, (besides my ragged room )the sub with it's dual 15-inch is swept full-range and has output way up to 3khz. That means unnecessary burden on Dirac, and one of the reasons I want to try crossover before DLCT.

 

But this setup, with steep crossovers in jriver, worked excellent.

[dallasjustice]

Okay. I understand what you are doing. You've set up DLCT using a custom 2.1 configuration. The problem you have integrating the sub this way is that the test sweep will not run the sub and the R/L at the same time in a full range sweep. Remember, All MCH configurations with a sub in DLCT are made strictly for home theater LFE integration. Therefore, your test sweep was for home theater use, not crossover. You can't integrate a sub with main R/L without doing a full range test sweep which includes the sub in said sweep. The sweep for the sub and for R/L are seperate. You can't integrate a subwoofer with R/L using crossovers with seperate test sweeps.

 

Dallas, I don't need to demonstrate anything to myself when it comes to my previous 2.1 system. It worked extremly well. Of-course the subwoofer was Dirac-processed..

Explain the routing..? Channel 1 (Left), Channel 2 (Right), Channel 4 (sub), standard routing.

Just to make you happy, I started DLCT and loaded one of my previous projects. I'm gonna show you every screen except the mic-screen, which is of no importance in this case.

 

[ATTACH=CONFIG]13184[/ATTACH][ATTACH=CONFIG]13185[/ATTACH][ATTACH=CONFIG]13186[/ATTACH]

[ATTACH=CONFIG]13187[/ATTACH][ATTACH=CONFIG]13188[/ATTACH]

 

As you can see on picture 4 and 5, (besides my ragged room )the sub with it's dual 15-inch is swept full-range and has output way up to 3khz. That means unnecessary burden on Dirac, and one of the reasons I want to try crossover before DLCT.

 

But this setup, with steep crossovers in jriver, worked excellent.

[retro]

You've set up DLCT using a custom 2.1 configuration. The problem you have integrating the sub this way is that the test sweep will not run the sub and the R/L at the same time in a full range sweep.

 

Why would it and why should it..? I also have Audiolense XO since many years and even though I haven't used it for a while, it also runs each channel separately. I also (!) have Trinnov in the shape of the Sherwood R-972, same thing there, it runs each channel separately.

 

Remember, All MCH configurations with a sub in DLCT are made strictly for home theater LFE integration. Therefore, your test sweep was for home theater use, not crossover. You can't integrate a sub with main R/L without doing a full range test sweep which includes the sub in said sweep. The sweep for the sub and for R/L are seperate. You can't integrate a subwoofer with R/L using crossovers with seperate test sweeps.

 

There's no technical difference in integrating a sub for home theater vs. integrating it into a 2-ch system. Yes, the sweeps are separate. I don't get why you're of the opinion they need/should to be combined..? Do you have a source for that..?? Like I wrote above, both Audiolense and Trinnov use separate sweeps. I have read that Dirac in the Datasat unit also use separate sweeps, then the crossovers are/should be applied.

 

Flak, where are you when we need you..

[dallasjustice]

You can't measure your combined response, at least in DIRAC. Of course, DIRAC would know best but I really don't think DIRAC was ever intended to be used this way to create crossovers. If it sounds great, I am happy for you. I would be curious to see the final result in REW. I do appreciate you posting your screenshots and I fully understand what you are doing. So, thanks for that.

 

 

Why would it and why should it..? I also have Audiolense XO since many years and even though I haven't used it for a while, it also runs each channel separately. I also (!) have Trinnov in the shape of the Sherwood R-972, same thing there, it runs each channel separately.

 

 

 

There's no technical difference in integrating a sub for home theater vs. integrating it into a 2-ch system. Yes, the sweeps are separate. I don't get why you're of the opinion they need/should to be combined..? Do you have a source for that..?? Like I wrote above, both Audiolense and Trinnov use separate sweeps. I have read that Dirac in the Datasat unit also use separate sweeps, then the crossovers are/should be applied.

 

Flak, where are you when we need you..

[retro]

You can't measure your combined response, at least in DIRAC. Of course, DIRAC would know best but I really don't think DIRAC was ever intended to be used this way to create crossovers. If it sounds great, I am happy for you. I would be curious to see the final result in REW. I do appreciate you posting your screenshots and I fully understand what you are doing. So, thanks for that.

 

It is working, I promise you. For me and for my friends. I would rather think your approach has more downsides, no offense..

 

Flavio, it would be great if you can comment on this..please..?

[retro]

I have PM'd flak for comments and info in this thread.

[dallasjustice]

It is working, I promise you. For me and for my friends. I would rather think your approach has more downsides, no offense..

 

Flavio, it would be great if you can comment on this..please..?

I don't doubt that it works. Of course, I still think it much better to use a digital crossover AFTER DAP so that DIRAC can build a single filter for each speaker using the combined response.

[flak]

 

Flak, where are you when we need you..

 

Hello Retro and Dallasjustice,

 

some time ago I've been talking with Dallasjustice about this subject and I told him that I had no personal experience on alternative solutions to the downstream digital crossover that I'm personally using, but that I was very interested in his findings about third party software digital crossovering solutions.

So I had not answered yet 'cause I have more to learn than to teach... in any case in order to give some support I have asked to a colleague, Jakob Agren, and these are his comments:

----------------------------------------------------------------------------------------------

1. I have no experience with asio bridge. If it can indeed make the DLCT measure through a crossover in for instance jRiver, that would work. The setup for a 2.2 system would be 2 channel project in the DLCT, a 2->4 upmix with cross overs in jRiver to a 4 channel output device. There can be issues with the phase correction of the subwoofer is in a totally different location, especially if crossed high as the DLCT is not aware its actually 2 different speakers. The result would be a compensation valid for both the subwoofers and the full range speakers. Retro says this is possible, I don’t know if it is, its completely up to asio bridge and jriver to make this work. Also jriver would have to be used for playback obviously to apply the cross over used when measuring.

 

2. Use a custom project (3 or 4 channel) with extra channels for subwoofers.

This works as well. There is (as Dallas Justice argues) no automatic mechanism to control what happens to the sum of the subwoofer and the full range speaker, this will have to be handled manually through parameters in the cross overs. The playback software can chose to play back on channels 1 and 2 for only the full range speakers that are compensated according to whatever target was used, and may opt to use channels 3 and/or 4 to play things on the subwoofers. The content to the subwoofers has to be generated by some other software such as jriver, and the user has to make sure all speakers add up nicely manually.

 

Difference between 1 and 2 then is:

1 will generate as many filters as there are input channels. The filters are optimized for the combination of the full range speaker and the subwoofer.

2 will generate one filter for each speaker, and the user is responsible for somehow providing a reasonable input signals to all of the output channels.

Both solutions require a media player that can do the upmix, either before or after the application of the Dirac Live filters. Obviously number 1 (cross overs after filter application) require measurements with crossovers in the signal path. Number 2 requires more knowledge, time patience etc about how to manually integrate (compensated) subwoofers with the full range speakers.

 

As I see the thread the disagreement is on 2 things:

 

1. Can you ever measure through computer based digital cross overs? I don’t know, retro says yes with asiobridge. We don’t support that, but if it works that’s great.

 

2. Dallas justice is right that there is no automatic compensation for the total system when compensating each speaker individually and doing cross overs before filter application. That does not mean it cannot sound good, or that it is a bad idea.

--------------------------------------------------------------------------------------------------------------------------

 

Thanks for posting such interesting threads

Flavio

[dallasjustice]

Flavio,

Please tell Jakob I said thank you for the thoughtful response. Of course, retro and I wouldn't be fussin' over the best way to use 3rd party software/hardware to implement a crossover if DIRAC LIVE would implement crossover functionality in DLCT and DAP. I know; easier said than done. :-]

 

Hello Retro and Dallasjustice,

 

some time ago I've been talking with Dallasjustice about this subject and I told him that I had no personal experience on alternative solutions to the downstream digital crossover that I'm personally using, but that I was very interested in his findings about third party software digital crossovering solutions.

So I had not answered yet 'cause I have more to learn than to teach... in any case in order to give some support I have asked to a colleague, Jakob Agren, and these are his comments:

----------------------------------------------------------------------------------------------

1. I have no experience with asio bridge. If it can indeed make the DLCT measure through a crossover in for instance jRiver, that would work. The setup for a 2.2 system would be 2 channel project in the DLCT, a 2->4 upmix with cross overs in jRiver to a 4 channel output device. There can be issues with the phase correction of the subwoofer is in a totally different location, especially if crossed high as the DLCT is not aware its actually 2 different speakers. The result would be a compensation valid for both the subwoofers and the full range speakers. Retro says this is possible, I don’t know if it is, its completely up to asio bridge and jriver to make this work. Also jriver would have to be used for playback obviously to apply the cross over used when measuring.

 

2. Use a custom project (3 or 4 channel) with extra channels for subwoofers.

This works as well. There is (as Dallas Justice argues) no automatic mechanism to control what happens to the sum of the subwoofer and the full range speaker, this will have to be handled manually through parameters in the cross overs. The playback software can chose to play back on channels 1 and 2 for only the full range speakers that are compensated according to whatever target was used, and may opt to use channels 3 and/or 4 to play things on the subwoofers. The content to the subwoofers has to be generated by some other software such as jriver, and the user has to make sure all speakers add up nicely manually.

 

Difference between 1 and 2 then is:

1 will generate as many filters as there are input channels. The filters are optimized for the combination of the full range speaker and the subwoofer.

2 will generate one filter for each speaker, and the user is responsible for somehow providing a reasonable input signals to all of the output channels.

Both solutions require a media player that can do the upmix, either before or after the application of the Dirac Live filters. Obviously number 1 (cross overs after filter application) require measurements with crossovers in the signal path. Number 2 requires more knowledge, time patience etc about how to manually integrate (compensated) subwoofers with the full range speakers.

 

As I see the thread the disagreement is on 2 things:

 

1. Can you ever measure through computer based digital cross overs? I don’t know, retro says yes with asiobridge. We don’t support that, but if it works that’s great.

 

2. Dallas justice is right that there is no automatic compensation for the total system when compensating each speaker individually and doing cross overs before filter application. That does not mean it cannot sound good, or that it is a bad idea.

--------------------------------------------------------------------------------------------------------------------------

 

Thanks for posting such interesting threads

Flavio

[retro]

Guess my approach got it's blessing from Dirac then..

 

Thanks Flavio and Jakob!

[dallasjustice]

Guess my approach got it's blessing from Dirac then..

 

Thanks Flavio and Jakob!

Yes! You should name it. Let's call it: Retro's variations on a theme of DIRAC