Audiolense solutions for correction of a multi-sub array.

[2234rew]

I agree. And when you consider the capacity to move air....

This must be one of the most sophisticated setups you've seen?

Maybe up there with a commercial cinema complex perhaps LOL

 

[Sonnie Parker]

My wife thinks the home theater is primarily used for playing measurement sweeps.

 

[Jiri_ch]

For some unknow reason my wife thinks the same.....

BTW: Hulkss I'm assume that the horns are PSE 144 wondering if you have some more in the back and also are you running them passive or active
J

 

[hulkss]

This must be one of the most sophisticated setups you've seen?
Maybe up there with a commercial cinema complex perhaps LOL

Thanks for the compliment. I was beginning to think of it more as hodgepodge until I got these great results with Audiolense. I was using DSP solutions that did not involve convolution to get the latency down as the family likes to watch online streaming media and high latency is a lip-sync issue. I tested the new AL convolver and I decided to give Audiolense another try. Now I can have both a TTD solution for my recorded media where (JRiver manages lip-sync) and a Min Delay solution for streaming.

 

[hulkss]

For some unknow reason my wife thinks the same.....

BTW: Hulkss I'm assume that the horns are PSE 144 wondering if you have some more in the back and also are you running them passive or active.

Yes, they are PSE-144's, active, and with some upgrades. Only three of them, they would not make good surround speakers.

 

[juicehifi]

This must be one of the most sophisticated setups you've seen?

Maybe up there with a commercial cinema complex perhaps LOL

Absolutely.

It makes me proud to see Audiolense playing a part in such an exceptional system.

It is perhaps the most really cool bass rig I’ve seen in a multichannel system. And multichannel sounds clearly better IMO.

There are more really cool bass arrangements out there, but AFAIK they may all be beyond capable of delivering anything that has been recorded in audio.

 

[juicehifi]

My

Thanks for the compliment. I was beginning to think of it more as hodgepodge until I got these great results with Audiolense. I was using DSP solutions that did not involve convolution to get the latency down as the family likes to watch online streaming media and high latency is a lip-sync issue. I tested the new AL convolver and I decided to give Audiolense another try. Now I can have both a TTD solution for my recorded media where (JRiver manages lip-sync) and a Min Delay solution for streaming.

Have you tested if Jriver synchronises the video to a TTD correction in the Al conolver?

 

[Joe Fan]

Wow this is great write up and awesome system.
Are you using Motu's thunderbolt or USB to connect to your pc?
Just wondering if thunderbolt work with Audiolense.

 

[hulkss]

Have you tested if Jriver synchronises the video to a TTD correction in the Al conolver?

I have only tested an AL low latency filter with online streaming and JRiver audio. It worked great. The only thing JRiver can do in this case is delay audio which is not what we need.

JRiver can delay video that is being rendered by JRiver, usually from the computer's DVD/Blu-ray player or a hard drive file. Delay is automatically corrected for JRiver DSP including the JRiver convolver. The AL Convolver would have to pass a latency value to JRiver for automatic correction.

This menu is where JRiver A/V sync would be set manually to work with the AL Convolver. +/- values are allowed to advance or retard.

 

[hulkss]

Wow this is great write up and awesome system.
Are you using Motu's thunderbolt or USB to connect to your pc?
Just wondering if thunderbolt work with Audiolense.

I have used both USB and Thunderbolt with windows 10 and they both work well for audio playback. I had a weird feedback problem with Thunderbolt the other day when trying to record measurements and I had to switch to USB. My computer is Thunderbolt 3 and the MOTU is Thunderbolt 2. I have to use a Startech adapter which may be the problem.

Thunderbolt works only with ASIO on Windows in my system, USB works with WASAPI or ASIO. There is very little difference between Thunderbolt and USB. On the 16a, a USB connection can handle 24 channels at 192 kHz sample rate or 64 channels at 48 kHz. Thunderbolt is about double that. Slightly less latency may be possible with Thunderbolt but the adapter may undo that small difference. I've read that Thunderbolt is definitely best if you're using a Mac.

 

[juicehifi]

I have only tested an AL low latency filter with online streaming and JRiver audio. It worked great. The only thing JRiver can do in this case is delay audio which is not what we need.

JRiver can delay video that is being rendered by JRiver, usually from the computer's DVD/Blu-ray player or a hard drive file. Delay is automatically corrected for JRiver DSP including the JRiver convolver. The AL Convolver would have to pass a latency value to JRiver for automatic correction.

This menu is where JRiver A/V sync would be set manually to work with the AL Convolver. +/- values are allowed to advance or retard.

View attachment 39084

Audiolense convolver passes the delay to jriver if jriver asks for it. So I am anxious to find out if jriver will delay the video automatically

 

[Omid]

I have a question: You're generating a filter for the 5 subs, without including your main speakers. Once you have a dsp filter for the combined subs, how to integrate it with the speakers?

 

[hulkss]

I have a question: You're generating a filter for the 5 subs, without including your main speakers. Once you have a dsp filter for the combined subs, how to integrate it with the speakers?

Audiolense generates bass "offloading" filters automatically for each of the 5 subwoofer channels and each of the front main speakers. That's three mains x five subs for fifteen bass routing filters. Fortunately my Ryzen 5900X processor can process all these filter paths (30 total with 65536 taps) at about 12% capacity using all 24 physical and virtual cores with zero latency in the AL convolver.

Here's the full setup:

Here's a full filter set view (just a sample for illustration I'm still working it):

 

[Omid]

Right, I do the same (but much less complex). I was looking at the filters Bernt worked on (it looks like the file only includes the subwoofers). It yields a very nice step response, but doesn’t include the main speakers.
When I setup my system as 2.1 (the sub has 2 drivers, the main speakers each have a mid-woofer and tweeter) and I generate a step response, AL Shows me a step response for the main speakers, and overlays a second wave for the subwoofer. I would’ve loved to see a combined step response. I realize I can do that through REW, but I would just take too long for trying many different iterations.

 

[hulkss]

Just select simulated result only, then generate the step response. I believe you get a result including the sub if the speaker chosen is routing bass to the sub.

 

[Omid]

Looking at your picture, I assumed that the graph called 'step response eva nr FrontLeft' in the legend didn't include the subwoofer (because of the separate subwoofer step response graph). But I think you're right: it must be the combo of main + sub.

 

[hulkss]

I'm getting close to bringing the full system back online. I decided to do a system check by generating set of crossover filters only. I set partial correction at zero Hz and exported an AL correction file. I fired up JRiver with that correction in the new AL Convolver. In my MOTU interface I looped back an REW measurement signal (sent into JRiver) to verify the systems electrical signal response (all digital, I did not loop through DA/AD).

Here's a look at the Left Bass channel which should be 80 - 350 Hz. YUCK.....not what I was hoping for.

This turned out to be a buffer problem. I found that a 10 ms buffer out of JRriver to the convolver and a 512 sample buffer (48 kHz) out of the convolver into the MOTU worked well. Performance was the same on USB or Thunderbolt. USB seemed more robust.

Here is the full set of left channel XO filters communicating properly: Nice!

I mixed these four in the MOTU to see the summed signal with no corrections.

I really don't know if this has any real meaning to the system? AL corrects the acoustic response of the system and does XO together somehow.

Well, in any case, I'm glad I did a signal check.

 

[specht007]

Here is the result measuring all subs together as a group through the MSO filters (the MSO filters are loaded into my MOTU 16a audio interface).

Very impressive!

I am about to integrate 5 subs, currently I have 4 and run them as cascaded stereo subs in AL. I was initially planning to integrate the 5th sub as a part of a 3 sub array <50Hz with MSO for the low end via a MiniDSP 2x4 HD and use my 2 Rythmik F8s as stereo midrage subs >50<200 Hz.

But since I am using a Motu Ultralite MK4 as the Main DAC I probably don’t need the MiniDSP? I did not know that you could upload the MSO filters or values into the Motu. Could you share how to do that?

 

[hulkss]

I enter the filters manually into the MOTU 16a Pro Audio Control Software Mixer. I'm no longer using MSO. It does not make sense to use it with AL correcting each individual sub separately. AL will "wipe out" the MSO filters.
MSO works well if AL puts a final filter over the whole sub group combined as one.

 

[specht007]

I enter the filters manually into the MOTU 16a Pro Audio Control Software Mixer. I'm no longer using MSO. It does not make sense to use it with AL correcting each individual sub separately. AL will "wipe out" the MSO filters.
MSO works well if AL puts a final filter over the whole sub group combined as one.

View attachment 39113

Got it.Thanks!

 

[hulkss]

Audiolense convolver passes the delay to jriver if jriver asks for it. So I am anxious to find out if jriver will delay the video automatically

No luck with automatic video delay in JRiver. With a 1950 ms latency convolution filter in the AL convolver, I had to enter 1600 ms into JRiver to correct lip-sync. This was a quick check with my computer monitor.

 

[juicehifi]

Thanks. I will see a msn about a dog.

 

[hulkss]

Hi Bernt,
I took the "preliminary" full system TTD correction I used for latency checking and ran some quick tests on it looking for problems as described in this paper link: "METHODS TO ELIMINATE THE BASS CANCELLATION BETWEEN LFE AND MAIN CHANNELS".

I checked the L, R, C, & LFE channels sending sweeps through individual and combinations of channels measuring the acoustic response at 10-100 Hz. I found the Right channel to have inverted polarity.

Here's the magnitude response of L, R, & Sub individually in green; R + Sub in blue (problem); R inverted + Sub, & L + Sub in Red. Reds show perfect +6dB addition. How should I proceed to correct and avoid this? I did notice some occasional inverted step responses while experimenting with filters.

Left & Right Magnitude & Phase

 

[juicehifi]

If you use the same bass offloading frequency for left and right speakers, they will use the same subwoofer filters. And they will be optimiset for the left speaker. So simply add 0.1Hz to the right speaker’s offloading frequency.

 

[hulkss]

That is what I did in the set-up. I have 15 bass offloading filters

Not, sure but I may have noticed the polarity flipping with different TTD window lengths.