Audiolense solutions for correction of a multi-sub array.

[juicehifi]

If this appeared as a problem in the simulations I could have another look. Is this based on the measurement you sent me?

 

[hulkss]

I'll send you the measurement of the full system that I am using.

I just checked a minimum delay solution. L,R, & C play together perfectly, but they don't play well with LFE. The bass routing works fine for each channel. LFE does not sum properly with the offloaded bass. I believe this is because the offloaded bass signal has a LPF at 80 Hz. The LFE does not. This is the problem described in the paper I referenced. I need to fix this somehow.

I believe that TTD and linear phase XO may preclude this problem as shown in the earlier post.

 

[hulkss]

Ha! I fixed that problem in the post above. Don't forget to use -10 dB on the LFE channel relative to others when generating test signals. Here's with a "sub optimal" min delay filter set for proof of concept:

I was lucky to find an easy solution. Something similar was done by a Pro Speaker company called Genelec.

1. I unchecked 10 dB amplification of LFE
2. I reduced the level of all channels except LFE by 10 dB
3. I mixed the LFE with the center channel
4. I muted the LFE channel going to the Sub
5. I added 10 db of Gain to the AL Convolver

Here is the entire solution not including the convolver gain:

 

[juicehifi]

A subwoofer is a separate speaker when it processes LFE and no measures have been taken to ensure integration between LFE and the main speakers.

If that becomes a problem, simply route the LFE signal to the main speakers and leave it to the bass ofloading, which is properly synchronised with the main speakers.

 

[jjazdk]

I would like to request a feature for multi-sub-setups.

1. Consider a setup with multiple subwoofers of the exact same type, but different distances to MLP.
Audiolense will equalize them to be equally loud in MLP, which will put more stress on the subwoofers that are the furthest away from MLP.

2. Consider a setup with multiple subwoofers, of different capability (cone area).
Audiolense will equalize them to be equally loud in MLP, which will put more stress on the subwoofers that have the least cone area.

I would like the possibility to level adjust the subwoofers independently, to account for these differences. F.x. if I measure my subwoofers, and can see that one sub has an overall 5dB less output volume than the others, then I would want to keep that difference and not put an extra 5dB gain/stress on that particular subwoofer.

 

[hulkss]

I would like to request a feature for multi-sub-setups.

1. Consider a setup with multiple subwoofers of the exact same type, but different distances to MLP.
Audiolense will equalize them to be equally loud in MLP, which will put more stress on the subwoofers that are the furthest away from MLP.

2. Consider a setup with multiple subwoofers, of different capability (cone area).
Audiolense will equalize them to be equally loud in MLP, which will put more stress on the subwoofers that have the least cone area.

I would like the possibility to level adjust the subwoofers independently, to account for these differences. F.x. if I measure my subwoofers, and can see that one sub has an overall 5dB less output volume than the others, then I would want to keep that difference and not put an extra 5dB gain/stress on that particular subwoofer.

This suggestion has merit. I was lucky and could compensate with a high pass filter on the Khorn channels. I am now pushing the Epik subs with more gain. In my sub line array, equal output sounds great and there is enough headroom for the corrections.

 

[hulkss]

A subwoofer is a separate speaker when it processes LFE and no measures have been taken to ensure integration between LFE and the main speakers.

If that becomes a problem, simply route the LFE signal to the main speakers and leave it to the bass ofloading, which is properly synchronised with the main speakers.

I can say from listening tests that this is a HUGE issue, every bit as big as those -15 dB valleys in bass response show in the post above.

That said, this fix does nothing for a typical move soundtrack. Why? The LFE signal only contains "special effects" and is totally uncorrelated with the other channels. That is what it was designed to do.

I mainly watch multi-channel live music concert recordings from Blu-ray or DVD. What do the mix engineers do? Well I can tell you that they do not follow the rules of the motion picture industry.

I have examined the sound tracks on several disks. In some cases they put the same bass track on Left, Right and LFE. This is where the fix above is absolutely essential for proper bass response.

Another issue is that the mix may have music in the LFE signal at frequencies well above what you want to hear from your sub woofer. Here is the LFE signal from a concert video (a snapshot of the JRiver signal analyzer). If you solo this channel, you can hear this is the bass guitar and kick drum tracks on the LFE channel!

In the "mix fix" suggested above, the higher frequency bass will not be lost, distorted, or improperly localized by multiple subs. The LFE signal will play through the center speaker above the offloading frequency that you have selected in AL.

I played a few minutes of a concert through the Jriver convolver with, and then without the "Mix Fix" while recording the output to .wav files. I then analyzed and differenced the two audio signals using BEQdesigner software. Here is the loss of bass matching the 15 dB drops shown with sine sweeps.

 

[juicehifi]

That was a neat solution you made up there with routing the LFE through the center, hulkss,

The difference that your last chart displays - that is the difference between a regular LFE and the center channel solution, right? Is everything due to poor integration or is it in part due to a low cutoff on the subwoofer? In any case I like your approach much better. I believe you are moving the frontier here.

Another advantage you get from running the LFE through a full range channel is faster risetime and a timing that is more in line with the rest of the speakers. In my time with the LFE I have always wondered whether the sub should start to play at the same time as the other speakers or whether it should peak at the same time. That problem goes away with your solution.

 

[hulkss]

That was a neat solution you made up there with routing the LFE through the center, hulkss,

The difference that your last chart displays - that is the difference between a regular LFE and the center channel solution, right? Is everything due to poor integration or is it in part due to a low cutoff on the subwoofer? In any case I like your approach much better. I believe you are moving the frontier here.

Another advantage you get from running the LFE through a full range channel is faster risetime and a timing that is more in line with the rest of the speakers. In my time with the LFE I have always wondered whether the sub should start to play at the same time as the other speakers or whether it should peak at the same time. That problem goes away with your solution.

I recognized the problem when experimenting with BEQdesigner software. Then I found the paper I referenced and in it was hidden a one line comment that Genelec routed LFE through the center speaker. Then I applied the flexibility of your routing matrix for a solution. It really does work well without any downside as far as I can tell (you do usually need to return some digital gain after the AL filters).

It is the cutoff on the LP bass routing filter that is the problem. Here is a plot of what happens when you sum an offloaded channel with LFE:

Most sub arrays need to avoid speaker localization with a low cut-off frequency and the upper bass on LFE is lost. Mixing the LFE channel into the center channel before AL processing gets mid bass and harmonics (when present on LFE) into the center speaker and avoids the problems of upper bass going into a distributed array of sub woofers. This provides proper imaging and precludes the comb filtering problem of multiple sources.

I have also learned to not offload bass from surround speakers. It can ruin the "clean" bass on the main channels. Place your subs to work well with the main speakers.

Bass offloading of surrounds splits the middle of the bass frequency range into widely separated sources and the acoustic recombination is a mess. Just use a target for the surrounds with a bass level they can handle. They will sound much better.

 

[juicehifi]

hulkss,

I\ve been doing some testing with your new setup. Audiolense is not handling the the situation with partly parallel subs as it should. Frankly I think this is the first time it has been tested. I wil get out a fix as soon as I can.

 

[hulkss]

hulkss,

I've been doing some testing with your new setup. Audiolense is not handling the the situation with partly parallel subs as it should. Frankly I think this is the first time it has been tested. I will get out a fix as soon as I can.

Hi Bernt,

While you are working on Audiolense, here is the wish list that has been identified in this thread so far:

1. Ability to set an output level adjustment for individual drivers of a sub array (prior to the combined array correction). (no work around has been identified)
2. Ability use a fullrange speaker as part of a sub array. (I pick "small" and route a sub driver channel in my MOTU mixer)
3. Maximize digital gain in the correction when mixing LFE into the Center Channel. (add gain in JRiver DSP or the AL Convolver)
4. Ability to not have an LFE cutoff filter. (I pick a high frequency as a work around)
5. I would like to be able to prevent treble boost on the sub channel only. (I use a separate sub target now)
6. Some bugs in step response and LFE cutoff frequency not implementing when requested.

More software leads to the need for more software.

 

[juicehifi]

Hi Hulkss,

I'm afraid most of these will have to wait until later. I have too much on my plate. In the meantime I can only offer a few pointers.

1) You can stack the weaker subs together and run them in parallel in a sub woofer. Alternatively, you can feed two or three subs - or all 5 for that matter - from the same sound card output by means of some line level cable splitting. You may also be able to do it in through your mixer's sound card.
2) You can put anything in a sub, but if you want full range output you may as well call it a regular speaker. It can till handle LFE. The only thing that sets a sub apart from other speakers is the lack of output above a few 100 Hz.
3) I believe you can work around that in the in-out matrix. Route the LFE to the center channel at whatever comparable spl level you prefer.
4) If you make a separate target for the sub, an lfe cutoff will not be used.
5) I do not recall at the moment, but I don't think it applies there.

I hope this helps ... and I believe these issues of yours will be less pressing when I get Audiolense in order for the subwoofer configuration you showed above.

 

[hulkss]

Hi Hulkss,

I'm afraid most of these will have to wait until later. I have too much on my plate. In the meantime I can only offer a few pointers.

1) You can stack the weaker subs together and run them in parallel in a sub woofer. Alternatively, you can feed two or three subs - or all 5 for that matter - from the same sound card output by means of some line level cable splitting. You may also be able to do it in through your mixer's sound card.
2) You can put anything in a sub, but if you want full range output you may as well call it a regular speaker. It can till handle LFE. The only thing that sets a sub apart from other speakers is the lack of output above a few 100 Hz.
3) I believe you can work around that in the in-out matrix. Route the LFE to the center channel at whatever comparable spl level you prefer.
4) If you make a separate target for the sub, an lfe cutoff will not be used.
5) I do not recall at the moment, but I don't think it applies there.

I hope this helps ... and I believe these issues of yours will be less pressing when I get Audiolense in order for the subwoofer configuration you showed above.

Hi Bernt,

1. That may work for me, I'm not lacking sub output, but I could combine the four Epik subs into one instead of two and two.
2. The issue here is bass routing. I would like AudioLense to do a 5 driver sub array as: Left Full Range & Sub; a Sub with Drivers 0,1,2.....; Right Full Range & Sub and get the Bass below the offloading frequency into all 5 subs including from Right and Left, not just from the small speakers. I can't see a way to do that in AL.
3. The LFE expects +10 dB acoustic gain. To mix properly with the center channel and system, all non-LFE input signals must be dropped 10 dBFS. The +10 dB amplifications of LFE is now NOT used in the Setup. If we crank the volume going into our convolver to 100%, only the channels mixed with the LFE signal can go to 0 dBFS, all the others max at -10 dBFS. This should be considered when setting the maximum gains of the filters. Filters without LFE signal content can safely go to +10 dBFS.
4. Good to know
5. OK, I'll post any bugs I can confirm.

 

[juicehifi]

2 So you want mono-bass. Why not just have a separate subwoofer and route all speakers through it? It will look different than your solution but it will do what you want to get done. You can send the LFE to one or two or three of these speakers. One may be the best bet here, since you have some phase differences between the speakers even after correction.

Something to think about: By handling the LFE this way you may end up hearing stuff that the mastering engineer did not hear ... provided he mixed with a regular subwoofer in the studio.

 

[hulkss]

2 So you want mono-bass. Why not just have a separate subwoofer and route all speakers through it? It will look different than your solution but it will do what you want to get done. You can send the LFE to one or two or three of these speakers. One may be the best bet here, since you have some phase differences between the speakers even after correction.

I don't think I want or have mono bass (I do want mono sub-bass). All front speaker content below 80 Hz (which can't be localized) is going to a 5 driver sub array. Left, right, and center keep all bass content above 80 Hz which can be localized. The LFE, which is mono by definition, is also sent to the entire sub array with content above 80 Hz going to the Center channel where it can be localized properly and played with improved fidelity. The sub array is just a big subwoofer.

 

[hulkss]

Something to think about: By handling the LFE this way you may end up hearing stuff that the mastering engineer did not hear ... provided he mixed with a regular subwoofer in the studio.

I was initially concerned about this. I used BEQdesigner to analyze the bass content of a few dozen multichannel recordings and I have used JRiver DSP to solo the LFE channel of many recordings. I have yet to find a recording that is degraded by the process I am using. I did find one movie that had high frequency "garbage" on the LFE channel. It could only be salvaged with a low pass filter on the LFE channel.

 

[HerbertWest]

This thread is super interesting. It is really great to read this constructive discussion about how to best use Audiolense for such a complex rig, and for multi-subs in general.
I have a question: not sure if I fully understand "2. Ability use a fullrange speaker as part of a sub array. (I pick "small" and route a sub driver channel in my MOTU mixer)". If you don't mind, I'd like to check if it can apply to my case.

I have a simple "2.0" setup; the L/R channels are configured in Audiolense as full-range speaker with 2 drivers each:
1) the speaker itself (crossover at 70 hz);
2) a 12" subwoofer.
Overall, a pair of speaker and stereo subs, but corrected in AL by pairing each speaker with its own 12" dedicated sub. All of this is driven by a RME Digiface.

Given the promising results I am getting with stereo listening, I was considering adding the rest of my 7.1 system without removing the subs from their stereo config (basically, part of a full range L/R speaker). My surround speakers are "small".

My reading is that I have three options:
Option 1) I could simply create an "artificial" subwoofer channel in my RME card, and bind it to the OUT channels driving the "real" subwoofers via RME TotalMix.This would be transparent to Audiolense, which would see it as a sub.
Option 2) "2. Ability use a fullrange speaker as part of a sub array." - if this feature were implemented, AL could probably route the surround bass + LFE to the L/R speakers (specifically, the subwoofer drivers of these "artificial" speakers), and achieve the same result as 1)
Option 3) Add a third physical subwoofer, independent from the current two bound to L/R channels, that takes care of surround bass + LFE.

Am I reading option 2 right?

 

[hulkss]

You have a pair of two-way speakers and two subs if I read you correctly. The woofer of each main speaker may be used in a sub array like I did. Set-up two two-way small speakers and a four driver sub in Audiolense. The two woofers and two subs are the sub drivers. You will need to mix the left "woofer sub" signal and left "two-way woofer" signals in your RME as they go to the same physical driver. Same on the right side. If your two-way woofers are weak on the bottom, you can high pass them like I did with my Khorns.

 

[jjazdk]

I think what you want is achieved by this channel routing in AL.

This thread is super interesting. It is really great to read this constructive discussion about how to best use Audiolense for such a complex rig, and for multi-subs in general.
I have a question: not sure if I fully understand "2. Ability use a fullrange speaker as part of a sub array. (I pick "small" and route a sub driver channel in my MOTU mixer)". If you don't mind, I'd like to check if it can apply to my case.

I have a simple "2.0" setup; the L/R channels are configured in Audiolense as full-range speaker with 2 drivers each:
1) the speaker itself (crossover at 70 hz);
2) a 12" subwoofer.
Overall, a pair of speaker and stereo subs, but corrected in AL by pairing each speaker with its own 12" dedicated sub. All of this is driven by a RME Digiface.

Given the promising results I am getting with stereo listening, I was considering adding the rest of my 7.1 system without removing the subs from their stereo config (basically, part of a full range L/R speaker). My surround speakers are "small".

My reading is that I have three options:
Option 1) I could simply create an "artificial" subwoofer channel in my RME card, and bind it to the OUT channels driving the "real" subwoofers via RME TotalMix.This would be transparent to Audiolense, which would see it as a sub.
Option 2) "2. Ability use a fullrange speaker as part of a sub array." - if this feature were implemented, AL could probably route the surround bass + LFE to the L/R speakers (specifically, the subwoofer drivers of these "artificial" speakers), and achieve the same result as 1)
Option 3) Add a third physical subwoofer, independent from the current two bound to L/R channels, that takes care of surround bass + LFE.

Am I reading option 2 right?

 

[juicehifi]

I was initially concerned about this. I used BEQdesigner to analyze the bass content of a few dozen multichannel recordings and I have used JRiver DSP to solo the LFE channel of many recordings. I have yet to find a recording that is degraded by the process I am using. I did find one movie that had high frequency "garbage" on the LFE channel. It could only be salvaged with a low pass filter on the LFE channel.

I just published a new release of Audiolense that handles your partial subwoofer crossover configuration properly. It is looking pretty good.

 

[jjazdk]

I just published a new release of Audiolense that handles your partial subwoofer crossover configuration properly. It is looking pretty good.

Interesting..
What improvements have you made to a "partial subwoofer crossover configuration", I did not find anything in the release note apart from bug fixes?

 

[juicehifi]

It was a bug fix. that enabled hulkss to high pass two subs that ran in parallel with three others.

Btw, jjazdk, did you read my examination of hulkss three subwoofer configurations above? And if so what do you think of my conclusion?

 

[hulkss]

I think what you want is achieved by this channel routing in AL.

View attachment 39231

My issue is getting front left sub-bass into the front right speaker and vice-versa. The front left and right act as full range & sub but I could not reroute bass from a full range + sub speaker.

I have them now as small speakers. I can't rout bass to them, so, I do that in my MOTU mixer as a work around.

 

[hulkss]

I just published a new release of Audiolense that handles your partial subwoofer crossover configuration properly. It is looking pretty good.

Great!! Thank you for the fantastic software support!! My Audio system is getting better and better with Audiolense.

 

[jjazdk]

My issue is getting front left sub-bass into the front right speaker and vice-versa. The front left and right act as full range & sub but I could not reroute bass from a full range + sub speaker.

I have them now as small speakers. I can't rout bass to them, so, I do that in my MOTU mixer as a work around.

My reply was to HerbertWest's setup, not yours :-)

Regarding your setup, I agree.. I have somewhat similar challenges in my own setup.