acoustic simulation with AuVi 0.7

[Camilla Tetti]

Hi everyone, I’m working on my Master’s thesis involving the acoustic simulation of a recording studio, and I’ve run into a few geometric modeling issues.

First, I’m unsure about the best way to handle doors and windows. To maintain a manifold (watertight) mesh, should I model them as integrated parts of the shell or as separate intersecting geometries? 

Second, I have several floating ceiling clouds and cylindrical diffusers. Since the ceiling panels are suspended, both faces are exposed. I’m having trouble with surface normal orientation—the software seems to ignore reflections on the 'back' side of the planes. What is the best practice here? Should I give them a small thickness or is there a way to make the surfaces double-sided?

Thank you very much in advance. 

Here's a photo of the space I have to model and simulate. Of course the ceiling is missing now. 

[VI-Suite]

Hi.

To answer your first question, on the one hand your glazing surfaces could simply be faces in line with the walls as the window reveals probably won't have a significant impact on the results, on the other, as you don't seem to have any complex window frame geometry in your model, you could simple extrude out your in line window surfaces to create your reveals.

For the second question the way I generally I recommend is to do boolean difference operations on the room shell with your internal geometry. If both the room shell and interior objects are manifold then you can do a manifold boolean difference and the interior geometry will be taken away from the room volume and leave you with a new manifold volume of the room that now contains the surfaces of the interior geometry. Use the 'transfer' option for the materials in the "Solver Options" section of the boolean modifier if you want the materials from the interior geometry to propagate to the new surfaces on the shell mode. 

Cheers

Ryan 

[VI-Suite]

Just one other thing to keep in mind.

A bug was found in pyroomacoustics, and for now you should keep your scatter coefficients to 0. This bug has been fixed in pyroomacoustics 0.9.0, but the VI-Suite is still using 0.77. The update to 0.9.0 should happen with my next master commit, but you will need to reinstall the VI-Suite to update the version.

Cheers

Ryan

[Camilla Tetti]

Thank you for your answer. I'm gonna try what you suggested me. 
Do you already know when your gonna commit the new version updates?

Cheers

Camilla

[VI-Suite]

They were committed yesterday to master branch. I've only tested it on linux and Mac so far.

[VI-Suite]

Seems pyroomacoustics 0.9.0 breaks Blender 5.0 on windows. 5.1 alpha works though and when I replace msvcp files in 5.0's Blender.crt folder with the ones in the 5.1's folder it works. Whether this creates other problems though I don't know.

[Camilla Tetti]

Using pyroomacoustics 0.9.0 on Blender 4.5.5 LTS on Apple M3 it works!

I just noticed that when I save the IR, after generating it from the AuVi simulation node, that the exporting setting are 16kHz. 

I'm trying to make a comparison 

between the IR measured in the real recording studio and the IR generated from blender simulation. 

Do you know if there is a way to export at 44kHz the IR files? 

Otherwise is there a chance to obtain an excel showing the acoustical parameters measured from the geometry node? 

Because I can see only the RT60 showing in the VI metrics.
Thank you very much!

[VI-Suite]

Ah. Very good.

To simulate at 44kHz you need another octave band with a centre frequency at 16kHz. You can change all the references to 16000 with 44000 in the vi_operators file in the AuVi related operators. Add 16000 to the c_freqs list, replace all 7s with number 8s around line 4500 in the vi_operators file and add a number to the Lsf list, although I don't know what that number is it's only needed for STI calculation. You would probably have to do the same in the auvi_func file too.

I have added TSL and STI display to the metrics node. Should be in my next commit to master. For now you can create mesh faces at your microphone locations, specify their material as an LiVi/AuVi sensor, and after simulation do a point numerical display of TSL/STI for those faces.

Cheers

Ryan

[Khamilla]

Hi Ryan,
Thank you for your reply!

I tried implementing the changes you suggested, except for the Lsf list, because I wasn’t sure what value to add there. After making the modifications, I encountered this error.

Do you have any idea what might be causing it?

Also, I couldn’t find the corresponding settings to modify in the auvi_func file. Could you clarify where exactly those changes should be applied?

Thanks again for your help.

Best regards,
Camilla

--
You received this message because you are subscribed to the Google Groups "VI-Suite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to vi-suite+u...@googlegroups.com.
To view this discussion, visit https://groups.google.com/d/msgid/vi-suite/2637e536-7bc7-493a-953a-22eddb9b181dn%40googlegroups.com.

[VI-Suite]

Looks like you have either not added 16000 to the c_freqs list, or have not changed the 7's to 8's around line 4500 for the scattering data, or you have set different scattering values for the octave bands of your material. If it's the latter, then thinking about it, you will have to turn on Flat response. Turn on Flat response for the absorption data too.

The Lsf list will need an additional value but if you're not doing an STI analysis the number can be anything. If you are, you will have to find out what this number is. In the  auvi_func file you will find the ISO standard that details Lsf numbers.

In the auvi_func file there are references to 16000 that your should change to 44000 and there is another Lsf list that the additional number will also have to be added to.

Cheers

Ryan

[Khamilla]

Hello again Ryan,

I updated the VI-Suite add-on using your latest commit.

Before installing it (with the previous add-on version), I finally managed to obtain the IR at 44.1 kHz by setting a flat response for the scattering coefficients of all materials, as you suggested.

However, after installing the version from your recent commit and repeating the same modifications to the Python code, I’m now getting the following error in the screenshot.

Do you think this could be related to a conflict with the STI calculation?

Cheers,
Camilla

To view this discussion, visit https://groups.google.com/d/msgid/vi-suite/edaaa701-82be-4371-a775-620dbc6b5ce5n%40googlegroups.com.

[VI-Suite]

HI.

Yes, it's because I added STI to the metrics node. As you're just generating impulse responses you can comment out that line with a #.

Cheers

Ryan

[Khamilla]

Hello Ryan

I was wondering, since I'm comparing this Add-on with Odeon, if there is a chance to have the wav at 44.100 but with the possibility to not have flat data on absorption and scatter of the materials of the mesh.

If you could tell me if there is a way to fix this problem would be so helpful. 

Cheers

Camilla T.

To view this discussion, visit https://groups.google.com/d/msgid/vi-suite/c33d7477-8535-48ed-8531-4b2b1f17f276n%40googlegroups.com.