DIY 3D Speaker Scanner - the Mathematics and Everything Else

[Kessito]

...and build in algorithms (spericalcapExtrapolation, what Geddes does whit his software) to do upscaling of sparse measured data; in plots below 4th order upscaled to 35th order
Oh man, this is so great!

Basically, what is missing for a fully functional system:
1 - adapting the Bessel funtions to Hankel functions, so we can do SFS (this one is for you NTK!)
2 - adapting the inverse functions so htat he Hankel functions can be seperated(?)
3- automated measurments with motor control, I have this already
4 - windowing for the combining with direct measruement for the HF (I have this covered, already build that years ago)
5 - automated sticth the combined responses (HF and SFS
6 - tying it all together, which will be relatively low effort, since the data format is well defined and well documented, the rest is ALL THERE!!
7 - making a gui; I propose to making an executable (so you dont need matlab) with xml file control, than any gui wizzards oout there can easily make a nice gui

 

[mwmkravchenko]

BOOM:
https://www.ak.tu-berlin.de/menue/publications/open_research_tools/aktools/

TU Berlin has a MATLAB toolbox which has almost ALL needed functionality onboard!!!

INCLUDING the transformating from magnitude/phase data to the spherical harmonics and back!! (basically you can already do everything but thesoundfield seperation)

this means that the only piece NTK has to add (modify some of these functions) is the part where the secondary soundfield spherical harmonics are created and subtracted from the direct part!

And the good news: this is all highly verfied/documented open source/open use university research code

SUPER EXCITED WITH TIS FIND!!!!!!

@NTK: no excuses left now for you

Can you see my happy dance from there? I'

Yeah! working; HRTF measurments vs the same measurments transformed to spherical N=10 and back!

View attachment 118980



And N=35 and back , and its fast too..

View attachment 118979

Kool!

 

[mwmkravchenko]

BOOM:
https://www.ak.tu-berlin.de/menue/publications/open_research_tools/aktools/

TU Berlin has a MATLAB toolbox which has almost ALL needed functionality onboard!!!

INCLUDING the transformating from magnitude/phase data to the spherical harmonics and back!! (basically you can already do everything but thesoundfield seperation)

this means that the only piece NTK has to add (modify some of these functions) is the part where the secondary soundfield spherical harmonics are created and subtracted from the direct part!

And the good news: this is all highly verfied/documented open source/open use university research code

SUPER EXCITED WITH TIS FIND!!!!!!

@NTK: no excuses left now for you

I'm a Mathcad nitwit. To many other things are taking up my meeger amount of claping brain cells. But I know what you are taking about. Oh this is good. Verified how is my QC OCD side coming up with the inevitable questions.

Just great. Now I have to learn how to use Mathlab. (thanks FEW)

Hey a little epiphany moment. Have you MathCAD mavens tried out the code on Octave? Free, supposed to be syntax compatible? Or is the software our benefactor working on stand alone?

 

[Few]

Super-exciting indeed. Plus I agree it would be interesting to know if it works on Octave (just to avoid confusion, I’m sure mwmkravchenk meant Matlab rather than MathCAD—an easy typo to make). Good to see an old diyAudio buddy over here, by the way!

 

[mwmkravchenko]

Super-exciting indeed. Plus I agree it would be interesting to know if it works on Octave (just to avoid confusion, I’m sure mwmkravchenk meant Matlab rather than MathCAD—an easy typo to make). Good to see an old diyAudio buddy over here, by the way!

Super-exciting indeed. Plus I agree it would be interesting to know if it works on Octave (just to avoid confusion, I’m sure mwmkravchenk meant Matlab rather than MathCAD—an easy typo to make). Good to see an old diyAudio buddy over here, by the way!

Only a little tired from today's fun and games. Thanks for the gentle Wake up!

 

[Kessito]

Only a little tired from today's fun and games. Thanks for the gentle Wake up!

You are of course free to try it yourself, i think not because of the dlls that are used for measuring.
I will not do it, my time is sparse and I dont feel like learning the ins and outs of another software.
I think making an executable( or serie of executables) which can be configured with a xml or txt file will be most convinient, than you dont have to install either mathlab or octave

 

[mwmkravchenko]

You are of course free to try it yourself, i think not because of the dlls that are used for measuring.
I will not do it, my time is sparse and I dont feel like learning the ins and outs of another software.
I think making an executable( or serie of executables) which can be configured with a xml or txt file will be most convinient, than you dont have to install either mathlab or octave

Sensible. I agree if there are dll call outs it will not work in Octave. There will be a lot of issues for sure.

The stand alone idea is awesome. Isn't XML cross platform? Browser based?

 

[somebodyelse]

It looks like Octave might be missing some required bits:

warning: AKtools needs the Signal Processing Toolbox which is currently not installed. This might limit the functionality of AKtools.
warning: called from
    AKtoolsStart at line 106 column 9
warning: AKtools needs the Statistics and Machine Learning Toolbox which is currently not installed. This might limit the functionality of AKtools.
warning: called from
    AKtoolsStart at line 106 column 9
warning: the 'userpath' function is not yet implemented in Octave

Someone with some Octave experience may know whether this is likely to be something that can be worked around, or a showstopper.

 

[Few]

I noticed in the AKtools readme file that they welcome suggestions for new functionality:
"We welcome any feedback on bugs inside AKtools as well as features that could be added. You can contact us via [email protected]-berlin.de"
If they knew of this group's interest, perhaps they'd be willing to add in some of the missing pieces Kessito identified. It would be great to have everything fully integrated. Quite a find in any case!

Few

 

[Few]

They also have a white paper on the package but it requires AES membership or $$ for access. I'd love a copy if anyone would like to share...

 

[Kessito]

I got (digitally) this book which is referened to a lot in the toolkit:
Fundamentals of Spherical Array Processing

Book published 2019 in Springer Topics in Signal Processing

Authors: Boaz Rafaely

https://doi.org/10.1007/978-3-319-99561-8

Is is not a download link, it is an actual book you'd have to buy, it's illegal for me to share my copy

Highly recommended!

 

[Kessito]

An extra cool additional find in the toolbox:
Because of the hrtf driven background of the research, it has the tools to virtually listen around your speakers IN a defined room with absorbtion coefficients, while moving around in realtime!

 

[Valhalla]

I got (digitally) this book which is referened to a lot in the toolkit:
Fundamentals of Spherical Array Processing

Book published 2019 in Springer Topics in Signal Processing

Authors: Boaz Rafaely

https://doi.org/10.1007/978-3-319-99561-8

Is is not a download link, it is an actual book you'd have to buy, it's illegal for me to share my copy

Highly recommended!

We have bought this book and got it in our library and I've never read it. I can lend it to anyone who thinks it will provide any kinda help to move forward in this project to take a look at. Just pm me.

 

[mwmkravchenko]

They also have a white paper on the package but it requires AES membership or $$ for access. I'd love a copy if anyone would like to share...

Point me in the right direction. I have that required membership.

 

[Few]

Does the link not work for you? I just tested it again and it works for me, but there may be some funny business I’m not catching.

 

[mwmkravchenko]

Does the link not work for you? I just tested it again and it works for me, but there may be some funny business I’m not catching.

Again I missed the forest for the trees!

OK I have the file. I'll figure out how to set up a shared file on my Google drive and put it up.

 

[mwmkravchenko]

Try this. I think I have it set up correctly.

https://drive.google.com/file/d/1YK9AxwTOnHEZek6qeA8mBLXNfnGsCtrJ/view?usp=sharing

 

[Few]

Many thanks! I have infinite sympathy/empathy as we navigate these tricky times. I appreciate your efforts! I’ll be super-interested to read what they have to say.

By the way, I tried to email the ak-tools folks to alert them to this thread, but my email was rejected. I must have an email address typo hidden somewhere. I’ll keep at it.

Few

 

[Few]

Just tested the link. It works; thanks!

 

[mwmkravchenko]

Many thanks! I have infinite sympathy/empathy as we navigate these tricky times. I appreciate your efforts! I’ll be super-interested to read what they have to say.

By the way, I tried to email the ak-tools folks to alert them to this thread, but my email was rejected. I must have an email address typo hidden somewhere. I’ll keep at it.

Few

I always try to take a step back and take time to truly get a correct perspective on where I am at and what is going on. Have I been through worse, is it going to get better. Can I do something about it? Simple questions that lay to rest some things and make me think of preparations for others. We do what we can with what we have. And sometimes we even have some fun!