Lambda1k = lim(J,inf, 1/J*SUM(j=1 to J) norm(M1j(k))^2)

M1j and M2j are the DFT of noises n1(l) and n2(l), but what “j” means?

Can you help me understanding that formula?

Thank you very much!

So i don’t use the R matrix because i don’t need to change system coordinate, every microphones are the same coordinate system, so i simply use the formula l=t*a

To calculate (aj x ak) i’m using the cross function cross(aj,ak), is it right?

Reading Brandstain thesis, the t’s are the distance from the center of the array to the source location.
I calculate all possible intersections, i find all the “t”, but no one of this have a value close the real distance from the center of a mic and the source location.

Some example:
The real distances are about 4.1 – 4.8
The t’s values are about 14.9 – 40.8

Likely, your coordinate change is wrong. If you’re using the TDOA method, your initial solution is with respect to the center of the microphone cluster. I have a post that details how I handled this process. When I wrote this post, I had just taken a course in linear algebra, which helped a lot.

As you can tell, it’s been awhile since I did this, but if i remember correctly, taking the inverse of the a-coefficient matrix and multiplying it by the c-coefficients should give you the answer you’re looking for. A good linear algebra textbook should point you in the right direction (so to speak.)

When you estimate the location of the sourcem how you calculate Wjk?

I’ve tried it without noise, but the 4 line (from 4 set of microphones) doesn’t intersects, they go in 4 different directions….

I’d glad if you help me with this.

Thank you

Happy testing