k-Wave

1. Chesteta
   

   Member
   Joined: Mar '11
   Posts: 4

   Hello, this is more of a 'general wave theory' question:
   I am trying to create head-related transfer functions (impulse responses) using CT scan data of my head, by setting a point source of pressure and measuring within the ear canal of the model. I figure it would be more computationally efficient to reverse the process and use many sensors placed around the head and create an impulse inside of the ear (so I can measure impulse responses from many angles at once), does this 'work' on the physics side of things?

   More plainly: can one swap an impulse source and the sensor in the model and achieve the same impulse response? - if not the same impulse response, would the frequency response at least be the same?

   From setting up sub-woofers, I know that one way to do it is to set the sub-woofer in the listening position and then crawl around and find where the bass 'sounds good' (has an even frequency response) - my thinking is that this applies to all frequencies...

   Thanks for your knowledge and help!

   Posted 12 years ago #

2. Bradley Treeby
   

   Developer
   Joined: Mar '10
   Posts: 1,643

   Hi Chesteta,

   The term you're looking for is "reciprocity", and the answer is yes. In fact, this idea is already used both experimentally and computationally to derive HRTFs. Two relevant references are here and here.

   Good luck with your simulations!

   Brad.

   Posted 12 years ago #

3. Chesteta
   

   Member
   Joined: Mar '11
   Posts: 4

   Thanks for the info! Will I get the same impulses directly, or do I need to do additional processing?

   Posted 12 years ago #

4. Bradley Treeby
   

   Developer
   Joined: Mar '10
   Posts: 1,643

   Hi Chesteta,

   You should get the same thing directly. Here's a simple example to try:

   clear all;

   % create the computational grid
   PML_size = 20;
   Nx = 128 - 2*PML_size;
   Ny = 128 - 2*PML_size;
   dx = 0.1e-3;
   dy = 0.1e-3;
   kgrid = makeGrid(Nx, dx, Ny, dy);

   % define the properties of the propagation medium
   medium.sound_speed = 1500 + (rand(Nx, Ny) - 0.5)*300;
   medium.density = medium.sound_speed./1.5;

   % define initial pressure distribution and binary sensor mask
   p0 = zeros(Nx, Ny);
   p0(Nx/4, Ny/4) = 1;
   sensor_mask = zeros(Nx, Ny);
   sensor_mask(3*Nx/4, 3*Ny/4) = 1;

   % set simulation options
   input_args = {'LogScale', true, 'DisplayMask', p0 | sensor_mask, ...
   'PMLInside', false, 'PlotPML', false};

   % assign the source and sensor and run the simulation
   source.p0 = p0;
   sensor.mask = sensor_mask;
   sensor_data_12 = kspaceFirstOrder2D(kgrid, medium, source, sensor, input_args{:});

   % switch the source and sensor and run the simulation again
   source.p0 = sensor_mask;
   sensor.mask = p0;
   sensor_data_21 = kspaceFirstOrder2D(kgrid, medium, source, sensor, input_args{:});

   % plot the impulse response
   figure;
   plot(sensor_data_12, 'k-');
   hold on;
   plot(sensor_data_21, 'r--');
   legend('Impulse Response P1 to P2', 'Impulse Response P2 to P1');

   Kind regards,

   Brad.

   Posted 12 years ago #

RSS feed for this topic

Reply

You must log in to post.