Add multiple aborption coefficients into virtual impulse response

Question

I've got some code which creates a virtual impulse response of a room with arbitrary dimensions. I want to know, how can I add more than one absorption coefficient into it?

function h = rir(fs, mic, n, r, rm, src)
    %   RIR   Room Impulse Response.
    %   [h] = RIR(FS, MIC, N, R, RM, SRC) performs a room impulse
    %         response calculation by means of the mirror image method.
    %
    %      FS  = sample rate.
    %      MIC = row vector giving the x,y,z coordinates of
    %            the microphone.  
    %      N   = The program will account for (2*N+1)^3 virtual sources 
    %      R   = reflection coefficient for the walls, in general -1<R<1.
    %      RM  = row vector giving the dimensions of the room.  
    %      SRC = row vector giving the x,y,z coordinates of 
    %            the sound source.
    %
    %   EXAMPLE:
    %
    %      >>fs=44100;
    %      >>mic=[19 18 1.6];
    %      >>n=12;
    %      >>r=0.3;
    %      >>rm=[20 19 21];
    %      >>src=[5 2 1];
    %      >>h=rir(fs, mic, n, r, rm, src);
    %
    %   NOTES:
    %
    %   1) All distances are in meters.
    %   2) The output is scaled such that the largest value of the 
    %      absolute value of the output vector is equal to one.
    %   3) To implement this filter, you will need to do a fast 
    %      convolution.  The program FCONV.m will do this. It can be 
    %      found on the Mathworks File Exchange at
    %      www.mathworks.com/matlabcentral/fileexchange/.  It can also 
    %      be found at http://www.sgm-audio.com/research/rir/fconv.m
    %   4) A paper has been written on this model.  It is available at:
    %      http://www.sgm-audio.com/research/rir/rir.html
    %      
    %
    %Version 3.4.2
    %Copyright  2003 Stephen G. McGovern

    %Some of the following comments are references to equations the my paper.

    nn = -n:1:n;                            % Index for the sequence
    rms = nn+0.5-0.5*(-1).^nn;              % Part of equations 2,3,& 4
    srcs = (-1).^(nn);                      % part of equations 2,3,& 4
    xi = srcs*src(1)+rms*rm(1)-mic(1);      % Equation 2 
    yj = srcs*src(2)+rms*rm(2)-mic(2);      % Equation 3 
    zk = srcs*src(3)+rms*rm(3)-mic(3);      % Equation 4 

    [i, j, k] = meshgrid(xi, yj, zk);           % convert vectors to 3D matrices
    d = sqrt(i.^2+j.^2+k.^2);               % Equation 5
    time = round(fs*d/343)+1;               % Similar to Equation 6

    [e, f, g] = meshgrid(nn, nn, nn);         % convert vectors to 3D matrices
    c = r.^(abs(e)+abs(f)+abs(g));          % Equation 9
    e = c./d;                               % Equivalent to Equation 10

    h = full(sparse(time(:), 1, e(:)));       % Equivalent to equation 11
    h = h/max(abs(h));                      % Scale output

end

At the moment, the variable 'r' is the absorption coefficient for the whole room. But for the assignment I'm doing, I'm taking the RT60 of a theoretical, cuboid room that has various absorption coefficients for various surfaces.

FYI, I've looked for the paper Stephen McGovern mentions. I've found a couple of copies on archive.org but none with images! Which is kind of useless because all the images have the relevant equations.

score 0 · Accepted Answer · answered Apr 16 '18 at 14:16

A few points:

This is very poor model to approximate a real room impulse response, so you you should make sure that it can meet your needs before putting work into it. I think this one here may be already better: https://www.mathworks.com/matlabcentral/fileexchange/25965-fast-simulation-of-acoustic-room-impulse-responses--image-source-method-
The notion of a frequency independent reverb time (or absorption) is seriously flawed. In order to get something remotely realistic, you need to bake in some frequency dependency. Brute force would to be calculate one RIR for each octave, band-pass filter and summing them up.
If you want to expand this code, you should probably look at equation 9.

A quick hack could like like this

c = sqrt(r(1)*r(2)).^abs(e) .* 
    sqrt(r(3)*r(4)).^abs(f) .*    
    sqrt(r(5)*r(6)).^abs(g)

where $r$ is a vector with the reflection coefficients for each wall matching the x,y,z coefficients. That will be "close enough" for higher order reflection but pretty wrong for lower order reflections, especially if the reflections of opposite walls are fairly different.

Thank you! That's what I was thinking of doing, re. brute force RIRs. How might I do this? Would following the steps detailed here work? Calculation of Reverberation Time (RT60) from the Impulse Response
Also, re. the reflection coefficient hack. Why are you multiplying the coefficient pairs together? — Sophie Fitz, Apr 17 '18 at 13:25
Brute force: follow the steps in my answer: run 10 models, one for each ocatve, then apply a bandpass filter and sum the results over all octaves. — Hilmar, Apr 17 '18 at 19:29
Each individual reflection is modeled by a single multiplication with r(n). Multiple reflections are just consecutive multiplications — Hilmar, Apr 17 '18 at 19:31

Add multiple aborption coefficients into virtual impulse response

1 Answers1