Merge branch 'ObsFiltUpdate' into 'main'

SNR in observation filter calculation feature and repository version update

See merge request in-nova/in-nova!1

CHANGELOG

-v0.1.0
-This is the initial version of the project.
\ No newline at end of file

CHANGELOG.md

+### v0.2.0 ###
+**Utilities**
++ Added a single papameter bisection method.
+* Improved rotation matrix calculation function with degrees and radian calculations available.
++ Added function to calculate a double-sided spectrum from a single-sided spectrum.
++ Added a function to calculate the intersection of three vectors.
++ Added a pair of functions to check for even or odd elements.
++ Added a function to pick values from a vector based on given probabilities.
++ Added a function to pick randomly unique rows from a matrix.
++ Added a Heaviside step function.
++ Added function to swap argument values. 
+
+**Optimisation**
++ Added a MATLAB Memetic Algorithm (MA) implementation.
+* The algorithm calls provided functions and can solve generic problems.
+
+**Control**
++ Added tonal control in the frequency domain.
+* Control is contained in a single function.
+* Can be used with or without virtual sensing.
+* Implementations of optimal control and FxLMS (still frequency domain) calculations are available.
+
+**Signal Processing - Array Processing**
++ Added array manifold (steering vector) calculation function.
+
+**Signal Processing - Generic**
++ Added frequency band calculation function with 1/1 octave and 1/3 octave bands.
++ Add fractional delay filter impulse response generation function.
+
+**Virtual Sensing - Generic**
++ Added a multiple coherence calculation function.
+
+** Virtual Sensing - Remote Microphone Technique**
+* Divide the observation filter and estimation with observation filter in two functions. The return arguments are split appropriately to the function they relate.
++ Added option to include noise in the monitoring microphone power spectral density matrix with specified SNR value.
+
+**Sound Fields**
++ Added plane wave calculation function.
++ Added Circular Harmonics calculation function.
++ Added Spherical Harmonics calculation function.
++ Added Discrete Circular Fourier Transform (DCFT) calculation function.
++ Added Inverse Discrete Circular Fourier Transform (IDCFT) calculation function.
++ Added Discrete Spherical Fourier Transform (DSFT) calculation function.
++ Added Inverse Discrete Spherical Fourier Transform (IDSFT) calculation function.
++ Added function to extrapolate a sound field in the Spherical Harmonics domain.
++ Added function to calculate sound field generated by a point source in the Spherical Harmonics domain (truncated order).
++ Added function to calculate sound field generated by a plane wave in the Spherical Harmonics domain (truncated order).  
+
+
+### v0.1.0 ###
+* This is the initial version of the project.

README.md

@@ -72,7 +72,7 @@ The project is licensed under the ***MIT License***, which is a rather unrestric
 
 
 ## Versioning ##
-The project uses [semantic versioning](https://semver.org/) and the current version is **v0.1.0**.
+The project uses [semantic versioning](https://semver.org/) and the current version is **v0.2.0**.
 
 
 #### **Important**

Virtual Sensing/README.md

@@ -40,7 +40,7 @@ Pm = ptSrcField(vPos, mPos, 7.5e2); % Primary-to-monitoring transfer function (f
 Pe = ptSrcField(vPos, ePos, 7.5e2); % Primary-to-virtual transfer function (f = 750 Hz)
 
 % Perform estimation and acquire the normalised square estimation error
-Oopt = obsFilt(Pe, Pm, [], 1e2); % Use regularisation factor 100 to calculate the observation filter
+Oopt = obsFilt(Pe, Pm, [], 1e2, 15); % Use regularisation factor 100 to calculate the observation filter and add noise with Signal-to-Noise Ratio of 15 dB.
 [~, ~, ~, normSqrErr] = obsFileEst(Pm, Oopt, Pe, []); % Perform estimation
 ```
 

Virtual Sensing/Remote Microphone Technique/MATLAB/Functions/obsFilt.m

@@ -3,7 +3,7 @@
 % Author: Achilles Kappis
 % e-mail: axilleaz@protonmail.com
 %
-% Date: 15/08/2024 (DD/MM/YYYY)
+% Date: 01/09/2024 (DD/MM/YYYY)
 %
 % Copyright: MIT
 % --------------------------------------------------
@@ -36,6 +36,17 @@
 %                               regularisation factor for all microphones.
 %                               [Default: 0]
 % 
+% snrVal [numeric] (Optional): This is the Root-Mean-Square (RMS)
+%                              Signal-to-Noise Ratio (SNR) in the
+%                              monitoring microphone signals. This must
+%                              be either a scalar indicating the common
+%                              SNR value of the microphone signals, or a
+%                              vector of dimensions Kx1, containing the SNR
+%                              values of each microphone. The values must
+%                              be real or Inf for the noiseless case. See
+%                              the Notes below for references on how this
+%                              value is calculated. [Default: Inf]
+% 
 % --------------------------------------------------
 % Output
 % 
@@ -59,12 +70,16 @@
 % --------------------------------------------------
 % Notes
 % 
+% - The SNR at the monitoring microphones calculation is based on the
+%   paper: "Combining the remote microphone technique with head-tracking
+%   for local active sound control" by W. Jung, S. J. Elliott and J. Cheer.
+% 
 % --------------------------------------------------
-function [oOpt, Sme, Smm, condNum] = obsFilt(Pe, Pm, srcCsd, regFacs)
+function [oOpt, Sme, Smm, condNum] = obsFilt(Pe, Pm, srcCsd, regFacs, snrVal)
     % ====================================================
     % Check for number of arguments
     % ====================================================
-    narginchk(2, 4);
+    narginchk(2, 5);
     nargoutchk(0, 4);
 
     % ====================================================
@@ -92,6 +107,21 @@ function [oOpt, Sme, Smm, condNum] = obsFilt(Pe, Pm, srcCsd, regFacs)
         regFacs = 0;
     end
 
+    if nargin > 4 && ~isempty(snrVal)
+        validateattributes(snrVal, "numeric", {'vector', 'real', 'nonnan', 'nonempty'}, mfilename, "Normalised Root-Mean-Square Singal-to-Noise Ratio of the monitoring microphones", 5);
+
+        % Make sure snrVal has correct dimensions
+        if ~isscalar(snrVal) && numel(snrVal) ~= size(Pm, 1)
+            error("SNR must be either a scalar or its length must match the number of monitoring microphones.");
+        end
+
+        if sum(snrVal == -Inf) > 0
+            error("SNR value cannot be equal to -Inf");
+        end
+    else
+        snrVal = Inf;
+    end
+
 
     % ====================================================
     % Calculate optimal filters
@@ -102,14 +132,24 @@ function [oOpt, Sme, Smm, condNum] = obsFilt(Pe, Pm, srcCsd, regFacs)
     Sme = Pe * tmpVal; % Virtual-Monitor mics cross-spectra
     Smm = Pm * tmpVal; % Monitor-Monitor mics cross-spectra
 
+    % Regularisation matrix
     if isscalar(regFacs)
-        regMat = eye(size(Smm)) .* regFacs; % Regularisation matrix
+        regMat = eye(size(Smm)) .* regFacs;
     else
         regMat = diag(regFacs);
     end
 
+    % Signal-to-Noise Ratio at the monitoring microphones
+    if ~isinf(snrVal)
+        snrVal = 10^(-snrVal/10); % Convert deciBel to linear
+        snrMtx = (snrVal.^2) .* norm(Smm, 'fro')/(size(Pm, 1)^2); % Calculate the appropriate SNR amplitude values
+    else
+        snrMtx = 1;
+    end
+    snrMtx = snrMtx .* eye(size(Pm, 1));
+
     % Calcualte observation filters
-    invQty = Smm + regMat;
+    invQty = Smm + snrMtx + regMat;
     oOpt = Sme/invQty;
 
     % Condition number of auto-correlation (power spectrum) matrix