Created a rational resample MATLAB/Octave function

ratresamp.m

@@ -0,0 +1,64 @@
+function output = ratresamp (signal, stretchFactor, squeezeFactor)
+% -- ratresamp (S, L, M)
+%
+% Resamples a matrix S where each column represents one channel of a
+% sampled signal.
+%
+% The signal will be stretched by a factor of L and squeezed by a
+% factor of M. Thus the new sample rate will be L/M times the original
+% sample rate. Both L and M must be integers greater than zero.
+%
+% The frequency spectrum of the signal will be filtered to prevent
+% aliasing from occurring.
+
+
+if (nargin != 3)
+
+	error([...
+		"Invalid call to ratresamp. Correct usage is:"...
+		"\n\n -- ratresamp (S, L, M)"...
+	]);
+
+elseif (...
+	stretchFactor != round(stretchFactor) || ...
+	squeezeFactor != round(squeezeFactor) || ...
+	stretchFactor < 1 || squeezeFactor < 1 ...
+)
+
+	error("L and M must be integers greater than zero.")
+
+end
+
+
+inputLength = length(signal)(1);
+outputLength = inputLength*stretchFactor/squeezeFactor;
+
+
+	% ## STRETCHMENT STAGE
+
+	% Inserts L - 1 zeros between each sample
+	stretchedInput((1:inputLength)*stretchFactor,:) = stretchFactor*signal(1:inputLength,:);
+
+
+	% ## FILTRATION STAGE
+	outputFft = fft(stretchedInput);
+
+	% Depending on which is greater, applies a low-pass filter...
+	if (stretchFactor > squeezeFactor);
+		%...to remove spectrum images...
+		passband = round(inputLength/2);
+	else
+		%...or to prevent aliasing.
+		passband = round(outputLength/2);
+	end
+	% This way, both spectrum and aliasing are removed with just one filter
+	outputFft(passband:end - passband,:) = 0;
+	stretchedInput = real(ifft(outputFft));
+
+
+	% ## SHRINKAGE PHASE
+
+	% Takes 1 out of M samples in the stretched signal
+	output(1:outputLength,:) = stretchedInput((1:outputLength)*squeezeFactor,:);
+
+endfunction