MatlabAudioBuffer

This class aims at providing a convenient interface for working with block-wise (audio) signals. It can be used with single or multi-channel signals and handles audio files and signal vectors/matrices. It inherits from matlab.System and will, in a future release, be usable as a SIMULINK model. At the moment, some infos are missing to be included in SIMULINK.

Dependencies

No dependencies to toolboxes.

The code has been tested with MATLAB R2015a/15b/16a/16b on Windows 7 and Xubuntu 15.10 and 16.04 LTS.

Installation

Put the file AudioBuffer.m into your MATLAB path or the directory of your MATLAB project.

Usage

The usage of the class is straight forward. The user can specify a signal vector or matrix with corresponding sampling rate or a filename. The supported file formats depend directly on those which the audioread() function of the user's MATLAB version supports. These include typically wav, flac and mp3.

The following example shows a typical usage of the class.

obj = AudioBuffer(...
    signal, sampleRate, ...
    'BlockLengthSec', 32e-3, ...
    'OverlapRatio', 0.5, ...
    'IdxChannels', 1, ...
    'WindowFunction', @(x) hann(x, 'periodic') ...
    );
    
% or:
% obj = AudioBuffer( ...
%     filename, [], ...
%     'BlockLengthSec', 32e-3, ...
%     'OverlapRatio', 0.5, ...
%     'IdxChannels', 1, ...
%     'WindowFunction', @(x) hann(x, 'periodic') ...
%     );

numBlocks = obj.NumBlocks;
blockSize = obj.BlockSize;

signalBlocks = zeros(blockSize, numBlocks);
for iBlock = 1:numBlocks
    % iteratively return one signal block
    signalBlocks(:, iBlock) = obj.step();
end

% If desired, use the internal WOLA method to
% reconstruct the unblocked signal
reconstructedSignal = obj.WOLA(signalBlocks);

% this is the error between original
% and reconstructed signal.
norm(signal - reconstructedSignal)^2

% ans =
%
%      0.0021

License

The code is distributed under BSD 3-Clause.