How can I read .ogg datasets and apply LSTM classification (my code included)?

1 vue (au cours des 30 derniers jours)
Pooyan Mobtahej
Pooyan Mobtahej le 25 Oct 2020
I need to read .ogg audio files (datasets) in Matlab, get its size for crreating array and and apply LSTM classification. I have 2 classes normal and anomalies. Could you give me a code to be able to read audio files and do classification? You can separate our data into three parts. For example, 80% of all normal and anomaly signals for training (2 classes), 10% for validation, and 10% for testing.
You can create sample sounds .ogg files for your code. And the code shall be able to read different audio datasets.
I have used the following code but need advise for me proper modification:
normal = zeros(100,1000); %return matrix
anomaly = zeros(100,1000);
%assuming you have 100 audio files of 1000 samples each, for both normal and anomaly classes,
%you can use a loop to read the files and split the data for training / Validation / Testing.
for i = 1:100
normal_name = strcat('normal_',num2str(i),'.ogg'); %preserves them in cell arrays
anomoly_name = strcat('anomaly_',num2str(i),'.ogg');
%files named for anomaly_0 to anomaly_100
normal(i) = audioread(normal_name);
anomaly(i) = audioread(anomaly_name);
end
%The above arrays can be split for training/Validation/Testing data set
%Normal=NNoice
%Anomaly=ANoice
%NNoise = 2*rand([100,1000]) - 1; %changed N
NLabels = repelem(categorical("normal"),100,1);
ANoise = filter(1,[1,-0.999],NNoise);
ANoise = ANoise./max(abs(ANoise),[],'all');
ALabels = repelem(categorical("anomaly"),100,1);
%pNoise = pinknoise([N,1000]);n%removed pink noices
%Labels = repelem(categorical("pink"),1000,1)
sound(NNoise(:,1),fs)
melSpectrogram(NNoise(:,1),fs)
title('Normal')
sound(ANoise(:,1),fs)
melSpectrogram(ANoise(:,1),fs)
title('Anomaly')
%sound(pNoise(:,1),fs)
%melSpectrogram(pNoise(:,1),fs)
%title('Pink Noise')
featuresTrain = extract(aFE,audioTrain);
[numHopsPerSequence,numFeatures,numSignals] = size(featuresTrain)
audioTrain = [NNoise(:,1:80),ANoise(:,1:80)];%change
labelsTrain = [ALabels(1:80);NLabels(1:80)];
audioValidation = [NNoise(:,81:end),ANoise(:,81:end)]; %changed according 100 samples
labelsValidation = [NLabels(81:end);ALabels(81:end)];
aFE = audioFeatureExtractor("SampleRate",fs, ...
"SpectralDescriptorInput","melSpectrum", ...
"spectralCentroid",true, ...
"spectralSlope",true);
featuresTrain = permute(featuresTrain,[2,1,3]);
featuresTrain = squeeze(num2cell(featuresTrain,[1,2]));
numSignals = numel(featuresTrain)
[numFeatures,numHopsPerSequence] = size(featuresTrain{1})
featuresValidation = extract(aFE,audioValidation);
featuresValidation = permute(featuresValidation,[2,1,3]);
featuresValidation = squeeze(num2cell(featuresValidation,[1,2]));
layers = [ ...
sequenceInputLayer(numFeatures)
lstmLayer(50,"OutputMode","last")
fullyConnectedLayer(numel(unique(labelsTrain)))
softmaxLayer
classificationLayer];
options = trainingOptions("adam", ...
"Shuffle","every-epoch", ...
"ValidationData",{featuresValidation,labelsValidation}, ...
"Plots","training-progress", ...
"Verbose",false);
net = trainNetwork(featuresTrain,labelsTrain,layers,options);
NNoiseTest = 2*rand([100,1]) - 1;
classify(net,extract(aFE,NNoiseTest)')
ANoiseTest = filter(1,[1,-0.999],ANoiseTest);
ANoiseTest= ANoiseTest./max(abs(ANoiseTest),[],'all');
classify(net,extract(aFE,ANoiseTest)')
%pNoiseTest = pinknoise(N);
%classify(net,extract(aFE,pNoiseTest)')

Réponses (0)

Catégories

En savoir plus sur AI for Signals dans Help Center et File Exchange

Produits


Version

R2020b

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by