Skip to content

Commit

Permalink
Read parameters from config file
Browse files Browse the repository at this point in the history
  • Loading branch information
@mkholghi
mkholghi committed Jul 18, 2018
1 parent 433ff34 commit c70b164
Show file tree
Hide file tree
Showing 4 changed files with 133 additions and 24 deletions.
147 changes: 128 additions & 19 deletions src/AnalysisPrograms/MahnooshSandpit.cs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -14,6 +14,7 @@ namespace AnalysisPrograms
using Accord.MachineLearning;
using Accord.Math;
using Accord.Statistics;
using Acoustics.Shared.ConfigFile;
using Acoustics.Shared.Csv;
using AudioAnalysisTools.DSP;
using AudioAnalysisTools.StandardSpectrograms;
Expand All @@ -36,6 +37,7 @@ public void Execute(Arguments arguments)
var inputPath = Path.Combine(inputDir, "PatchSamplingSegments");
var trainSetPath = Path.Combine(inputDir, "TrainSet");
var testSetPath = Path.Combine(inputDir, "TestSet");
var configPath = @"C:\Work\GitHub\audio-analysis\src\AnalysisConfigFiles\FeatureLearningConfig.yml";
var outputMelImagePath = Path.Combine(resultDir, "MelScaleSpectrogram.png");
var outputNormMelImagePath = Path.Combine(resultDir, "NormalizedMelScaleSpectrogram.png");
var outputNoiseReducedMelImagePath = Path.Combine(resultDir, "NoiseReducedMelSpectrogram.png");
Expand All @@ -52,19 +54,41 @@ public void Execute(Arguments arguments)

// get the nyquist value from the first wav file in the folder of recordings
int nq = new AudioRecording(Directory.GetFiles(inputPath, "*.wav")[0]).Nyquist;

int nyquist = nq; // 11025;
int frameSize = 1024;
int finalBinCount = 128; // 256; // 100; // 40; // 200; //
int hertzInterval = 1000;
FreqScaleType scaleType = FreqScaleType.Mel;

//+++++
var configFile = configPath.ToFileInfo();

if (configFile == null)
{
throw new FileNotFoundException("No config file argument provided");
}
else if (!configFile.Exists)
{
throw new ArgumentException($"Config file {configFile.FullName} not found");
}

//var configuration = ConfigFile.Deserialize<AcousticIndices.AcousticIndicesConfig>(configFile);
var configuration = ConfigFile.Deserialize<FeatureLearningConfig>(configFile);
//+++++

int frameSize = configuration.FrameSize; // 1024;
int finalBinCount = configuration.FinalBinCount; // 128; // 256; // 100; // 40; // 200; //
int hertzInterval = configuration.HertzInterval; // 1000;
FreqScaleType scaleType = configuration.FrequencyScaleType; // FreqScaleType.Mel;
var freqScale = new FrequencyScale(scaleType, nyquist, frameSize, finalBinCount, hertzInterval);
var fst = freqScale.ScaleType;

var sonoConfig = new SonogramConfig
{
WindowSize = frameSize,
// since each 24 frames duration is equal to 1 second

// the duration of each frame (according to the default value (i.e., 1024) of frame size) is 0.04644 seconds
// The question is how many single-frames (i.e., patch height is equal to 1) should be selected to form one second
// The "WindowOverlap" is calculated to answer this question
// each 24 single-frames duration is equal to 1 second
// note that the "WindowOverlap" value should be recalculated if frame size is changed
// this has not yet been considered in the Config file!
WindowOverlap = 0.1028,
DoMelScale = (scaleType == FreqScaleType.Mel) ? true : false,
MelBinCount = (scaleType == FreqScaleType.Mel) ? finalBinCount : frameSize / 2,
Expand All @@ -73,16 +97,16 @@ public void Execute(Arguments arguments)

// MinFreBin and MaxFreqBin to work with arbitrary frequency bin bounds.
// The default value is minFreqBin = 1 and maxFreqBin = finalBinCount.
// For any other arbitrary frequency bin bounds, these two parameters need to be manually set .
// For any other arbitrary frequency bin bounds, these two parameters need to be manually set.

// Black Rail call is between 1000 Hz and 3000 Hz, which is mapped to Mel value [1000, 1876]
// Hence, we only work with freq bins between [40, 76]
int minFreqBin = 1; //24; //35; //40; //
int maxFreqBin = finalBinCount; //95; //85; //80; //76;
int numFreqBand = 1;
int patchWidth = (maxFreqBin - minFreqBin + 1) / numFreqBand; //finalBinCount / numFreqBand;
int patchHeight = 1; // 2; // 4; // 16; // 6; // Frame size
int numRandomPatches = 4; //40; //8; //16; // 80; // 2; //10; //100; // 30; // 500; //20; //
int minFreqBin = configuration.MinFreqBin; // 24; //1; //35; //40; //
int maxFreqBin = configuration.MaxFreqBin; // 95; //103; //109; //finalBinCount; //85; //80; //76;
int numFreqBand = configuration.NumFreqBand; // 1;
int patchWidth = configuration.PatchWidth; // (maxFreqBin - minFreqBin + 1) / numFreqBand; //finalBinCount / numFreqBand;
int patchHeight = configuration.PatchHeight; // 1; // 2; // 4; // 16; // 6; // Frame size
int numRandomPatches = configuration.NumRandomPatches; // 4; //40; //8; //16; // 80; // 2; //10; //100; // 30; // 500; //20; //
// int fileCount = Directory.GetFiles(folderPath, "*.wav").Length;

// Define variable number of "randomPatch" lists based on "numFreqBand"
Expand Down Expand Up @@ -123,7 +147,12 @@ public void Execute(Arguments arguments)

// DO NOISE REDUCTION
// sonogram.Data = SNR.NoiseReduce_Median(sonogram.Data, nhBackgroundThreshold: 2.0);
sonogram.Data = PcaWhitening.NoiseReduction(sonogram.Data);
//sonogram.Data = PcaWhitening.NoiseReduction(sonogram.Data);

if (configuration.DoNoiseReduction)
{
sonogram.Data = PcaWhitening.NoiseReduction(sonogram.Data);
}

// check whether the full band spectrogram is needed or a matrix with arbitrary freq bins
if (minFreqBin != 1 || maxFreqBin != finalBinCount)
Expand Down Expand Up @@ -154,7 +183,7 @@ public void Execute(Arguments arguments)
}

// convert list of random patches matrices to one matrix
int numberOfClusters = 256; //8; //128; //500; //10; //16; //20; // 500; // 128; // 64; // 32; // 50;
int numClusters = configuration.NumClusters; // 256; //8; //128; //500; //10; //16; //20; // 500; // 128; // 64; // 32; // 50;
List<double[][]> allBandsCentroids = new List<double[][]>();
List<KMeansClusterCollection> allClusteringOutput = new List<KMeansClusterCollection>();

Expand All @@ -163,10 +192,10 @@ public void Execute(Arguments arguments)
double[,] patchMatrix = randomPatches[i];

// Apply PCA Whitening
var whitenedSpectrogram = PcaWhitening.Whitening(patchMatrix);
var whitenedSpectrogram = PcaWhitening.Whitening(patchMatrix, configuration.DoWhitening);

// Do k-means clustering
var clusteringOutput = KmeansClustering.Clustering(whitenedSpectrogram.Reversion, numberOfClusters);
var clusteringOutput = KmeansClustering.Clustering(whitenedSpectrogram.Reversion, numClusters);
// var clusteringOutput = KmeansClustering.Clustering(patchMatrix, noOfClusters, pathToClusterCsvFile);

// writing centroids to a csv file
Expand Down Expand Up @@ -236,7 +265,7 @@ public void Execute(Arguments arguments)
//var recording2Path = Path.Combine(trainSetPath, "SM304264_0+1_20160421_054539_29-30min.wav"); // an example from the train set
//var recording2Path = Path.Combine(testSetPath, "SM304264_0+1_20160423_054539_29-30min.wav"); // an example from the test set
// check whether there is any file in the folder/subfolders
if (Directory.GetFiles(trainSetPath, "*", SearchOption.AllDirectories).Length == 0) // testSetPath
if (Directory.GetFiles(testSetPath, "*", SearchOption.AllDirectories).Length == 0) // trainSetPath
{
throw new ArgumentException("The folder of recordings is empty...");
}
Expand All @@ -250,7 +279,7 @@ public void Execute(Arguments arguments)
Dictionary<string, List<double[,]>> allFilesStdFeatureVectors = new Dictionary<string, List<double[,]>>();
Dictionary<string, List<double[,]>> allFilesSkewnessFeatureVectors = new Dictionary<string, List<double[,]>>();

foreach (string filePath in Directory.GetFiles(trainSetPath, "*.wav")) // testSetPath
foreach (string filePath in Directory.GetFiles(testSetPath, "*.wav")) // trainSetPath
{
FileInfo fileInfo = filePath.ToFileInfo();

Expand All @@ -268,6 +297,10 @@ public void Execute(Arguments arguments)
// DO NOISE REDUCTION
// sonogram.Data = SNR.NoiseReduce_Median(sonogram.Data, nhBackgroundThreshold: 2.0);
sonogram.Data = PcaWhitening.NoiseReduction(sonogram.Data);
if (configuration.DoNoiseReduction)
{
sonogram.Data = PcaWhitening.NoiseReduction(sonogram.Data);
}

// check whether the full band spectrogram is needed or a matrix with arbitrary freq bins
if (minFreqBin != 1 || maxFreqBin != finalBinCount)
Expand Down Expand Up @@ -698,6 +731,82 @@ public void Execute(Arguments arguments)
//*****
}

public class FeatureLearningConfig : Config
{
public const FreqScaleType DefaultFrequencyScaleType = FreqScaleType.Mel;

public const int DefaultHertzInterval = 1000;

public const int DefaultFrameSize = 1024;

public const int DefaultFinalBinCount = 128;

public const int DefaultMinFreqBin = 1;

public const int DefaultMaxFreqBin = DefaultFinalBinCount;

public const int DefaultNumFreqBand = 1;

public const int DefaultPatchWidth = (DefaultMaxFreqBin - DefaultMinFreqBin + 1) / DefaultNumFreqBand;

public const int DefaultPatchHeight = 1;

public const int DefaultNumRandomPatches = 4;

public const int DefaultNumClusters = 256;

public const bool DefaultDoNoiseReduction = true;

public const bool DefaultDoWhitening = true;

/// <summary>
/// Initializes a new instance of the <see cref="FeatureLearningConfig"/> class.
/// CONSTRUCTOR
/// </summary>
public FeatureLearningConfig()
{
this.FrequencyScaleType = DefaultFrequencyScaleType;
this.HertzInterval = DefaultHertzInterval;
this.FrameSize = DefaultFrameSize;
this.FinalBinCount = DefaultFinalBinCount;
this.MinFreqBin = DefaultMinFreqBin;
this.MaxFreqBin = DefaultMaxFreqBin;
this.NumFreqBand = DefaultNumFreqBand;
this.PatchWidth = DefaultPatchWidth;
this.PatchHeight = DefaultPatchHeight;
this.NumRandomPatches = DefaultNumRandomPatches;
this.NumClusters = DefaultNumClusters;
this.DoNoiseReduction = DefaultDoNoiseReduction;
this.DoWhitening = DefaultDoWhitening;
}

public FreqScaleType FrequencyScaleType { get; set; }

public int HertzInterval { get; set; }

public int FrameSize { get; set; }

public int FinalBinCount { get; set; }

public int MinFreqBin { get; set; }

public int MaxFreqBin { get; set; }

public int NumFreqBand { get; set; }

public int PatchWidth { get; set; }

public int PatchHeight { get; set; }

public int NumRandomPatches { get; set; }

public int NumClusters { get; set; }

public bool DoNoiseReduction { get; set; }

public bool DoWhitening { get; set; }
}

[Command(
CommandName,
Description = "Temporary entry point for unsupervised feature learning")]
Expand Down
4 changes: 2 additions & 2 deletions src/AudioAnalysisTools/DSP/PcaWhitening.cs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -28,7 +28,7 @@ public class Output
public int Components { get; set; }
}

public static Output Whitening(double[,] matrix)
public static Output Whitening(double[,] matrix, bool doWhitening)
{
if (matrix == null)
{
Expand All @@ -43,7 +43,7 @@ public static Output Whitening(double[,] matrix)
{
// the "Center" method only subtracts the mean.
Method = PrincipalComponentMethod.Center,
Whiten = true,
Whiten = doWhitening,
};

pca.Learn(jaggedArray);
Expand Down
4 changes: 2 additions & 2 deletions tests/Acoustics.Test/AudioAnalysisTools/DSP/PcaWhiteningTests.cs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -68,7 +68,7 @@ public void PcaWhiteningDefault()
sonogram.Data = dataMatrix;

// DO PCA WHITENING
var whitenedSpectrogram = PcaWhitening.Whitening(sonogram.Data);
var whitenedSpectrogram = PcaWhitening.Whitening(sonogram.Data, true);

// DO UNIT TESTING
// check if the dimensions of the reverted spectrogram (second output of the pca whitening) is equal to the input matrix
Expand Down Expand Up @@ -124,7 +124,7 @@ public void TestPcaWhitening()
double[,] sequentialPatchMatrix = sequentialPatches.ToMatrix();

// DO PCA WHITENING
var whitenedSpectrogram = PcaWhitening.Whitening(sequentialPatchMatrix);
var whitenedSpectrogram = PcaWhitening.Whitening(sequentialPatchMatrix, true);

// reconstructing the spectrogram from sequential patches and the projection matrix obtained from random patches
var projectionMatrix = whitenedSpectrogram.ProjectionMatrix;
Expand Down
2 changes: 1 addition & 1 deletion tests/Acoustics.Test/AudioAnalysisTools/DSP/UnsupervisedFeatureLearningTest.cs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -182,7 +182,7 @@ public void TestFeatureLearning()
double[,] patchMatrix = randomPatches[i];

// Apply PCA Whitening
var whitenedSpectrogram = PcaWhitening.Whitening(patchMatrix);
var whitenedSpectrogram = PcaWhitening.Whitening(patchMatrix, true);

// Do k-means clustering
var clusteringOutput = KmeansClustering.Clustering(whitenedSpectrogram.Reversion, numberOfClusters);
Expand Down

0 comments on commit c70b164

Please sign in to comment.