public List<bool[]> CreateFingerprintsFromLogSpectrum(
double[][] logarithmizedSpectrum, IStride stride, int fingerprintLength, int overlap, int topWavelets)
{
DbgTimer t = new DbgTimer();
t.Start ();
// Cut the logaritmic spectrogram into smaller spectrograms with one stride between each
List<double[][]> spectralImages = SpectrumService.CutLogarithmizedSpectrum(logarithmizedSpectrum, stride, fingerprintLength, overlap);
// Then apply the wavelet transform on them to later reduce the resolution
// do this in place
WaveletService.ApplyWaveletTransformInPlace(spectralImages);
// Then for each of the wavelet reduce the resolution by only keeping the top wavelets
// and ignore the magnitude of the top wavelets.
// Instead, we can simply keep the sign of it (+/-).
// This information is enough to keep the extract perceptual characteristics of a song.
List<bool[]> fingerprints = new List<bool[]>();
foreach (var spectralImage in spectralImages)
{
bool[] image = FingerprintDescriptor.ExtractTopWavelets(spectralImage, topWavelets);
fingerprints.Add(image);
}
Dbg.WriteLine ("Created {1} Fingerprints from Log Spectrum - Execution Time: {0} ms", t.Stop().TotalMilliseconds, fingerprints.Count);
return fingerprints;
}
public static float[] DecodeUsingMplayerAndSox(string fileIn, int srate, int secondsToAnalyze)
{
lock (_locker) {
using (Process tosoxreadable = new Process())
{
fileIn = Regex.Replace(fileIn, "%20", " ");
DbgTimer t = new DbgTimer();
t.Start();
String curdir = System.Environment.CurrentDirectory;
Dbg.WriteLine("Decoding: " + fileIn);
String tempFile = System.IO.Path.GetTempFileName();
String soxreadablewav = tempFile + ".wav";
Dbg.WriteLine("Temporary wav file: " + soxreadablewav);
tosoxreadable.StartInfo.FileName = "./NativeLibraries\\mplayer\\mplayer.exe";
tosoxreadable.StartInfo.Arguments = " -quiet -vc null -vo null -ao pcm:fast:waveheader \"" + fileIn + "\" -ao pcm:file=\\\"" + soxreadablewav + "\\\"";
tosoxreadable.StartInfo.UseShellExecute = false;
tosoxreadable.StartInfo.RedirectStandardOutput = true;
tosoxreadable.StartInfo.RedirectStandardError = true;
tosoxreadable.Start();
tosoxreadable.WaitForExit();
int exitCode = tosoxreadable.ExitCode;
// 0 = succesfull
// 1 = partially succesful
// 2 = failed
if (exitCode != 0)
{
string standardError = tosoxreadable.StandardError.ReadToEnd();
Console.Out.WriteLine(standardError);
return(null);
}
#if DEBUG
string standardOutput = tosoxreadable.StandardOutput.ReadToEnd();
Console.Out.WriteLine(standardOutput);
#endif
float[] floatBuffer = null;
if (File.Exists(soxreadablewav))
{
floatBuffer = DecodeUsingSox(soxreadablewav, srate, secondsToAnalyze);
try
{
File.Delete(tempFile);
File.Delete(soxreadablewav);
}
catch (IOException io)
{
Console.WriteLine(io);
}
}
Dbg.WriteLine("Decoding Execution Time: " + t.Stop().TotalMilliseconds + " ms");
return(floatBuffer);
}
}
}
public static float[] DecodeUsingMplayer(string fileIn, int srate)
{
lock (_locker) {
using (Process towav = new Process())
{
fileIn = Regex.Replace(fileIn, "%20", " ");
DbgTimer t = new DbgTimer();
t.Start();
String curdir = System.Environment.CurrentDirectory;
Dbg.WriteLine("Decoding: " + fileIn);
String tempFile = System.IO.Path.GetTempFileName();
String wav = tempFile + ".wav";
Dbg.WriteLine("Temporary wav file: " + wav);
towav.StartInfo.FileName = "./NativeLibraries\\mplayer\\mplayer.exe";
towav.StartInfo.Arguments = " -quiet -ao pcm:fast:waveheader \"" + fileIn + "\" -format floatle -af resample=" + srate + ":0:2,pan=1:0.5:0.5 -channels 1 -vo null -vc null -ao pcm:file=\\\"" + wav + "\\\"";
towav.StartInfo.UseShellExecute = false;
towav.StartInfo.RedirectStandardOutput = true;
towav.StartInfo.RedirectStandardError = true;
towav.Start();
towav.WaitForExit();
int exitCode = towav.ExitCode;
// 0 = succesfull
// 1 = partially succesful
// 2 = failed
if (exitCode != 0)
{
string standardError = towav.StandardError.ReadToEnd();
Console.Out.WriteLine(standardError);
return(null);
}
#if DEBUG
string standardOutput = towav.StandardOutput.ReadToEnd();
Console.Out.WriteLine(standardOutput);
#endif
RiffRead riff = new RiffRead(wav);
riff.Process();
float[] floatBuffer = riff.SoundData[0];
try
{
File.Delete(tempFile);
//File.Delete(wav);
}
catch (IOException io)
{
Console.WriteLine(io);
}
Dbg.WriteLine("Decoding Execution Time: " + t.Stop().TotalMilliseconds + " ms");
return(floatBuffer);
}
}
}
/// <summary>
/// Computes a Scms model from the MFCC representation of a song.
/// </summary>
/// <param name="mfcc">Comirva.Audio.Util.Maths.Matrix mfcc</param>
/// <returns></returns>
public static Scms GetScms(Comirva.Audio.Util.Maths.Matrix mfccs, string name)
{
DbgTimer t = new DbgTimer();
t.Start();
Comirva.Audio.Util.Maths.Matrix mean = mfccs.Mean(2);
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE) {
if (Analyzer.DEBUG_OUTPUT_TEXT) mean.WriteText(name + "_mean.txt");
mean.DrawMatrixGraph(name + "_mean.png");
}
#endif
// Covariance
Comirva.Audio.Util.Maths.Matrix covarMatrix = mfccs.Cov(mean);
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE) {
if (Analyzer.DEBUG_OUTPUT_TEXT) covarMatrix.WriteText(name + "_covariance.txt");
covarMatrix.DrawMatrixGraph(name + "_covariance.png");
}
#endif
// Inverse Covariance
Comirva.Audio.Util.Maths.Matrix covarMatrixInv;
try {
covarMatrixInv = covarMatrix.InverseGausJordan();
} catch (Exception) {
Dbg.WriteLine("MatrixSingularException - Scms failed!");
return null;
}
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE) {
if (Analyzer.DEBUG_OUTPUT_TEXT) covarMatrixInv.WriteAscii(name + "_inverse_covariance.ascii");
covarMatrixInv.DrawMatrixGraph(name + "_inverse_covariance.png");
}
#endif
// Store the Mean, Covariance, Inverse Covariance in an optimal format.
int dim = mean.Rows;
Scms s = new Scms(dim);
int l = 0;
for (int i = 0; i < dim; i++) {
s.mean[i] = (float) mean.MatrixData[i][0];
for (int j = i; j < dim; j++) {
s.cov[l] = (float) covarMatrix.MatrixData[i][j];
s.icov[l] = (float) covarMatrixInv.MatrixData[i][j];
l++;
}
}
Dbg.WriteLine("Compute Scms - Execution Time: {0} ms", t.Stop().TotalMilliseconds);
return s;
}
public Matrix Apply(ref Matrix m)
{
DbgTimer t = new DbgTimer();
t.Start();
Matrix mel = new Matrix(filterWeights.rows, m.columns);
/*
* // Performance optimization of ...
* mel = filterWeights.Multiply(m);
* for (int i = 0; i < mel.rows; i++) {
* for (int j = 0; j < mel.columns; j++) {
* mel.d[i, j] = (mel.d[i, j] < 1.0f ? 0 : (float)(10.0 * Math.Log10(mel.d[i, j])));
* //mel.d[i, j] = (float)(10.0 * Math.Log10(mel.d[i, j]));
* }
* }
*/
int mc = m.columns;
int mr = m.rows;
int melcolumns = mel.columns;
int fwc = filterWeights.columns;
int fwr = filterWeights.rows;
unsafe
{
fixed(float *md = m.d, fwd = filterWeights.d, meld = mel.d)
{
for (int i = 0; i < mc; i++)
{
for (int k = 0; k < fwr; k++)
{
int idx = k * melcolumns + i;
int kfwc = k * fwc;
for (int j = 0; j < mr; j++)
{
meld[idx] += fwd[kfwc + j] * md[j * mc + i];
}
meld[idx] = (meld[idx] < 1.0f ?
0 : (float)(10.0 * Math.Log10(meld[idx])));
}
}
}
}
Matrix mfcc = dct.Multiply(mel);
Dbg.WriteLine("mfcc (MfccLessOptimized) Execution Time: " + t.Stop().TotalMilliseconds + " ms");
return(mfcc);
}
public static float[] Decode(string fileIn, int srate, int secondsToAnalyze)
{
DbgTimer t = new DbgTimer();
t.Start();
float[] floatBuffer = null;
// check if file exists
if (fileIn != null && fileIn != "") {
FileInfo fi = new FileInfo(fileIn);
if (!fi.Exists) {
Console.Out.WriteLine("No file found {0}!", fileIn);
return null;
}
}
// Try to use Un4Seen Bass
BassProxy bass = BassProxy.Instance;
double duration = bass.GetDurationInSeconds(fileIn);
if (duration > 0) {
Dbg.WriteLine("Using BASS to decode the file ...");
// duration in seconds
if (duration > secondsToAnalyze) {
// find segment to extract
double startSeconds = (duration/2-(secondsToAnalyze/2));
if (startSeconds < 0) {
startSeconds = 0;
}
floatBuffer = bass.ReadMonoFromFile(fileIn, srate, secondsToAnalyze*1000, (int) (startSeconds*1000));
// if this failes, the duration read from the tags was wrong or it is something wrong with the audio file
if (floatBuffer == null) {
IOUtils.LogMessageToFile(Mir.WARNING_FILES_LOG, fileIn);
}
} else {
// return whole file
floatBuffer = bass.ReadMonoFromFile(fileIn, srate, 0, 0);
// if this failes, the duration read from the tags was wrong or it is something wrong with the audio file
if (floatBuffer == null) {
IOUtils.LogMessageToFile(Mir.WARNING_FILES_LOG, fileIn);
}
}
}
// Bass failed reading or never even tried, so use another alternative
if (floatBuffer == null) {
Dbg.WriteLine("Using MPlayer and SOX to decode the file ...");
fileIn = Regex.Replace(fileIn, "%20", " ");
floatBuffer = DecodeUsingMplayerAndSox(fileIn, srate, secondsToAnalyze);
}
return floatBuffer;
}
/// <summary>
/// Computes a Scms model from the MFCC representation of a song.
/// </summary>
/// <param name="mfcc">Comirva.Audio.Util.Maths.Matrix mfcc</param>
/// <returns></returns>
public static Scms GetScmsNoInverse(Comirva.Audio.Util.Maths.Matrix mfccs, string name)
{
DbgTimer t = new DbgTimer();
t.Start();
Comirva.Audio.Util.Maths.Matrix mean = mfccs.Mean(2);
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE)
{
if (Analyzer.DEBUG_OUTPUT_TEXT)
{
mean.WriteText(name + "_mean.txt");
}
mean.DrawMatrixGraph(name + "_mean.png");
}
#endif
// Covariance
Comirva.Audio.Util.Maths.Matrix covarMatrix = mfccs.Cov(mean);
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE)
{
if (Analyzer.DEBUG_OUTPUT_TEXT)
{
covarMatrix.WriteText(name + "_covariance.txt");
}
covarMatrix.DrawMatrixGraph(name + "_covariance.png");
}
#endif
Comirva.Audio.Util.Maths.Matrix covarMatrixInv = new Comirva.Audio.Util.Maths.Matrix(covarMatrix.Rows, covarMatrix.Columns);
// Store the Mean, Covariance, Inverse Covariance in an optimal format.
int dim = mean.Rows;
Scms s = new Scms(dim);
int l = 0;
for (int i = 0; i < dim; i++)
{
s.mean[i] = (float)mean.MatrixData[i][0];
for (int j = i; j < dim; j++)
{
s.cov[l] = (float)covarMatrix.MatrixData[i][j];
s.icov[l] = (float)covarMatrixInv.MatrixData[i][j];
l++;
}
}
Dbg.WriteLine("GetScmsNoInverse - Execution Time: {0} ms", t.Stop().TotalMilliseconds);
return(s);
}
public Matrix Apply(ref Matrix m)
{
DbgTimer t = new DbgTimer();
t.Start();
Matrix mel = new Matrix(filterWeights.rows, m.columns);
int mc = m.columns;
int melcolumns = mel.columns;
int fwc = filterWeights.columns;
int fwr = filterWeights.rows;
unsafe
{
fixed(float *md = m.d, fwd = filterWeights.d, meld = mel.d)
{
for (int i = 0; i < mc; i++)
{
for (int k = 0; k < fwr; k++)
{
int idx = k * melcolumns + i;
int kfwc = k * fwc;
// The filter weights matrix is mostly 0.
// So only multiply non-zero elements!
for (int j = fwFT[k, 0]; j < fwFT[k, 1]; j++)
{
meld[idx] += fwd[kfwc + j] * md[j * mc + i];
}
meld[idx] = (meld[idx] < 1.0f ?
0 : (float)(10.0 * Math.Log10(meld[idx])));
}
}
}
}
try {
Matrix mfcc = dct.Multiply(mel);
long stop = 0;
t.Stop(ref stop);
Dbg.WriteLine("Mirage - mfcc Execution Time: {0}ms", stop);
return(mfcc);
} catch (MatrixDimensionMismatchException) {
throw new MfccFailedException();
}
}
/// <summary>
/// Find Similar Tracks to an audio file using its file path
/// </summary>
/// <param name="searchForPath">audio file path</param>
/// <param name="db">database</param>
/// <param name="analysisMethod">analysis method (SCMS or MandelEllis)</param>
/// <param name="numToTake">max number of entries to return</param>
/// <param name="percentage">percentage below and above the duration in ms when querying (used if between 0.1 - 0.9)</param>
/// <param name="distanceType">distance method to use (KullbackLeiblerDivergence is default)</param>
/// <returns>a dictinary list of key value pairs (filepath and distance)</returns>
public static Dictionary<KeyValuePair<int, string>, double> SimilarTracks(string searchForPath, Db db, Analyzer.AnalysisMethod analysisMethod, int numToTake=25, double percentage=0.2, AudioFeature.DistanceType distanceType = AudioFeature.DistanceType.KullbackLeiblerDivergence)
{
DbgTimer t = new DbgTimer();
t.Start();
FileInfo fi = new FileInfo(searchForPath);
AudioFeature seedAudioFeature = null;
AudioFeature[] audioFeatures = null;
switch (analysisMethod) {
case Analyzer.AnalysisMethod.MandelEllis:
seedAudioFeature = Analyzer.AnalyzeMandelEllis(fi);
audioFeatures = new MandelEllis[100];
break;
case Analyzer.AnalysisMethod.SCMS:
seedAudioFeature = Analyzer.AnalyzeScms(fi);
audioFeatures = new Scms[100];
break;
}
// Get all tracks from the DB except the seedSongs
IDataReader r = db.GetTracks(null, seedAudioFeature.Duration, percentage);
// store results in a dictionary
var NameDictionary = new Dictionary<KeyValuePair<int, string>, double>();
int[] mapping = new int[100];
int read = 1;
double dcur;
while (read > 0) {
read = db.GetNextTracks(ref r, ref audioFeatures, ref mapping, 100, analysisMethod);
for (int i = 0; i < read; i++) {
dcur = seedAudioFeature.GetDistance(audioFeatures[i], distanceType);
// convert to positive values
dcur = Math.Abs(dcur);
NameDictionary.Add(new KeyValuePair<int,string>(mapping[i], audioFeatures[i].Name), dcur);
}
}
// sort by non unique values
var sortedDict = (from entry in NameDictionary orderby entry.Value ascending select entry)
.Take(numToTake)
.ToDictionary(pair => pair.Key, pair => pair.Value);
Console.Out.WriteLine(String.Format("Found Similar to ({0}) in {1} ms", seedAudioFeature.Name, t.Stop().TotalMilliseconds));
return sortedDict;
}
public static Scms Analyze (string file_path)
{
DbgTimer t = new DbgTimer ();
t.Start ();
Matrix stftdata = ad.Decode (file_path);
Matrix mfccdata = mfcc.Apply (ref stftdata);
Scms scms = Scms.GetScms (mfccdata);
long stop = 0;
t.Stop (ref stop);
Dbg.WriteLine ("Mirage - Total Execution Time: {0}ms", stop);
return scms;
}
// EDIT !
public static Matrix AnalyzeMFCC (string file_path)
{
Hyena.Log.Debug("AnalyzeMFCC called for " + file_path);
DbgTimer t = new DbgTimer ();
t.Start ();
Matrix stftdata = ad.Decode (file_path);
Matrix mfccdata = mfcc.Apply (ref stftdata);
long stop = 0;
t.Stop (ref stop);
Dbg.WriteLine ("Mirage - Total Execution Time: {0}ms", stop);
return mfccdata;
}
public static Scms Analyze(string file_path)
{
DbgTimer t = new DbgTimer();
t.Start();
Matrix stftdata = ad.Decode(file_path);
Matrix mfccdata = mfcc.Apply(ref stftdata);
Scms scms = Scms.GetScms(mfccdata);
long stop = 0;
t.Stop(ref stop);
Dbg.WriteLine("Mirage - Total Execution Time: {0}ms", stop);
return(scms);
}
// Computes a Scms model from the MFCC representation of a song.
public static Scms GetScms(Matrix mfcc)
{
DbgTimer t = new DbgTimer();
t.Start();
// Mean
Vector m = mfcc.Mean();
// Covariance
Matrix c = mfcc.Covariance(m);
// Inverse Covariance
Matrix ic;
try {
ic = c.Inverse();
} catch (MatrixSingularException) {
throw new ScmsImpossibleException();
}
// Store the Mean, Covariance, Inverse Covariance in an optimal format.
int dim = m.rows;
Scms s = new Scms(dim);
int l = 0;
for (int i = 0; i < dim; i++)
{
s.mean[i] = m.d[i, 0];
for (int j = i; j < dim; j++)
{
s.cov[l] = c.d[i, j];
s.icov[l] = ic.d[i, j];
l++;
}
}
long stop = 0;
t.Stop(ref stop);
Dbg.WriteLine("Mirage - scms created in: {0}ms", stop);
return(s);
}
public Matrix Apply(ref Matrix m)
{
DbgTimer t = new DbgTimer ();
t.Start ();
Matrix mel = new Matrix (filterWeights.rows, m.columns);
int mc = m.columns;
int melcolumns = mel.columns;
int fwc = filterWeights.columns;
int fwr = filterWeights.rows;
unsafe {
fixed (float* md = m.d, fwd = filterWeights.d, meld = mel.d) {
for (int i = 0; i < mc; i++) {
for (int k = 0; k < fwr; k++) {
int idx = k*melcolumns + i;
int kfwc = k*fwc;
// The filter weights matrix is mostly 0.
// So only multiply non-zero elements!
for (int j = fwFT[k,0]; j < fwFT[k,1]; j++) {
meld[idx] += fwd[kfwc + j] * md[j*mc + i];
}
meld[idx] = (meld[idx] < 1.0f ?
0 : (float)(10.0 * Math.Log10(meld[idx])));
}
}
}
}
try {
Matrix mfcc = dct.Multiply (mel);
long stop = 0;
t.Stop (ref stop);
Dbg.WriteLine ("Mirage - mfcc Execution Time: {0}ms", stop);
return mfcc;
} catch (MatrixDimensionMismatchException) {
throw new MfccFailedException ();
}
}
/// <summary>
/// Apply the STFT on the audiodata
/// </summary>
/// <param name="audiodata">Audiodata to apply the STFT on</param>
/// <returns>A matrix with the result of the STFT</returns>
public Matrix Apply(float[] audiodata)
{
DbgTimer t = new DbgTimer();
t.Start();
// calculate how many hops (bands) we have using the current overlap (hopsize)
int hops = (audiodata.Length - winsize)/ hopsize;
// Create a Matrix with "winsize" Rows and "hops" Columns
// Matrix[Row, Column]
Matrix stft = new Matrix(winsize/2 +1, hops);
for (int i = 0; i < hops; i++) {
fft.ComputeMirageMatrixUsingFftw(ref stft, i, audiodata, i*hopsize);
}
Dbg.WriteLine("Stft (ComputeMirageMatrix) Execution Time: " + t.Stop().TotalMilliseconds + " ms");
return stft;
}
/// <summary>
/// Apply the STFT on the audiodata
/// </summary>
/// <param name="audiodata">Audiodata to apply the STFT on</param>
/// <returns>A matrix with the result of the STFT</returns>
public Matrix Apply(float[] audiodata)
{
DbgTimer t = new DbgTimer();
t.Start();
// calculate how many hops (bands) we have using the current overlap (hopsize)
int hops = (audiodata.Length - winsize) / hopsize;
// Create a Matrix with "winsize" Rows and "hops" Columns
// Matrix[Row, Column]
Matrix stft = new Matrix(winsize / 2 + 1, hops);
for (int i = 0; i < hops; i++)
{
fft.ComputeMirageMatrixUsingFftw(ref stft, i, audiodata, i * hopsize);
}
Dbg.WriteLine("Stft (ComputeMirageMatrix) Execution Time: " + t.Stop().TotalMilliseconds + " ms");
return(stft);
}
private static bool AnalyseAndAddScmsUsingFingerprints(List<double[][]> spectralImages,
List<bool[]> fingerprints,
WorkUnitParameterObject param,
Db db,
int trackId,
bool doOutputDebugInfo=DEFAULT_DEBUG_INFO)
{
DbgTimer t = new DbgTimer();
t.Start ();
// Insert Statistical Cluster Model Similarity Audio Feature
string fileName = param.FileName;
// Merge the arrays in the List using Linq
var result = spectralImages.SelectMany(i => i).ToArray();
Comirva.Audio.Util.Maths.Matrix scmsMatrix = new Comirva.Audio.Util.Maths.Matrix(result);
if (doOutputDebugInfo) {
scmsMatrix.DrawMatrixImage(String.Format("{0}_spectral.png", fileName));
}
#region Store in a Statistical Cluster Model Similarity class.
Scms audioFeature = Scms.GetScms(scmsMatrix, fileName);
if (audioFeature != null) {
// Store bitstring hash as well
audioFeature.BitString = GetBitString(scmsMatrix);
// Store duration
audioFeature.Duration = (long) param.DurationInMs;
// Store file name
audioFeature.Name = param.PathToAudioFile;
// Add to database
int id = trackId;
if (db.AddTrack(audioFeature) == -1) {
Console.Out.WriteLine("Failed! Could not add audio feature to database ({0})!", fileName);
return false;
}
} else {
return false;
}
#endregion
Dbg.WriteLine ("AnalyseAndAddScmsUsingFingerprints2 - Execution Time: {0} ms", t.Stop().TotalMilliseconds);
return true;
}
private static bool AnalyseAndAddScmsUsingFingerprints(List<bool[]> fingerprints,
WorkUnitParameterObject param,
Db db,
int trackId,
bool doOutputDebugInfo=DEFAULT_DEBUG_INFO)
{
DbgTimer t = new DbgTimer();
t.Start ();
// Insert Statistical Cluster Model Similarity Audio Feature
string fileName = param.FileName;
int fingerprintWidth = param.FingerprintingConfiguration.FingerprintLength;
int fingerprintHeight = param.FingerprintingConfiguration.LogBins;
int fingerprintCount = 0;
foreach (bool[] fingerprint in fingerprints) {
fingerprintCount++;
Comirva.Audio.Util.Maths.Matrix scmsMatrix = new Comirva.Audio.Util.Maths.Matrix(fingerprintWidth, fingerprintHeight);
for (int i = 0; i < fingerprintWidth /*128*/; i++) {
for (int j = 0; j < fingerprintHeight /*32*/; j++) {
// Negative Numbers = 01
// Positive Numbers = 10
// Zeros = 00
bool v1 = fingerprint[(2 * fingerprintHeight * i) + (2 * j)];
bool v2 = fingerprint[(2 * fingerprintHeight * i) + (2 * j) + 1];
if (v1) {
scmsMatrix.MatrixData[i][j] = 2.0;
} else if (v2) {
scmsMatrix.MatrixData[i][j] = 0.0;
} else {
scmsMatrix.MatrixData[i][j] = 1.0;
}
}
}
if (doOutputDebugInfo) {
scmsMatrix.DrawMatrixImage(String.Format("{0}_fingerprint_{1}.png", fileName, fingerprintCount), fingerprintWidth, fingerprintHeight);
}
#region Store in a Statistical Cluster Model Similarity class.
Scms audioFeature = Scms.GetScmsNoInverse(scmsMatrix, fileName);
if (audioFeature != null) {
// Store bitstring hash as well
audioFeature.BitString = GetBitString(fingerprint);
// Store duration
audioFeature.Duration = (long) param.DurationInMs;
// Store file name
audioFeature.Name = param.PathToAudioFile;
// Add to database
int id = trackId;
if (db.AddTrack(audioFeature) == -1) {
Console.Out.WriteLine("Failed! Could not add audio feature to database ({0})!", fileName);
return false;
}
} else {
return false;
}
#endregion
}
Dbg.WriteLine ("AnalyseAndAddScmsUsingFingerprints - Execution Time: {0} ms", t.Stop().TotalMilliseconds);
return true;
}
public static Scms AnalyzeScms(FileInfo filePath, bool doOutputDebugInfo=DEFAULT_DEBUG_INFO, bool useHaarWavelet = true)
{
DbgTimer t = new DbgTimer();
t.Start ();
FindSimilar.AudioProxies.BassProxy bass = FindSimilar.AudioProxies.BassProxy.Instance;
float[] audiodata = AudioFileReader.Decode(filePath.FullName, SAMPLING_RATE, SECONDS_TO_ANALYZE);
if (audiodata == null || audiodata.Length == 0) {
Dbg.WriteLine("Error! - No Audio Found");
return null;
}
// Name of file being processed
string name = StringUtils.RemoveNonAsciiCharacters(Path.GetFileNameWithoutExtension(filePath.Name));
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE) {
if (DEBUG_OUTPUT_TEXT) WriteAscii(audiodata, name + "_audiodata.ascii");
if (DEBUG_OUTPUT_TEXT) WriteF3Formatted(audiodata, name + "_audiodata.txt");
}
#endif
if (doOutputDebugInfo) {
DrawGraph(MathUtils.FloatToDouble(audiodata), name + "_audiodata.png");
}
// Calculate duration in ms
double duration = (double) audiodata.Length / SAMPLING_RATE * 1000;
// Explode samples to the range of 16 bit shorts (–32,768 to 32,767)
// Matlab multiplies with 2^15 (32768)
// e.g. if( max(abs(speech))<=1 ), speech = speech * 2^15; end;
MathUtils.Multiply(ref audiodata, AUDIO_MULTIPLIER); // 65536
// zero pad if the audio file is too short to perform a mfcc
if (audiodata.Length < WINDOW_SIZE * 8)
{
int lenNew = WINDOW_SIZE * 8;
Array.Resize<float>(ref audiodata, lenNew);
}
// 2. Windowing
// 3. FFT
Comirva.Audio.Util.Maths.Matrix stftdata = stftMirage.Apply(audiodata);
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE) {
if (DEBUG_OUTPUT_TEXT) {
stftdata.WriteAscii(name + "_stftdata.ascii");
stftdata.WriteCSV(name + "_stftdata.csv", ";");
}
}
#endif
if (doOutputDebugInfo) {
// same as specgram(audio*32768, 2048, 44100, hanning(2048), 1024);
stftdata.DrawMatrixImageLogValues(name + "_specgram.png", true);
// spec gram with log values for the y axis (frequency)
stftdata.DrawMatrixImageLogY(name + "_specgramlog.png", SAMPLING_RATE, 20, SAMPLING_RATE/2, 120, WINDOW_SIZE);
}
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE & false) {
#region Inverse STFT
double[] audiodata_inverse_stft = stftMirage.InverseStft(stftdata);
// divide
//MathUtils.Divide(ref audiodata_inverse_stft, AUDIO_MULTIPLIER);
MathUtils.Normalize(ref audiodata_inverse_stft);
if (DEBUG_OUTPUT_TEXT) {
WriteAscii(audiodata_inverse_stft, name + "_audiodata_inverse_stft.ascii");
WriteF3Formatted(audiodata_inverse_stft, name + "_audiodata_inverse_stft.txt");
}
DrawGraph(audiodata_inverse_stft, name + "_audiodata_inverse_stft.png");
float[] audiodata_inverse_float = MathUtils.DoubleToFloat(audiodata_inverse_stft);
bass.SaveFile(audiodata_inverse_float, name + "_inverse_stft.wav", Analyzer.SAMPLING_RATE);
#endregion
}
#endif
// 4. Mel Scale Filterbank
// Mel-frequency is proportional to the logarithm of the linear frequency,
// reflecting similar effects in the human's subjective aural perception)
// 5. Take Logarithm
// 6. DCT (Discrete Cosine Transform)
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE) {
#region Mel Scale and Log Values
Comirva.Audio.Util.Maths.Matrix mellog = mfccMirage.ApplyMelScaleAndLog(ref stftdata);
if (DEBUG_OUTPUT_TEXT) {
mellog.WriteCSV(name + "_mel_log.csv", ";");
}
if (doOutputDebugInfo) {
mellog.DrawMatrixImage(name + "_mel_log.png", 600, 400, true, true);
}
#endregion
#region Inverse Mel Scale and Log Values
if (false) {
Comirva.Audio.Util.Maths.Matrix inverse_mellog = mfccMirage.InverseMelScaleAndLog(ref mellog);
inverse_mellog.WriteCSV(name + "_mel_log_inverse.csv", ";");
inverse_mellog.DrawMatrixImageLogValues(name + "_mel_log_inverse.png", true);
double[] audiodata_inverse_mellog = stftMirage.InverseStft(inverse_mellog);
//MathUtils.Divide(ref audiodata_inverse_mellog, AUDIO_MULTIPLIER/100);
MathUtils.Normalize(ref audiodata_inverse_mellog);
if (DEBUG_OUTPUT_TEXT) {
WriteAscii(audiodata_inverse_mellog, name + "_audiodata_inverse_mellog.ascii");
WriteF3Formatted(audiodata_inverse_mellog, name + "_audiodata_inverse_mellog.txt");
}
DrawGraph(audiodata_inverse_mellog, name + "_audiodata_inverse_mellog.png");
float[] audiodata_inverse_mellog_float = MathUtils.DoubleToFloat(audiodata_inverse_mellog);
bass.SaveFile(audiodata_inverse_mellog_float, name + "_inverse_mellog.wav", Analyzer.SAMPLING_RATE);
}
#endregion
}
#endif
Comirva.Audio.Util.Maths.Matrix featureData = null;
if (useHaarWavelet) {
#region Wavelet Transform
int lastHeight = 0;
int lastWidth = 0;
featureData = mfccMirage.ApplyMelScaleWaveletCompression(ref stftdata, out lastHeight, out lastWidth);
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE) {
if (DEBUG_OUTPUT_TEXT) featureData.WriteAscii(name + "_waveletdata.ascii");
}
#endif
if (doOutputDebugInfo) {
featureData.DrawMatrixImageLogValues(name + "_waveletdata.png", true);
}
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE & false) {
#region Inverse Wavelet
// try to do an inverse wavelet transform
Comirva.Audio.Util.Maths.Matrix stftdata_inverse_wavelet = mfccMirage.InverseMelScaleWaveletCompression(ref featureData, lastHeight, lastWidth);
if (DEBUG_OUTPUT_TEXT) stftdata_inverse_wavelet.WriteCSV(name + "_specgramlog_inverse_wavelet.csv", ";");
stftdata_inverse_wavelet.DrawMatrixImageLogValues(name + "_specgramlog_inverse_wavelet.png", true);
double[] audiodata_inverse_wavelet = stftMirage.InverseStft(stftdata_inverse_wavelet);
MathUtils.Normalize(ref audiodata_inverse_wavelet);
if (DEBUG_OUTPUT_TEXT) WriteF3Formatted(audiodata_inverse_wavelet, name + "_audiodata_inverse_wavelet.txt");
DrawGraph(audiodata_inverse_wavelet, name + "_audiodata_inverse_wavelet.png");
bass.SaveFile(MathUtils.DoubleToFloat(audiodata_inverse_wavelet), name + "_inverse_wavelet.wav", Analyzer.SAMPLING_RATE);
#endregion
}
#endif
#endregion
} else {
#region DCT Transform
// It seems the Mirage way of applying the DCT is slightly faster than the
// Comirva way due to less loops
featureData = mfccMirage.ApplyMelScaleDCT(ref stftdata);
//featureData = mfccMirage.ApplyComirvaWay(ref stftdata);
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE) {
if (DEBUG_OUTPUT_TEXT) featureData.WriteAscii(name + "_mfccdata.ascii");
}
#endif
if (doOutputDebugInfo) {
featureData.DrawMatrixImageLogValues(name + "_mfccdata.png", true);
}
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE & false) {
#region Inverse MFCC
// try to do an inverse mfcc
Comirva.Audio.Util.Maths.Matrix stftdata_inverse_mfcc = mfccMirage.InverseMelScaleDCT(ref featureData);
if (DEBUG_OUTPUT_TEXT) stftdata_inverse_mfcc.WriteCSV(name + "_stftdata_inverse_mfcc.csv", ";");
stftdata_inverse_mfcc.DrawMatrixImageLogValues(name + "_specgramlog_inverse_mfcc.png", true);
double[] audiodata_inverse_mfcc = stftMirage.InverseStft(stftdata_inverse_mfcc);
MathUtils.Normalize(ref audiodata_inverse_mfcc);
if (DEBUG_OUTPUT_TEXT) WriteF3Formatted(audiodata_inverse_mfcc, name + "_audiodata_inverse_mfcc.txt");
DrawGraph(audiodata_inverse_mfcc, name + "_audiodata_inverse_mfcc.png");
bass.SaveFile(MathUtils.DoubleToFloat(audiodata_inverse_mfcc), name + "_inverse_mfcc.wav", Analyzer.SAMPLING_RATE);
#endregion
}
#endif
#endregion
}
// Store in a Statistical Cluster Model Similarity class.
// A Gaussian representation of a song
Scms audioFeature = Scms.GetScms(featureData, name);
if (audioFeature != null) {
// Store image if debugging
if (doOutputDebugInfo) {
audioFeature.Image = featureData.DrawMatrixImageLogValues(name + "_featuredata.png", true, false, 0, 0, true);
}
// Store bitstring hash as well
string hashString = GetBitString(featureData);
audioFeature.BitString = hashString;
// Store duration
audioFeature.Duration = (long) duration;
// Store file name
audioFeature.Name = filePath.FullName;
}
Dbg.WriteLine ("Mirage - Total Execution Time: {0} ms", t.Stop().TotalMilliseconds);
return audioFeature;
}
/// <summary>
/// Query the database for perceptually similar tracks using the sound fingerprinting methods
/// </summary>
/// <param name="filePath">input file</param>
/// <returns>a dictionary of similar tracks</returns>
public static Dictionary<Track, double> SimilarTracksSoundfingerprinting(FileInfo filePath, Repository repository)
{
DbgTimer t = new DbgTimer();
t.Start ();
// get work config from the audio file
WorkUnitParameterObject param = GetWorkUnitParameterObjectFromAudioFile(filePath);
param.FingerprintingConfiguration = fingerprintingConfigQuerying;
Dictionary<Track, double> candidates = repository.FindSimilarFromAudioSamples(param.FingerprintingConfiguration.NumberOfHashTables, param.FingerprintingConfiguration.NumberOfKeys, 1, param);
Dbg.WriteLine ("SimilarTracksSoundfingerprinting - Total Execution Time: {0} ms", t.Stop().TotalMilliseconds);
return candidates;
}
public static List<FindSimilar.QueryResult> SimilarTracksSoundfingerprintingList(FileInfo filePath, Repository repository)
{
DbgTimer t = new DbgTimer();
t.Start ();
// get work config from the audio file
WorkUnitParameterObject param = GetWorkUnitParameterObjectFromAudioFile(filePath);
param.FingerprintingConfiguration = fingerprintingConfigQuerying;
// TODO: i don't really know how the threshold tables work.
// 1 returns more similar hits
// 2 returns sometimes only the one we search for
// even 0 seem to work (like 1)
List<FindSimilar.QueryResult> candidates = repository.FindSimilarFromAudioSamplesList(param.FingerprintingConfiguration.NumberOfHashTables, param.FingerprintingConfiguration.NumberOfKeys, 0, param);
Dbg.WriteLine ("SimilarTracksSoundfingerprintingList - Total Execution Time: {0} ms", t.Stop().TotalMilliseconds);
return candidates;
}
public static AudioFeature AnalyzeMandelEllis(FileInfo filePath, bool doOutputDebugInfo=DEFAULT_DEBUG_INFO)
{
DbgTimer t = new DbgTimer();
t.Start ();
float[] audiodata = AudioFileReader.Decode(filePath.FullName, SAMPLING_RATE, SECONDS_TO_ANALYZE);
if (audiodata == null || audiodata.Length == 0) {
Dbg.WriteLine("Error! - No Audio Found");
return null;
}
#if DEBUG
DrawGraph(MathUtils.FloatToDouble(audiodata), "waveform.png");
#endif
// Calculate duration in ms
double duration = (double) audiodata.Length / SAMPLING_RATE * 1000;
// Normalize
//MathUtils.NormalizeInPlace(audiodata);
// Matlab multiplies with 2^15 (32768)
// Explode samples to the range of 16 bit shorts (–32,768 to 32,767)
// if( max(abs(speech))<=1 ), speech = speech * 2^15; end;
MathUtils.Multiply(ref audiodata, AUDIO_MULTIPLIER); // 65536
MandelEllisExtractor extractor = new MandelEllisExtractor(SAMPLING_RATE, WINDOW_SIZE, MFCC_COEFFICIENTS, MEL_COEFFICIENTS);
AudioFeature audioFeature = extractor.Calculate(MathUtils.FloatToDouble(audiodata));
if (audioFeature != null) {
// Store duration
audioFeature.Duration = (long) duration;
// Store file name
audioFeature.Name = filePath.Name;
}
Dbg.WriteLine ("MandelEllisExtractor - Total Execution Time: {0} ms", t.Stop().TotalMilliseconds);
return audioFeature;
}
/// <summary>
/// Return information from the Audio File
/// </summary>
/// <param name="filePath">filepath object</param>
/// <returns>a WorkUnitParameter object</returns>
public static WorkUnitParameterObject GetWorkUnitParameterObjectFromAudioFile(FileInfo filePath, bool doOutputDebugInfo=DEFAULT_DEBUG_INFO)
{
DbgTimer t = new DbgTimer();
t.Start ();
float[] audiodata = AudioFileReader.Decode(filePath.FullName, SAMPLING_RATE, SECONDS_TO_ANALYZE);
if (audiodata == null || audiodata.Length == 0) {
Dbg.WriteLine("Error! - No Audio Found");
return null;
}
// Name of file being processed
string fileName = StringUtils.RemoveNonAsciiCharacters(Path.GetFileNameWithoutExtension(filePath.Name));
#if DEBUG
if (DEBUG_INFO_VERBOSE) {
if (DEBUG_OUTPUT_TEXT) WriteAscii(audiodata, fileName + "_audiodata.ascii");
if (DEBUG_OUTPUT_TEXT) WriteF3Formatted(audiodata, fileName + "_audiodata.txt");
}
#endif
if (doOutputDebugInfo) {
DrawGraph(MathUtils.FloatToDouble(audiodata), fileName + "_audiodata.png");
}
// Calculate duration in ms
double duration = (double) audiodata.Length / SAMPLING_RATE * 1000;
// Explode samples to the range of 16 bit shorts (–32,768 to 32,767)
// Matlab multiplies with 2^15 (32768)
// e.g. if( max(abs(speech))<=1 ), speech = speech * 2^15; end;
MathUtils.Multiply(ref audiodata, AUDIO_MULTIPLIER);
// zero pad if the audio file is too short to perform a mfcc
if (audiodata.Length < (WINDOW_SIZE + OVERLAP))
{
int lenNew = WINDOW_SIZE + OVERLAP;
Array.Resize<float>(ref audiodata, lenNew);
}
// work config
WorkUnitParameterObject param = new WorkUnitParameterObject();
param.AudioSamples = audiodata;
param.PathToAudioFile = filePath.FullName;
param.MillisecondsToProcess = SECONDS_TO_ANALYZE * 1000;
param.StartAtMilliseconds = 0;
param.FileName = fileName;
param.DurationInMs = duration;
param.Tags = GetTagInfoFromFile(filePath.FullName);
Dbg.WriteLine ("Get Audio File Parameters - Execution Time: {0} ms", t.Stop().TotalMilliseconds);
return param;
}
/// <summary>
/// Computes a Scms model from the MFCC representation of a song.
/// </summary>
/// <param name="mfcc">Comirva.Audio.Util.Maths.Matrix mfcc</param>
/// <returns></returns>
public static Scms GetScms(Comirva.Audio.Util.Maths.Matrix mfccs, string name)
{
DbgTimer t = new DbgTimer();
t.Start();
Comirva.Audio.Util.Maths.Matrix mean = mfccs.Mean(2);
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE)
{
if (Analyzer.DEBUG_OUTPUT_TEXT)
{
mean.WriteText(name + "_mean.txt");
}
mean.DrawMatrixGraph(name + "_mean.png");
}
#endif
// Covariance
Comirva.Audio.Util.Maths.Matrix covarMatrix = mfccs.Cov(mean);
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE)
{
if (Analyzer.DEBUG_OUTPUT_TEXT)
{
covarMatrix.WriteText(name + "_covariance.txt");
}
covarMatrix.DrawMatrixGraph(name + "_covariance.png");
}
#endif
// Inverse Covariance
Comirva.Audio.Util.Maths.Matrix covarMatrixInv;
try {
covarMatrixInv = covarMatrix.InverseGausJordan();
} catch (Exception) {
Dbg.WriteLine("MatrixSingularException - Scms failed!");
return(null);
}
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE)
{
if (Analyzer.DEBUG_OUTPUT_TEXT)
{
covarMatrixInv.WriteAscii(name + "_inverse_covariance.ascii");
}
covarMatrixInv.DrawMatrixGraph(name + "_inverse_covariance.png");
}
#endif
// Store the Mean, Covariance, Inverse Covariance in an optimal format.
int dim = mean.Rows;
Scms s = new Scms(dim);
int l = 0;
for (int i = 0; i < dim; i++)
{
s.mean[i] = (float)mean.MatrixData[i][0];
for (int j = i; j < dim; j++)
{
s.cov[l] = (float)covarMatrix.MatrixData[i][j];
s.icov[l] = (float)covarMatrixInv.MatrixData[i][j];
l++;
}
}
Dbg.WriteLine("Compute Scms - Execution Time: {0} ms", t.Stop().TotalMilliseconds);
return(s);
}
/// <summary>
/// Computes a Scms model from the MFCC representation of a song.
/// </summary>
/// <param name="mfcc">Mirage.Matrix mfcc</param>
/// <returns></returns>
public static Scms GetScms(Matrix mfcc, string name)
{
DbgTimer t = new DbgTimer();
t.Start();
// Mean
Vector m = mfcc.Mean();
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE) {
m.WriteText(name + "_mean_orig.txt");
m.DrawMatrixGraph(name + "_mean_orig.png");
}
#endif
// Covariance
Matrix c = mfcc.Covariance(m);
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE) {
c.WriteText(name + "_covariance_orig.txt");
c.DrawMatrixGraph(name + "_covariance_orig.png");
}
#endif
// Inverse Covariance
Matrix ic;
try {
ic = c.Inverse();
} catch (MatrixSingularException) {
//throw new ScmsImpossibleException();
Dbg.WriteLine("MatrixSingularException - Scms failed!");
return null;
}
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE) {
ic.WriteAscii(name + "_inverse_covariance_orig.txt");
ic.DrawMatrixGraph(name + "_inverse_covariance_orig.png");
}
#endif
// Store the Mean, Covariance, Inverse Covariance in an optimal format.
int dim = m.rows;
Scms s = new Scms(dim);
int l = 0;
for (int i = 0; i < dim; i++) {
s.mean[i] = m.d[i, 0];
for (int j = i; j < dim; j++) {
s.cov[l] = c.d[i, j];
s.icov[l] = ic.d[i, j];
l++;
}
}
Dbg.WriteLine("(Mirage) - scms created in: {0} ms", t.Stop().TotalMilliseconds);
return s;
}
/// <summary>
/// Computes a Scms model from the MFCC representation of a song.
/// </summary>
/// <param name="mfcc">Mirage.Matrix mfcc</param>
/// <returns></returns>
public static Scms GetScms(Matrix mfcc, string name)
{
DbgTimer t = new DbgTimer();
t.Start();
// Mean
Vector m = mfcc.Mean();
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE)
{
if (Analyzer.DEBUG_OUTPUT_TEXT)
{
m.WriteText(name + "_mean_orig.txt");
}
m.DrawMatrixGraph(name + "_mean_orig.png");
}
#endif
// Covariance
Matrix c = mfcc.Covariance(m);
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE)
{
if (Analyzer.DEBUG_OUTPUT_TEXT)
{
c.WriteText(name + "_covariance_orig.txt");
}
c.DrawMatrixGraph(name + "_covariance_orig.png");
}
#endif
// Inverse Covariance
Matrix ic;
try {
ic = c.Inverse();
} catch (MatrixSingularException) {
//throw new ScmsImpossibleException();
Dbg.WriteLine("MatrixSingularException - Scms failed!");
return(null);
}
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE)
{
if (Analyzer.DEBUG_OUTPUT_TEXT)
{
ic.WriteAscii(name + "_inverse_covariance_orig.txt");
}
ic.DrawMatrixGraph(name + "_inverse_covariance_orig.png");
}
#endif
// Store the Mean, Covariance, Inverse Covariance in an optimal format.
int dim = m.rows;
Scms s = new Scms(dim);
int l = 0;
for (int i = 0; i < dim; i++)
{
s.mean[i] = m.d[i, 0];
for (int j = i; j < dim; j++)
{
s.cov[l] = c.d[i, j];
s.icov[l] = ic.d[i, j];
l++;
}
}
Dbg.WriteLine("(Mirage) - scms created in: {0} ms", t.Stop().TotalMilliseconds);
return(s);
}
public static float[] DecodeUsingSox(string fileIn, int srate, int secondsToAnalyze)
{
lock (_locker) {
using (Process toraw = new Process())
{
fileIn = Regex.Replace(fileIn, "%20", " ");
DbgTimer t = new DbgTimer();
t.Start();
String curdir = System.Environment.CurrentDirectory;
Dbg.WriteLine("Decoding: " + fileIn);
String tempFile = System.IO.Path.GetTempFileName();
String raw = tempFile + "_raw.wav";
Dbg.WriteLine("Temporary raw file: " + raw);
toraw.StartInfo.FileName = "./NativeLibraries\\sox\\sox.exe";
toraw.StartInfo.Arguments = " \"" + fileIn + "\" -r " + srate + " -e float -b 32 -G -t raw \"" + raw + "\" channels 1";
toraw.StartInfo.UseShellExecute = false;
toraw.StartInfo.RedirectStandardOutput = true;
toraw.StartInfo.RedirectStandardError = true;
toraw.Start();
toraw.WaitForExit();
int exitCode = toraw.ExitCode;
// 0 = succesfull
// 1 = partially succesful
// 2 = failed
if (exitCode != 0)
{
string standardError = toraw.StandardError.ReadToEnd();
Console.Out.WriteLine(standardError);
return(null);
}
#if DEBUG
string standardOutput = toraw.StandardOutput.ReadToEnd();
Console.Out.WriteLine(standardOutput);
#endif
float[] floatBuffer;
FileStream fs = null;
try {
FileInfo fi = new FileInfo(raw);
fs = fi.OpenRead();
int bytes = (int)fi.Length;
int samples = bytes / sizeof(float);
if ((samples * sizeof(float)) != bytes)
{
return(null);
}
// if the audio file is larger than seconds to analyze,
// find a proper section to exctract
if (bytes > secondsToAnalyze * srate * sizeof(float))
{
int seekto = (bytes / 2) - ((secondsToAnalyze / 2) * sizeof(float) * srate);
Dbg.WriteLine("Extracting section: seekto = " + seekto);
bytes = (secondsToAnalyze) * srate * sizeof(float);
fs.Seek((samples / 2 - (secondsToAnalyze / 2) * srate) * sizeof(float), SeekOrigin.Begin);
}
BinaryReader br = new BinaryReader(fs);
byte[] bytesBuffer = new byte[bytes];
br.Read(bytesBuffer, 0, bytesBuffer.Length);
int items = (int)bytes / sizeof(float);
floatBuffer = new float[items];
for (int i = 0; i < items; i++)
{
floatBuffer[i] = BitConverter.ToSingle(bytesBuffer, i * sizeof(float)); // * 65536.0f;
}
} catch (System.IO.FileNotFoundException) {
floatBuffer = null;
} finally {
if (fs != null)
{
fs.Close();
}
try
{
File.Delete(tempFile);
File.Delete(raw);
}
catch (IOException io)
{
Console.WriteLine(io);
}
Dbg.WriteLine("Decoding Execution Time: " + t.Stop().TotalMilliseconds + " ms");
}
return(floatBuffer);
}
}
}
/// <summary>
/// Method to analyse and add all the different types of audio features
/// </summary>
/// <param name="filePath">full file path</param>
/// <param name="db">Scms database (Mirage)</param>
/// <param name="repository">Soundfingerprinting Repository</param>
/// <param name="doOutputDebugInfo">decide whether to output debug info like spectrogram and audiofile (default value can be set)</param>
/// <param name="useHaarWavelet">decide whether to use haar wavelet compression or DCT compression</param>
/// <returns>true if successful</returns>
public static bool AnalyzeAndAddComplete(FileInfo filePath, Db db, Repository repository, bool doOutputDebugInfo=DEFAULT_DEBUG_INFO, bool useHaarWavelet = true) {
DbgTimer t = new DbgTimer();
t.Start ();
// get work config from the audio file
WorkUnitParameterObject param = GetWorkUnitParameterObjectFromAudioFile(filePath);
if (param == null) return false;
param.FingerprintingConfiguration = fingerprintingConfigCreation;
string fileName = param.FileName;
// build track
Track track = new Track();
track.Title = param.FileName;
track.TrackLengthMs = (int) param.DurationInMs;
track.FilePath = param.PathToAudioFile;
track.Tags = param.Tags;
track.Id = -1; // this will be set by the insert method
double[][] logSpectrogram;
List<bool[]> fingerprints;
if (repository.InsertTrackInDatabaseUsingSamples(track, param.FingerprintingConfiguration.NumberOfHashTables, param.FingerprintingConfiguration.NumberOfKeys, param, out logSpectrogram, out fingerprints)) {
// store logSpectrogram as Matrix
try {
Comirva.Audio.Util.Maths.Matrix logSpectrogramMatrix = new Comirva.Audio.Util.Maths.Matrix(logSpectrogram);
logSpectrogramMatrix = logSpectrogramMatrix.Transpose();
#region Output debugging information (Saving spectrograms and/or csv files)
if (doOutputDebugInfo) {
logSpectrogramMatrix.DrawMatrixImageLogValues(fileName + "_matrix_spectrogram.png", true);
if (DEBUG_OUTPUT_TEXT) {
logSpectrogramMatrix.WriteCSV(fileName + "_matrix_spectrogram.csv", ";");
}
// Save debug images using fingerprinting methods
SaveFingerprintingDebugImages(fileName, logSpectrogram, fingerprints, repository.FingerprintService, param.FingerprintingConfiguration);
}
#endregion
// Insert Statistical Cluster Model Similarity Audio Feature as well
if (!AnalyseAndAddScmsUsingLogSpectrogram(logSpectrogramMatrix, param, db, track.Id, doOutputDebugInfo, useHaarWavelet)) {
Dbg.WriteLine("AnalyzeAndAddComplete - Failed inserting Statistical Cluster Model Similarity Audio Feature");
// Failed, but ignore!
}
} catch (Exception e) {
Dbg.WriteLine("AnalyzeAndAddComplete - Failed creating Statistical Cluster Model Similarity Audio Feature");
Dbg.WriteLine(e.Message);
// Failed, but ignore!
}
} else {
// Failed
return false;
}
Dbg.WriteLine("AnalyzeAndAddComplete - Total Execution Time: {0} ms", t.Stop().TotalMilliseconds);
return true;
}
public static float[] DecodeUsingMplayerAndSox(string fileIn, int srate, int secondsToAnalyze) {
lock (_locker) {
using (Process tosoxreadable = new Process())
{
fileIn = Regex.Replace(fileIn, "%20", " ");
DbgTimer t = new DbgTimer();
t.Start();
String curdir = System.Environment.CurrentDirectory;
Dbg.WriteLine("Decoding: " + fileIn);
String tempFile = System.IO.Path.GetTempFileName();
String soxreadablewav = tempFile + ".wav";
Dbg.WriteLine("Temporary wav file: " + soxreadablewav);
tosoxreadable.StartInfo.FileName = "./NativeLibraries\\mplayer\\mplayer.exe";
tosoxreadable.StartInfo.Arguments = " -quiet -vc null -vo null -ao pcm:fast:waveheader \""+fileIn+"\" -ao pcm:file=\\\""+soxreadablewav+"\\\"";
tosoxreadable.StartInfo.UseShellExecute = false;
tosoxreadable.StartInfo.RedirectStandardOutput = true;
tosoxreadable.StartInfo.RedirectStandardError = true;
tosoxreadable.Start();
tosoxreadable.WaitForExit();
int exitCode = tosoxreadable.ExitCode;
// 0 = succesfull
// 1 = partially succesful
// 2 = failed
if (exitCode != 0) {
string standardError = tosoxreadable.StandardError.ReadToEnd();
Console.Out.WriteLine(standardError);
return null;
}
#if DEBUG
string standardOutput = tosoxreadable.StandardOutput.ReadToEnd();
Console.Out.WriteLine(standardOutput);
#endif
float[] floatBuffer = null;
if (File.Exists(soxreadablewav)) {
floatBuffer = DecodeUsingSox(soxreadablewav, srate, secondsToAnalyze);
try
{
File.Delete(tempFile);
File.Delete(soxreadablewav);
}
catch (IOException io)
{
Console.WriteLine(io);
}
}
Dbg.WriteLine("Decoding Execution Time: " + t.Stop().TotalMilliseconds + " ms");
return floatBuffer;
}
}
}
public static AudioFeature AnalyzeSoundfingerprinting(FileInfo filePath, bool doOutputDebugInfo=DEFAULT_DEBUG_INFO, bool useHaarWavelet = true)
{
DbgTimer t = new DbgTimer();
t.Start ();
float[] audiodata = AudioFileReader.Decode(filePath.FullName, SAMPLING_RATE, SECONDS_TO_ANALYZE);
if (audiodata == null || audiodata.Length == 0) {
Dbg.WriteLine("Error! - No Audio Found");
return null;
}
// Read TAGs using BASS
FindSimilar.AudioProxies.BassProxy bass = FindSimilar.AudioProxies.BassProxy.Instance;
Un4seen.Bass.AddOn.Tags.TAG_INFO tag_info = bass.GetTagInfoFromFile(filePath.FullName);
// Name of file being processed
string name = StringUtils.RemoveNonAsciiCharacters(Path.GetFileNameWithoutExtension(filePath.Name));
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE) {
if (DEBUG_OUTPUT_TEXT) WriteAscii(audiodata, name + "_audiodata.ascii");
if (DEBUG_OUTPUT_TEXT) WriteF3Formatted(audiodata, name + "_audiodata.txt");
}
#endif
if (doOutputDebugInfo) {
DrawGraph(MathUtils.FloatToDouble(audiodata), name + "_audiodata.png");
}
// Calculate duration in ms
double duration = (double) audiodata.Length / SAMPLING_RATE * 1000;
// zero pad if the audio file is too short to perform a mfcc
if (audiodata.Length < (fingerprintingConfig.WdftSize + fingerprintingConfig.Overlap))
{
int lenNew = fingerprintingConfig.WdftSize + fingerprintingConfig.Overlap;
Array.Resize<float>(ref audiodata, lenNew);
}
// Get fingerprint signatures using the Soundfingerprinting methods
// Get database
DatabaseService databaseService = DatabaseService.Instance;
IPermutations permutations = new LocalPermutations("Soundfingerprinting\\perms.csv", ",");
Repository repository = new Repository(permutations, databaseService, fingerprintService);
// Image Service
ImageService imageService = new ImageService(
fingerprintService.SpectrumService,
fingerprintService.WaveletService);
// work config
WorkUnitParameterObject param = new WorkUnitParameterObject();
param.FingerprintingConfiguration = fingerprintingConfig;
param.AudioSamples = audiodata;
param.PathToAudioFile = filePath.FullName;
param.MillisecondsToProcess = SECONDS_TO_ANALYZE * 1000;
param.StartAtMilliseconds = 0;
// build track
Track track = new Track();
track.Title = name;
track.TrackLengthMs = (int) duration;
track.FilePath = filePath.FullName;
track.Id = -1; // this will be set by the insert method
#region parse tag_info
if (tag_info != null) {
Dictionary<string, string> tags = new Dictionary<string, string>();
//if (tag_info.title != string.Empty) tags.Add("title", tag_info.title);
if (tag_info.artist != string.Empty) tags.Add("artist", tag_info.artist);
if (tag_info.album != string.Empty) tags.Add("album", tag_info.album);
if (tag_info.albumartist != string.Empty) tags.Add("albumartist", tag_info.albumartist);
if (tag_info.year != string.Empty) tags.Add("year", tag_info.year);
if (tag_info.comment != string.Empty) tags.Add("comment", tag_info.comment);
if (tag_info.genre != string.Empty) tags.Add("genre", tag_info.genre);
if (tag_info.track != string.Empty) tags.Add("track", tag_info.track);
if (tag_info.disc != string.Empty) tags.Add("disc", tag_info.disc);
if (tag_info.copyright != string.Empty) tags.Add("copyright", tag_info.copyright);
if (tag_info.encodedby != string.Empty) tags.Add("encodedby", tag_info.encodedby);
if (tag_info.composer != string.Empty) tags.Add("composer", tag_info.composer);
if (tag_info.publisher != string.Empty) tags.Add("publisher", tag_info.publisher);
if (tag_info.lyricist != string.Empty) tags.Add("lyricist", tag_info.lyricist);
if (tag_info.remixer != string.Empty) tags.Add("remixer", tag_info.remixer);
if (tag_info.producer != string.Empty) tags.Add("producer", tag_info.producer);
if (tag_info.bpm != string.Empty) tags.Add("bpm", tag_info.bpm);
//if (tag_info.filename != string.Empty) tags.Add("filename", tag_info.filename);
tags.Add("channelinfo", tag_info.channelinfo.ToString());
//if (tag_info.duration > 0) tags.Add("duration", tag_info.duration.ToString());
if (tag_info.bitrate > 0) tags.Add("bitrate", tag_info.bitrate.ToString());
if (tag_info.replaygain_track_gain != -100f) tags.Add("replaygain_track_gain", tag_info.replaygain_track_gain.ToString());
if (tag_info.replaygain_track_peak != -1f) tags.Add("replaygain_track_peak", tag_info.replaygain_track_peak.ToString());
if (tag_info.conductor != string.Empty) tags.Add("conductor", tag_info.conductor);
if (tag_info.grouping != string.Empty) tags.Add("grouping", tag_info.grouping);
if (tag_info.mood != string.Empty) tags.Add("mood", tag_info.mood);
if (tag_info.rating != string.Empty) tags.Add("rating", tag_info.rating);
if (tag_info.isrc != string.Empty) tags.Add("isrc", tag_info.isrc);
foreach(var nativeTag in tag_info.NativeTags) {
string[] keyvalue = nativeTag.Split('=');
tags.Add(keyvalue[0], keyvalue[1]);
}
track.Tags = tags;
}
#endregion
AudioFeature audioFeature = null;
double[][] logSpectrogram;
if (repository.InsertTrackInDatabaseUsingSamples(track, 25, 4, param, out logSpectrogram)) {
if (doOutputDebugInfo) {
imageService.GetLogSpectralImages(logSpectrogram, fingerprintingConfig.Stride, fingerprintingConfig.FingerprintLength, fingerprintingConfig.Overlap, 2).Save(name + "_specgram_logimages.png");
Comirva.Audio.Util.Maths.Matrix logSpectrogramMatrix = new Comirva.Audio.Util.Maths.Matrix(logSpectrogram);
logSpectrogramMatrix = logSpectrogramMatrix.Transpose();
logSpectrogramMatrix.DrawMatrixImageLogValues(name + "_specgram_logimage.png", true);
if (DEBUG_OUTPUT_TEXT) {
logSpectrogramMatrix.WriteCSV(name + "_specgram_log.csv", ";");
}
}
audioFeature = new DummyAudioFeature();
// Store duration
audioFeature.Duration = (long) duration;
// Store file name
audioFeature.Name = filePath.FullName;
} else {
// failed
}
Dbg.WriteLine ("Soundfingerprinting - Total Execution Time: {0} ms", t.Stop().TotalMilliseconds);
return audioFeature;
}
public double[][] CreateLogSpectrogram(
float[] samples, IWindowFunction windowFunction, AudioServiceConfiguration configuration)
{
DbgTimer t = new DbgTimer();
t.Start ();
if (configuration.NormalizeSignal)
{
NormalizeInPlace(samples);
}
int width = (samples.Length - configuration.WindowSize) / configuration.Overlap; /*width of the image*/
double[][] frames = new double[width][];
int[] logFrequenciesIndexes = GenerateLogFrequencies(configuration);
double[] window = windowFunction.GetWindow();
for (int i = 0; i < width; i++)
{
double[] complexSignal = new double[2 * configuration.WindowSize]; /*even - Re, odd - Img, thats how Exocortex works*/
// take 371 ms each 11.6 ms (2048 samples each 64 samples, samplerate 5512)
// or 256 ms each 16 ms (8192 samples each 512 samples, samplerate 32000)
for (int j = 0; j < configuration.WindowSize; j++)
{
// Weight by Hann Window
complexSignal[2 * j] = window[j] * samples[(i * configuration.Overlap) + j];
// need to clear out as fft modifies buffer (phase)
complexSignal[(2 * j) + 1] = 0;
}
lomonFFT.TableFFT(complexSignal, true);
frames[i] = ExtractLogBins(complexSignal, logFrequenciesIndexes, configuration.LogBins);
}
Dbg.WriteLine ("Create Log Spectrogram - Execution Time: {0} ms", t.Stop().TotalMilliseconds);
return frames;
}
// TODO: Rememeber to use another stride when querying
public static Dictionary<Track, double> SimilarTracksSoundfingerprinting(FileInfo filePath)
{
DbgTimer t = new DbgTimer();
t.Start ();
FindSimilar.AudioProxies.BassProxy bass = FindSimilar.AudioProxies.BassProxy.Instance;
float[] audiodata = AudioFileReader.Decode(filePath.FullName, SAMPLING_RATE, SECONDS_TO_ANALYZE);
if (audiodata == null || audiodata.Length == 0) {
Dbg.WriteLine("Error! - No Audio Found");
return null;
}
// Name of file being processed
string name = StringUtils.RemoveNonAsciiCharacters(Path.GetFileNameWithoutExtension(filePath.Name));
// Calculate duration in ms
double duration = (double) audiodata.Length / SAMPLING_RATE * 1000;
// Explode samples to the range of 16 bit shorts (–32,768 to 32,767)
// Matlab multiplies with 2^15 (32768)
// e.g. if( max(abs(speech))<=1 ), speech = speech * 2^15; end;
MathUtils.Multiply(ref audiodata, AUDIO_MULTIPLIER); // 65536
// zero pad if the audio file is too short to perform a mfcc
if (audiodata.Length < (fingerprintingConfig.WdftSize + fingerprintingConfig.Overlap))
{
int lenNew = fingerprintingConfig.WdftSize + fingerprintingConfig.Overlap;
Array.Resize<float>(ref audiodata, lenNew);
}
// Get fingerprint signatures using the Soundfingerprinting methods
// Get database
DatabaseService databaseService = DatabaseService.Instance;
IPermutations permutations = new LocalPermutations("Soundfingerprinting\\perms.csv", ",");
Repository repository = new Repository(permutations, databaseService, fingerprintService);
// work config
WorkUnitParameterObject param = new WorkUnitParameterObject();
param.FingerprintingConfiguration = fingerprintingConfig;
param.PathToAudioFile = filePath.FullName;
param.AudioSamples = audiodata;
param.MillisecondsToProcess = SECONDS_TO_ANALYZE * 1000;
param.StartAtMilliseconds = 0;
Dictionary<Track, double> candidates = repository.FindSimilarFromAudioSamples(25, 4, 2, param);
return candidates;
/*
// Use var keyword to enumerate dictionary
foreach (var pair in candidates)
{
Console.WriteLine("{0} - {1:0.00}",
pair.Key.Title,
pair.Value);
}
*/
Dbg.WriteLine ("Soundfingerprinting - Total Execution Time: {0} ms", t.Stop().TotalMilliseconds);
}
/// <summary>
/// Query the database for perceptually similar tracks using the sound fingerprinting methods
/// </summary>
/// <param name="filePath">input file</param>
/// <param name="repository">the database (repository)</param>
/// <param name="thresholdTables">Minimum number of hash tables that must be found for one signature to be considered a candidate (0 and 1 = return all candidates, 2+ = return only exact matches)</param>
/// <param name="optimizeSignatureCount">Reduce the number of signatures in order to increase the search performance</param>
/// <param name="doSearchEverything">disregard the local sensitivity hashes and search the whole database</param>
/// <param name="splashScreen">The "please wait" splash screen (or null)</param>
/// <returns>a list of query results objects (e.g. similar tracks)</returns>
public static List<FindSimilar.QueryResult> SimilarTracksSoundfingerprintingList(FileInfo filePath,
Repository repository,
int thresholdTables,
bool optimizeSignatureCount,
bool doSearchEverything,
SplashSceenWaitingForm splashScreen) {
DbgTimer t = new DbgTimer();
t.Start ();
if (splashScreen != null) splashScreen.SetProgress(0, "Reading audio file ...");
// get work config from the audio file
WorkUnitParameterObject param = GetWorkUnitParameterObjectFromAudioFile(filePath);
if (param == null) {
if (splashScreen != null) splashScreen.SetProgress(0, "Failed reading audio file!");
return null;
}
param.FingerprintingConfiguration = fingerprintingConfigQuerying;
if (splashScreen != null) splashScreen.SetProgress(1, "Successfully reading audio file!");
// This is how the threshold tables work:
// For each signature created from a query file we retrieve a number of candidates
// based on how many fingerprints that are associated to the same hash bucket.
// if the number of fingerprints associated to the same hash bucket is relatively high
// the likelyhood for this being an exact match is also very high.
// Therefore a value of 0 or 1 basically means return every track that has an association
// to the same hash bucket, while a number higher than that increases the accuracy for
// only matching identical matches.
// 0 and 1 returns many matches
// 2 returns sometimes only the one we search for (exact match)
List<FindSimilar.QueryResult> similarFiles = repository.FindSimilarFromAudioSamplesList(param.FingerprintingConfiguration.NumberOfHashTables,
param.FingerprintingConfiguration.NumberOfKeys,
thresholdTables,
param,
optimizeSignatureCount,
doSearchEverything,
splashScreen);
Dbg.WriteLine ("SimilarTracksSoundfingerprintingList - Total Execution Time: {0} ms", t.Stop().TotalMilliseconds);
return similarFiles;
}
public static AudioFeature AnalyzeMandelEllis(FileInfo filePath, bool doOutputDebugInfo=DEFAULT_DEBUG_INFO)
{
DbgTimer t = new DbgTimer();
t.Start ();
// get work config from the audio file
WorkUnitParameterObject param = GetWorkUnitParameterObjectFromAudioFile(filePath);
// Calculate the audio feature
AudioFeature audioFeature = mandelEllisExtractor.Calculate(MathUtils.FloatToDouble(param.AudioSamples));
if (audioFeature != null) {
// Store duration
audioFeature.Duration = (long) param.DurationInMs;
// Store file name
audioFeature.Name = filePath.Name;
}
Dbg.WriteLine ("MandelEllisExtractor - Total Execution Time: {0} ms", t.Stop().TotalMilliseconds);
return audioFeature;
}
public static float[] Decode(string fileIn, int srate, int secondsToAnalyze)
{
DbgTimer t = new DbgTimer();
t.Start();
float[] floatBuffer = null;
// check if file exists
if (fileIn != null && fileIn != "")
{
FileInfo fi = new FileInfo(fileIn);
if (!fi.Exists)
{
Console.Out.WriteLine("No file found {0}!", fileIn);
return(null);
}
}
// Try to use Un4Seen Bass
BassProxy bass = BassProxy.Instance;
double duration = bass.GetDurationInSeconds(fileIn);
if (duration > 0)
{
Dbg.WriteLine("Using BASS to decode the file ...");
// duration in seconds
if (duration > secondsToAnalyze)
{
// find segment to extract
double startSeconds = (duration / 2 - (secondsToAnalyze / 2));
if (startSeconds < 0)
{
startSeconds = 0;
}
floatBuffer = bass.ReadMonoFromFile(fileIn, srate, secondsToAnalyze * 1000, (int)(startSeconds * 1000));
// if this failes, the duration read from the tags was wrong or it is something wrong with the audio file
if (floatBuffer == null)
{
IOUtils.LogMessageToFile(Mir.WARNING_FILES_LOG, fileIn);
}
}
else
{
// return whole file
floatBuffer = bass.ReadMonoFromFile(fileIn, srate, 0, 0);
// if this failes, the duration read from the tags was wrong or it is something wrong with the audio file
if (floatBuffer == null)
{
IOUtils.LogMessageToFile(Mir.WARNING_FILES_LOG, fileIn);
}
}
}
// Bass failed reading or never even tried, so use another alternative
if (floatBuffer == null)
{
Dbg.WriteLine("Using MPlayer and SOX to decode the file ...");
fileIn = Regex.Replace(fileIn, "%20", " ");
floatBuffer = DecodeUsingMplayerAndSox(fileIn, srate, secondsToAnalyze);
}
return(floatBuffer);
}
public static AudioFeature AnalyzeScms(FileInfo filePath, bool doOutputDebugInfo=DEFAULT_DEBUG_INFO, bool useHaarWavelet = true)
{
DbgTimer t = new DbgTimer();
t.Start ();
// get work config from the audio file
WorkUnitParameterObject param = GetWorkUnitParameterObjectFromAudioFile(filePath, doOutputDebugInfo);
string fileName = param.FileName;
// used to save wave files in the debug inverse methods
FindSimilar.AudioProxies.BassProxy bass = FindSimilar.AudioProxies.BassProxy.Instance;
// 2. Windowing
// 3. FFT
Comirva.Audio.Util.Maths.Matrix stftdata = stftMirage.Apply(param.AudioSamples);
if (DEBUG_INFO_VERBOSE & DEBUG_OUTPUT_TEXT) {
stftdata.WriteAscii(fileName + "_stftdata.ascii");
stftdata.WriteCSV(fileName + "_stftdata.csv", ";");
}
if (doOutputDebugInfo) {
// same as specgram(audio*32768, 2048, 44100, hanning(2048), 1024);
//stftdata.DrawMatrixImageLogValues(fileName + "_specgram.png", true);
// spec gram with log values for the y axis (frequency)
stftdata.DrawMatrixImageLogY(fileName + "_specgramlog.png", SAMPLING_RATE, 20, SAMPLING_RATE/2, 120, WINDOW_SIZE);
}
if (DEBUG_DO_INVERSE_TESTS) {
#region Inverse STFT
double[] audiodata_inverse_stft = stftMirage.InverseStft(stftdata);
// divide
//MathUtils.Divide(ref audiodata_inverse_stft, AUDIO_MULTIPLIER);
MathUtils.Normalize(ref audiodata_inverse_stft);
if (DEBUG_OUTPUT_TEXT) {
WriteAscii(audiodata_inverse_stft, fileName + "_audiodata_inverse_stft.ascii");
WriteF3Formatted(audiodata_inverse_stft, fileName + "_audiodata_inverse_stft.txt");
}
DrawGraph(audiodata_inverse_stft, fileName + "_audiodata_inverse_stft.png");
float[] audiodata_inverse_float = MathUtils.DoubleToFloat(audiodata_inverse_stft);
bass.SaveFile(audiodata_inverse_float, fileName + "_inverse_stft.wav", Analyzer.SAMPLING_RATE);
#endregion
}
// 4. Mel Scale Filterbank
// Mel-frequency is proportional to the logarithm of the linear frequency,
// reflecting similar effects in the human's subjective aural perception)
// 5. Take Logarithm
// 6. DCT (Discrete Cosine Transform)
if (DEBUG_INFO_VERBOSE) {
#region Mel Scale and Log Values
Comirva.Audio.Util.Maths.Matrix mellog = mfccMirage.ApplyMelScaleAndLog(ref stftdata);
if (DEBUG_OUTPUT_TEXT) {
mellog.WriteCSV(fileName + "_mel_log.csv", ";");
}
if (doOutputDebugInfo) {
mellog.DrawMatrixImage(fileName + "_mel_log.png", 600, 400, true, true);
}
#endregion
#region Inverse Mel Scale and Log Values
if (DEBUG_DO_INVERSE_TESTS) {
Comirva.Audio.Util.Maths.Matrix inverse_mellog = mfccMirage.InverseMelScaleAndLog(ref mellog);
inverse_mellog.WriteCSV(fileName + "_mel_log_inverse.csv", ";");
inverse_mellog.DrawMatrixImageLogValues(fileName + "_mel_log_inverse.png", true);
double[] audiodata_inverse_mellog = stftMirage.InverseStft(inverse_mellog);
//MathUtils.Divide(ref audiodata_inverse_mellog, AUDIO_MULTIPLIER/100);
MathUtils.Normalize(ref audiodata_inverse_mellog);
if (DEBUG_OUTPUT_TEXT) {
WriteAscii(audiodata_inverse_mellog, fileName + "_audiodata_inverse_mellog.ascii");
WriteF3Formatted(audiodata_inverse_mellog, fileName + "_audiodata_inverse_mellog.txt");
}
DrawGraph(audiodata_inverse_mellog, fileName + "_audiodata_inverse_mellog.png");
float[] audiodata_inverse_mellog_float = MathUtils.DoubleToFloat(audiodata_inverse_mellog);
bass.SaveFile(audiodata_inverse_mellog_float, fileName + "_inverse_mellog.wav", Analyzer.SAMPLING_RATE);
}
#endregion
}
Comirva.Audio.Util.Maths.Matrix featureData = null;
if (useHaarWavelet) {
#region Wavelet Transform
int lastHeight = 0;
int lastWidth = 0;
featureData = mfccMirage.ApplyMelScaleAndWaveletCompress(ref stftdata, out lastHeight, out lastWidth);
if (DEBUG_INFO_VERBOSE & DEBUG_OUTPUT_TEXT) {
featureData.WriteAscii(fileName + "_waveletdata.ascii");
}
if (doOutputDebugInfo) {
featureData.DrawMatrixImageLogValues(fileName + "_waveletdata.png", true);
}
if (DEBUG_DO_INVERSE_TESTS) {
#region Inverse Wavelet
// try to do an inverse wavelet transform
Comirva.Audio.Util.Maths.Matrix stftdata_inverse_wavelet = mfccMirage.InverseMelScaleAndWaveletCompress(ref featureData, lastHeight, lastWidth);
if (DEBUG_OUTPUT_TEXT) stftdata_inverse_wavelet.WriteCSV(fileName + "_specgramlog_inverse_wavelet.csv", ";");
stftdata_inverse_wavelet.DrawMatrixImageLogValues(fileName + "_specgramlog_inverse_wavelet.png", true);
double[] audiodata_inverse_wavelet = stftMirage.InverseStft(stftdata_inverse_wavelet);
MathUtils.Normalize(ref audiodata_inverse_wavelet);
if (DEBUG_OUTPUT_TEXT) WriteF3Formatted(audiodata_inverse_wavelet, fileName + "_audiodata_inverse_wavelet.txt");
DrawGraph(audiodata_inverse_wavelet, fileName + "_audiodata_inverse_wavelet.png");
bass.SaveFile(MathUtils.DoubleToFloat(audiodata_inverse_wavelet), fileName + "_inverse_wavelet.wav", Analyzer.SAMPLING_RATE);
#endregion
}
#endregion
} else {
#region DCT Transform
// It seems the Mirage way of applying the DCT is slightly faster than the
// Comirva way due to less loops
featureData = mfccMirage.ApplyMelScaleDCT(ref stftdata);
//featureData = mfccMirage.ApplyComirvaWay(ref stftdata);
if (DEBUG_INFO_VERBOSE & DEBUG_OUTPUT_TEXT) {
featureData.WriteAscii(fileName + "_mfccdata.ascii");
}
if (doOutputDebugInfo) {
featureData.DrawMatrixImageLogValues(fileName + "_mfccdata.png", true);
}
if (DEBUG_DO_INVERSE_TESTS) {
#region Inverse MFCC
// try to do an inverse mfcc
Comirva.Audio.Util.Maths.Matrix stftdata_inverse_mfcc = mfccMirage.InverseMelScaleDCT(ref featureData);
if (DEBUG_OUTPUT_TEXT) stftdata_inverse_mfcc.WriteCSV(fileName + "_stftdata_inverse_mfcc.csv", ";");
stftdata_inverse_mfcc.DrawMatrixImageLogValues(fileName + "_specgramlog_inverse_mfcc.png", true);
double[] audiodata_inverse_mfcc = stftMirage.InverseStft(stftdata_inverse_mfcc);
MathUtils.Normalize(ref audiodata_inverse_mfcc);
if (DEBUG_OUTPUT_TEXT) WriteF3Formatted(audiodata_inverse_mfcc, fileName + "_audiodata_inverse_mfcc.txt");
DrawGraph(audiodata_inverse_mfcc, fileName + "_audiodata_inverse_mfcc.png");
bass.SaveFile(MathUtils.DoubleToFloat(audiodata_inverse_mfcc), fileName + "_inverse_mfcc.wav", Analyzer.SAMPLING_RATE);
#endregion
}
#endregion
}
// Store in a Statistical Cluster Model Similarity class.
// A Gaussian representation of a song
Scms audioFeature = Scms.GetScms(featureData, fileName);
if (audioFeature != null) {
// Store image if debugging
if (doOutputDebugInfo) {
audioFeature.Image = featureData.DrawMatrixImageLogValues(fileName + "_featuredata.png", true, false, 0, 0, true);
}
// Store bitstring hash as well
string hashString = GetBitString(featureData);
audioFeature.BitString = hashString;
// Store duration
audioFeature.Duration = (long) param.DurationInMs;
// Store file name
audioFeature.Name = filePath.FullName;
} else {
// failed creating the Scms class
Console.Out.WriteLine("Failed! Could not compute the Scms {0}!", fileName);
}
Dbg.WriteLine ("AnalyzeScms - Total Execution Time: {0} ms", t.Stop().TotalMilliseconds);
return audioFeature;
}
/// <summary>
/// Method to analyze and add using the soundfingerprinting methods
/// </summary>
/// <param name="filePath">full file path</param>
/// <param name="repository">Soundfingerprinting Repository</param>
/// <param name="doOutputDebugInfo">decide whether to output debug info like spectrogram and audiofile (default value can be set)</param>
/// <param name="useHaarWavelet">decide whether to use haar wavelet compression or DCT compression</param>
/// <returns>true if successful</returns>
public static bool AnalyzeAndAddSoundfingerprinting(FileInfo filePath, Repository repository, bool doOutputDebugInfo=DEFAULT_DEBUG_INFO, bool useHaarWavelet = true)
{
DbgTimer t = new DbgTimer();
t.Start ();
// get work config from the audio file
WorkUnitParameterObject param = GetWorkUnitParameterObjectFromAudioFile(filePath);
param.FingerprintingConfiguration = fingerprintingConfigCreation;
string fileName = param.FileName;
// build track
Track track = new Track();
track.Title = param.FileName;
track.TrackLengthMs = (int) param.DurationInMs;
track.FilePath = param.PathToAudioFile;
track.Tags = param.Tags;
track.Id = -1; // this will be set by the insert method
// Get fingerprint signatures using the Soundfingerprinting methods
double[][] logSpectrogram;
List<bool[]> fingerprints;
List<double[][]> spectralImages;
if (repository.InsertTrackInDatabaseUsingSamples(track, param.FingerprintingConfiguration.NumberOfHashTables, param.FingerprintingConfiguration.NumberOfKeys, param, out logSpectrogram, out fingerprints, out spectralImages)) {
// store logSpectrogram as Matrix
Comirva.Audio.Util.Maths.Matrix logSpectrogramMatrix = new Comirva.Audio.Util.Maths.Matrix(logSpectrogram);
logSpectrogramMatrix = logSpectrogramMatrix.Transpose();
#region Debug for Soundfingerprinting Method
if (doOutputDebugInfo) {
// Image Service
ImageService imageService = new ImageService(repository.FingerprintService.SpectrumService, repository.FingerprintService.WaveletService);
imageService.GetLogSpectralImages(logSpectrogram, fingerprintingConfigCreation.Stride, fingerprintingConfigCreation.FingerprintLength, fingerprintingConfigCreation.Overlap, 2).Save(fileName + "_specgram_logimages.png");
logSpectrogramMatrix.DrawMatrixImageLogValues(fileName + "_specgram_logimage.png", true);
if (DEBUG_OUTPUT_TEXT) {
logSpectrogramMatrix.WriteCSV(fileName + "_specgram_log.csv", ";");
}
}
#endregion
} else {
// failed
Console.Out.WriteLine("Failed! Could not compute the soundfingerprint {0}!", fileName);
return false;
}
Dbg.WriteLine ("AnalyzeAndAddSoundfingerprinting - Total Execution Time: {0} ms", t.Stop().TotalMilliseconds);
return true;
}
//private static Mfcc mfccOptimized = new Mfcc(WINDOW_SIZE, SAMPLING_RATE, MEL_COEFFICIENTS, MFCC_COEFFICIENTS);
//private static MFCC mfccComirva = new MFCC(SAMPLING_RATE, WINDOW_SIZE, MFCC_COEFFICIENTS, true, 20.0, SAMPLING_RATE/2, MEL_COEFFICIENTS);
#endif
#region Methods
public static bool AnalyzeAndAdd(FileInfo filePath, Db db, DatabaseService databaseService, bool doOutputDebugInfo=DEFAULT_DEBUG_INFO, bool useHaarWavelet = true)
{
DbgTimer t = new DbgTimer();
t.Start ();
float[] audiodata = AudioFileReader.Decode(filePath.FullName, SAMPLING_RATE, SECONDS_TO_ANALYZE);
if (audiodata == null || audiodata.Length == 0) {
Dbg.WriteLine("Error! - No Audio Found");
return false;
}
// Read TAGs using BASS
FindSimilar.AudioProxies.BassProxy bass = FindSimilar.AudioProxies.BassProxy.Instance;
Un4seen.Bass.AddOn.Tags.TAG_INFO tag_info = bass.GetTagInfoFromFile(filePath.FullName);
// Name of file being processed
string name = StringUtils.RemoveNonAsciiCharacters(Path.GetFileNameWithoutExtension(filePath.Name));
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE) {
if (DEBUG_OUTPUT_TEXT) WriteAscii(audiodata, name + "_audiodata.ascii");
if (DEBUG_OUTPUT_TEXT) WriteF3Formatted(audiodata, name + "_audiodata.txt");
}
#endif
if (doOutputDebugInfo) {
DrawGraph(MathUtils.FloatToDouble(audiodata), name + "_audiodata.png");
}
// Calculate duration in ms
double duration = (double) audiodata.Length / SAMPLING_RATE * 1000;
// Explode samples to the range of 16 bit shorts (–32,768 to 32,767)
// Matlab multiplies with 2^15 (32768)
// e.g. if( max(abs(speech))<=1 ), speech = speech * 2^15; end;
MathUtils.Multiply(ref audiodata, AUDIO_MULTIPLIER); // 65536
// zero pad if the audio file is too short to perform a mfcc
if (audiodata.Length < (fingerprintingConfig.WdftSize + fingerprintingConfig.Overlap))
{
int lenNew = fingerprintingConfig.WdftSize + fingerprintingConfig.Overlap;
Array.Resize<float>(ref audiodata, lenNew);
}
// Get fingerprint signatures using the Soundfingerprinting methods
IPermutations permutations = new LocalPermutations("Soundfingerprinting\\perms.csv", ",");
Repository repository = new Repository(permutations, databaseService, fingerprintService);
// Image Service
ImageService imageService = new ImageService(
fingerprintService.SpectrumService,
fingerprintService.WaveletService);
// work config
WorkUnitParameterObject param = new WorkUnitParameterObject();
param.FingerprintingConfiguration = fingerprintingConfig;
param.AudioSamples = audiodata;
param.PathToAudioFile = filePath.FullName;
param.MillisecondsToProcess = SECONDS_TO_ANALYZE * 1000;
param.StartAtMilliseconds = 0;
// build track
Track track = new Track();
track.Title = name;
track.TrackLengthMs = (int) duration;
track.FilePath = filePath.FullName;
track.Id = -1; // this will be set by the insert method
#region parse tag_info
if (tag_info != null) {
Dictionary<string, string> tags = new Dictionary<string, string>();
//if (tag_info.title != string.Empty) tags.Add("title", tag_info.title);
if (tag_info.artist != string.Empty) tags.Add("artist", tag_info.artist);
if (tag_info.album != string.Empty) tags.Add("album", tag_info.album);
if (tag_info.albumartist != string.Empty) tags.Add("albumartist", tag_info.albumartist);
if (tag_info.year != string.Empty) tags.Add("year", tag_info.year);
if (tag_info.comment != string.Empty) tags.Add("comment", tag_info.comment);
if (tag_info.genre != string.Empty) tags.Add("genre", tag_info.genre);
if (tag_info.track != string.Empty) tags.Add("track", tag_info.track);
if (tag_info.disc != string.Empty) tags.Add("disc", tag_info.disc);
if (tag_info.copyright != string.Empty) tags.Add("copyright", tag_info.copyright);
if (tag_info.encodedby != string.Empty) tags.Add("encodedby", tag_info.encodedby);
if (tag_info.composer != string.Empty) tags.Add("composer", tag_info.composer);
if (tag_info.publisher != string.Empty) tags.Add("publisher", tag_info.publisher);
if (tag_info.lyricist != string.Empty) tags.Add("lyricist", tag_info.lyricist);
if (tag_info.remixer != string.Empty) tags.Add("remixer", tag_info.remixer);
if (tag_info.producer != string.Empty) tags.Add("producer", tag_info.producer);
if (tag_info.bpm != string.Empty) tags.Add("bpm", tag_info.bpm);
//if (tag_info.filename != string.Empty) tags.Add("filename", tag_info.filename);
tags.Add("channelinfo", tag_info.channelinfo.ToString());
//if (tag_info.duration > 0) tags.Add("duration", tag_info.duration.ToString());
if (tag_info.bitrate > 0) tags.Add("bitrate", tag_info.bitrate.ToString());
if (tag_info.replaygain_track_gain != -100f) tags.Add("replaygain_track_gain", tag_info.replaygain_track_gain.ToString());
if (tag_info.replaygain_track_peak != -1f) tags.Add("replaygain_track_peak", tag_info.replaygain_track_peak.ToString());
if (tag_info.conductor != string.Empty) tags.Add("conductor", tag_info.conductor);
if (tag_info.grouping != string.Empty) tags.Add("grouping", tag_info.grouping);
if (tag_info.mood != string.Empty) tags.Add("mood", tag_info.mood);
if (tag_info.rating != string.Empty) tags.Add("rating", tag_info.rating);
if (tag_info.isrc != string.Empty) tags.Add("isrc", tag_info.isrc);
foreach(var nativeTag in tag_info.NativeTags) {
string[] keyvalue = nativeTag.Split('=');
tags.Add(keyvalue[0], keyvalue[1]);
}
track.Tags = tags;
}
#endregion
double[][] logSpectrogram;
if (repository.InsertTrackInDatabaseUsingSamples(track, 25, 4, param, out logSpectrogram)) {
// store logSpectrogram as Matrix
Comirva.Audio.Util.Maths.Matrix logSpectrogramMatrix = new Comirva.Audio.Util.Maths.Matrix(logSpectrogram);
logSpectrogramMatrix = logSpectrogramMatrix.Transpose();
#region Debug for Soundfingerprinting Method
if (doOutputDebugInfo) {
imageService.GetLogSpectralImages(logSpectrogram, fingerprintingConfig.Stride, fingerprintingConfig.FingerprintLength, fingerprintingConfig.Overlap, 2).Save(name + "_specgram_logimages.png");
logSpectrogramMatrix.DrawMatrixImageLogValues(name + "_specgram_logimage.png", true);
if (DEBUG_OUTPUT_TEXT) {
logSpectrogramMatrix.WriteCSV(name + "_specgram_log.csv", ";");
}
}
#endregion
#region Insert Statistical Cluster Model Similarity Audio Feature as well
Comirva.Audio.Util.Maths.Matrix scmsMatrix = null;
if (useHaarWavelet) {
#region Wavelet Transform
int lastHeight = 0;
int lastWidth = 0;
scmsMatrix = mfccMirage.ApplyWaveletCompression(ref logSpectrogramMatrix, out lastHeight, out lastWidth);
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE) {
if (DEBUG_OUTPUT_TEXT) scmsMatrix.WriteAscii(name + "_waveletdata.ascii");
}
#endif
if (doOutputDebugInfo) {
scmsMatrix.DrawMatrixImageLogValues(name + "_waveletdata.png", true);
}
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE) {
#region Inverse Wavelet
// try to do an inverse wavelet transform
Comirva.Audio.Util.Maths.Matrix stftdata_inverse_wavelet = mfccMirage.InverseWaveletCompression(ref scmsMatrix, lastHeight, lastWidth, logSpectrogramMatrix.Rows, logSpectrogramMatrix.Columns);
if (DEBUG_OUTPUT_TEXT) stftdata_inverse_wavelet.WriteCSV(name + "_specgramlog_inverse_wavelet.csv", ";");
stftdata_inverse_wavelet.DrawMatrixImageLogValues(name + "_specgramlog_inverse_wavelet.png", true);
#endregion
}
#endif
#endregion
} else {
#region DCT Transform
// It seems the Mirage way of applying the DCT is slightly faster than the
// Comirva way due to less loops
scmsMatrix = mfccMirage.ApplyDCT(ref logSpectrogramMatrix);
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE) {
if (DEBUG_OUTPUT_TEXT) scmsMatrix.WriteAscii(name + "_mfccdata.ascii");
}
#endif
if (doOutputDebugInfo) {
scmsMatrix.DrawMatrixImageLogValues(name + "_mfccdata.png", true);
}
#if DEBUG
if (Analyzer.DEBUG_INFO_VERBOSE) {
#region Inverse MFCC
// try to do an inverse mfcc
Comirva.Audio.Util.Maths.Matrix stftdata_inverse_mfcc = mfccMirage.InverseDCT(ref scmsMatrix);
if (DEBUG_OUTPUT_TEXT) stftdata_inverse_mfcc.WriteCSV(name + "_stftdata_inverse_mfcc.csv", ";");
stftdata_inverse_mfcc.DrawMatrixImageLogValues(name + "_specgramlog_inverse_mfcc.png", true);
#endregion
}
#endif
#endregion
}
// Store in a Statistical Cluster Model Similarity class.
// A Gaussian representation of a song
Scms audioFeature = Scms.GetScms(scmsMatrix, name);
if (audioFeature != null) {
// Store image if debugging
if (doOutputDebugInfo) {
audioFeature.Image = scmsMatrix.DrawMatrixImageLogValues(name + "_featuredata.png", true, false, 0, 0, true);
}
// Store bitstring hash as well
string hashString = GetBitString(scmsMatrix);
audioFeature.BitString = hashString;
// Store duration
audioFeature.Duration = (long) duration;
// Store file name
audioFeature.Name = filePath.FullName;
int id = track.Id;
if (db.AddTrack(ref id, audioFeature) == -1) {
Console.Out.WriteLine("Failed! Could not add audioFeature to database {0}!", name);
}
}
#endregion
} else {
// failed
return false;
}
Dbg.WriteLine ("AnalyzeAndAdd - Total Execution Time: {0} ms", t.Stop().TotalMilliseconds);
return true;
}
public static float[] DecodeUsingMplayer(string fileIn, int srate) {
lock (_locker) {
using (Process towav = new Process())
{
fileIn = Regex.Replace(fileIn, "%20", " ");
DbgTimer t = new DbgTimer();
t.Start();
String curdir = System.Environment.CurrentDirectory;
Dbg.WriteLine("Decoding: " + fileIn);
String tempFile = System.IO.Path.GetTempFileName();
String wav = tempFile + ".wav";
Dbg.WriteLine("Temporary wav file: " + wav);
towav.StartInfo.FileName = "./NativeLibraries\\mplayer\\mplayer.exe";
towav.StartInfo.Arguments = " -quiet -ao pcm:fast:waveheader \""+fileIn+"\" -format floatle -af resample="+srate+":0:2,pan=1:0.5:0.5 -channels 1 -vo null -vc null -ao pcm:file=\\\""+wav+"\\\"";
towav.StartInfo.UseShellExecute = false;
towav.StartInfo.RedirectStandardOutput = true;
towav.StartInfo.RedirectStandardError = true;
towav.Start();
towav.WaitForExit();
int exitCode = towav.ExitCode;
// 0 = succesfull
// 1 = partially succesful
// 2 = failed
if (exitCode != 0) {
string standardError = towav.StandardError.ReadToEnd();
Console.Out.WriteLine(standardError);
return null;
}
#if DEBUG
string standardOutput = towav.StandardOutput.ReadToEnd();
Console.Out.WriteLine(standardOutput);
#endif
RiffRead riff = new RiffRead(wav);
riff.Process();
float[] floatBuffer = riff.SoundData[0];
try
{
File.Delete(tempFile);
//File.Delete(wav);
}
catch (IOException io)
{
Console.WriteLine(io);
}
Dbg.WriteLine("Decoding Execution Time: " + t.Stop().TotalMilliseconds + " ms");
return floatBuffer;
}
}
}
/// <summary>
/// Find Similar Tracks to one or many audio files using their unique database id(s)
/// </summary>
/// <param name="id">an array of unique database ids for the audio files to search for similar matches</param>
/// <param name="exclude">an array of unique database ids to ignore (normally the same as the id array)</param>
/// <param name="db">database</param>
/// <param name="analysisMethod">analysis method (SCMS or MandelEllis)</param>
/// <param name="numToTake">max number of entries to return</param>
/// <param name="percentage">percentage below and above the duration in ms when querying (used if between 0.1 - 0.9)</param>
/// <param name="distanceType">distance method to use (KullbackLeiblerDivergence is default)</param>
/// <returns>a dictinary list of key value pairs (filepath and distance)</returns>
public static Dictionary<KeyValuePair<int, string>, double> SimilarTracks(int[] id, int[] exclude, Db db, Analyzer.AnalysisMethod analysisMethod, int numToTake=25, double percentage=0.2, AudioFeature.DistanceType distanceType = AudioFeature.DistanceType.KullbackLeiblerDivergence)
{
DbgTimer t = new DbgTimer();
t.Start();
AudioFeature[] seedAudioFeatures = null;
AudioFeature[] audioFeatures = null;
switch (analysisMethod) {
case Analyzer.AnalysisMethod.MandelEllis:
seedAudioFeatures = new MandelEllis[id.Length];
audioFeatures = new MandelEllis[100];
break;
case Analyzer.AnalysisMethod.SCMS:
seedAudioFeatures = new Scms[id.Length];
audioFeatures = new Scms[100];
break;
}
for (int i = 0; i < seedAudioFeatures.Length; i++) {
seedAudioFeatures[i] = db.GetTrack(id[i], analysisMethod);
}
// Get all tracks from the DB except the seedSongs
IDataReader r = db.GetTracks(exclude, seedAudioFeatures[0].Duration, percentage);
// store results in a dictionary
var NameDictionary = new Dictionary<KeyValuePair<int, string>, double>();
int[] mapping = new int[100];
int read = 1;
double d;
double dcur;
float count;
while (read > 0) {
read = db.GetNextTracks(ref r, ref audioFeatures, ref mapping, 100, analysisMethod);
for (int i = 0; i < read; i++) {
d = 0;
count = 0;
for (int j = 0; j < seedAudioFeatures.Length; j++) {
dcur = seedAudioFeatures[j].GetDistance(audioFeatures[i], distanceType);
// convert to positive values
dcur = Math.Abs(dcur);
d += dcur;
count++;
}
if (d > 0) {
NameDictionary.Add(new KeyValuePair<int,string>(mapping[i], audioFeatures[i].Name), d/count);
//NameDictionary.Add(new KeyValuePair<int,string>(mapping[i], String.Format("{0} ({1} ms)", audioFeatures[i].Name, audioFeatures[i].Duration)), d/count);
}
}
}
// sort by non unique values
var sortedDict = (from entry in NameDictionary orderby entry.Value ascending select entry)
.Take(numToTake)
.ToDictionary(pair => pair.Key, pair => pair.Value);
Console.Out.WriteLine(String.Format("Found Similar to ({0}) in {1} ms", String.Join(",", seedAudioFeatures.Select(p=>p.Name)), t.Stop().TotalMilliseconds));
return sortedDict;
}
public static float[] DecodeUsingSox(string fileIn, int srate, int secondsToAnalyze) {
lock (_locker) {
using (Process toraw = new Process())
{
fileIn = Regex.Replace(fileIn, "%20", " ");
DbgTimer t = new DbgTimer();
t.Start();
String curdir = System.Environment.CurrentDirectory;
Dbg.WriteLine("Decoding: " + fileIn);
String tempFile = System.IO.Path.GetTempFileName();
String raw = tempFile + "_raw.wav";
Dbg.WriteLine("Temporary raw file: " + raw);
toraw.StartInfo.FileName = "./NativeLibraries\\sox\\sox.exe";
toraw.StartInfo.Arguments = " \"" + fileIn + "\" -r "+srate+" -e float -b 32 -G -t raw \"" + raw + "\" channels 1";
toraw.StartInfo.UseShellExecute = false;
toraw.StartInfo.RedirectStandardOutput = true;
toraw.StartInfo.RedirectStandardError = true;
toraw.Start();
toraw.WaitForExit();
int exitCode = toraw.ExitCode;
// 0 = succesfull
// 1 = partially succesful
// 2 = failed
if (exitCode != 0) {
string standardError = toraw.StandardError.ReadToEnd();
Console.Out.WriteLine(standardError);
return null;
}
#if DEBUG
string standardOutput = toraw.StandardOutput.ReadToEnd();
Console.Out.WriteLine(standardOutput);
#endif
float[] floatBuffer;
FileStream fs = null;
try {
FileInfo fi = new FileInfo(raw);
fs = fi.OpenRead();
int bytes = (int)fi.Length;
int samples = bytes/sizeof(float);
if ((samples*sizeof(float)) != bytes)
return null;
// if the audio file is larger than seconds to analyze,
// find a proper section to exctract
if (bytes > secondsToAnalyze*srate*sizeof(float)) {
int seekto = (bytes/2) - ((secondsToAnalyze/2)*sizeof(float)*srate);
Dbg.WriteLine("Extracting section: seekto = " + seekto);
bytes = (secondsToAnalyze)*srate*sizeof(float);
fs.Seek((samples/2-(secondsToAnalyze/2)*srate)*sizeof(float), SeekOrigin.Begin);
}
BinaryReader br = new BinaryReader(fs);
byte[] bytesBuffer = new byte[bytes];
br.Read(bytesBuffer, 0, bytesBuffer.Length);
int items = (int)bytes/sizeof(float);
floatBuffer = new float[items];
for (int i = 0; i < items; i++) {
floatBuffer[i] = BitConverter.ToSingle(bytesBuffer, i * sizeof(float)); // * 65536.0f;
}
} catch (System.IO.FileNotFoundException) {
floatBuffer = null;
} finally {
if (fs != null)
fs.Close();
try
{
File.Delete(tempFile);
File.Delete(raw);
}
catch (IOException io)
{
Console.WriteLine(io);
}
Dbg.WriteLine("Decoding Execution Time: " + t.Stop().TotalMilliseconds + " ms");
}
return floatBuffer;
}
}
}
private static void Test()
{
mirageaudio_initgst();
string song1_filename = "/home/lorentz/Music/Library/Pachelbel/Johann Pachelbel - Canon And Gigue In D Major For 3 Violins And Basso Continuo.mp3";
string song2_filename = "/home/lorentz/Music/Library/Karajan Adagios/CD 1/Pachelbel - Canon in d Major (Kanon And Gigue in d Major = d Dur) av Johann Pachelbel.mp3";
Scms song1 = null;
Scms song2 = null;
DbgTimer t1 = new DbgTimer();
t1.Start();
int runs = 10;
for (int i = 0; i < runs; i++) {
Analyzer.Analyze(song1_filename);
}
long l1 = 0;
t1.Stop(ref l1);
Dbg.WriteLine("Analysis: " + runs + " times - " + l1 + "ms; " +
(double)l1/(double)runs + "ms per analysis");
song1 = Analyzer.Analyze(song1_filename);
song2 = Analyzer.Analyze(song2_filename);
ScmsConfiguration config = new ScmsConfiguration (Analyzer.MFCC_COEFFICIENTS);
Console.WriteLine("Distance = " + Scms.Distance (song1, song2, config));
DbgTimer t2 = new DbgTimer();
t2.Start();
runs = 100000;
for (int i = 0; i < runs; i++) {
Scms.Distance (song1, song2, config);
}
long l2 = 0;
t2.Stop(ref l2);
Dbg.WriteLine("Distance Computation: " + runs + " times - " + l2 + "ms; " +
(double)l2/(double)runs + "ms per comparison");
}
