public void AddVectorsFromDB (PCAnalyzer ana)
{
Dictionary<int, Mirage.Vector> vectorMap = null;
lock (BansheeLibraryAnalyzer.Singleton.db_synch) {
vectorMap = BansheeLibraryAnalyzer.Singleton.db.GetMirageMaxVectors ();
}
if (vectorMap == null)
throw new DatabaseException ("vectorMap is null!");
foreach (int bid in vectorMap.Keys) {
try {
TrackInfo ti = DatabaseTrackInfo.Provider.FetchSingle (bid) as TrackInfo;
if (ti == null) {
if (!BansheeLibraryAnalyzer.Singleton.updating_db)
new Thread (new ThreadStart (BansheeLibraryAnalyzer.Singleton.RemoveDeletedTracks))
.Start ();
continue;
}
if (!ana.AddEntry (bid, BansheeLibraryAnalyzer.Singleton.
ConvertMirageVector (vectorMap [bid]),
ti.Duration.TotalSeconds))
throw new Exception("AddEntry failed!");
} catch (Exception e) {
Hyena.Log.Exception ("NoNoise - PCA Problem", e);
}
}
}
/// <summary>
/// Checks for each track in the music library if there is already
/// MIR data in the database. If not, computes the MFCC matrix and
/// stores it in the database.
/// </summary>
private void PcaForMusicLibrary()
{
PCAnalyzer ana = new PCAnalyzer();
// if (gatherMIRdata) {
Banshee.Library.MusicLibrarySource ml = ServiceManager.SourceManager.MusicLibrary;
Mirage.Matrix mfcc;
Dictionary<int, Mirage.Matrix> mfccMap = db.GetMirageMatrices ();
for (int i = 0; i < ml.TrackModel.Count; i++) {
try {
TrackInfo ti = ml.TrackModel [i];
string absPath = ti.Uri.AbsolutePath;
int bid = ml.GetTrackIdForUri (ti.Uri);
// WARN: A bid could theoretically be inserted/deleted between GetMirageMatrices ()
// and CointainsMirDataForTrack () such that if and else fail
if (!db.ContainsMirDataForTrack (bid)) {
mfcc = Mirage.Analyzer.AnalyzeMFCC (absPath);
if (!db.InsertMatrix (mfcc, bid))
Hyena.Log.Error ("NoNoise - Matrix insert failed");
} else
mfcc = mfccMap[bid];
if (!ana.AddEntry (bid, ConvertMfccMean (mfcc.Mean ())))
throw new Exception("AddEntry failed!");
// if (!ana.AddEntry (bid, ConvertMfccMean(mfcc.Mean()), ti.Duration.TotalSeconds))
// throw new Exception("AddEntry failed!");
// if (!ana.AddEntry (bid, null, ti.Bpm, ti.Duration.TotalSeconds))
// throw new Exception("AddEntry failed!");
} catch (Exception e) {
Hyena.Log.Exception("NoNoise - MFCC/DB Problem", e);
}
}
// } else {
// Dictionary<int, Matrix> mfccMap = db.GetMirageMatrices ();
//
// foreach (int key in mfccMap.Keys) {
// try {
// if (!ana.AddEntry (key, ConvertMfccMean (mfccMap[key].Mean ())))
// throw new Exception ("AddEntry failed!");
// } catch (Exception e) {
// Hyena.Log.Exception ("NoNoise - PCA Problem", e);
// }
// }
// }
try {
ana.PerformPCA ();
// Hyena.Log.Debug(ana.GetCoordinateStrings ());
List<DataEntry> coords = ana.Coordinates;
db.ClearPcaData ();
if (!db.InsertPcaCoordinates (coords))
Hyena.Log.Error ("NoNoise - PCA coord insert failed");
} catch (Exception e) {
Hyena.Log.Exception("PCA Problem", e);
}
}
public void AddVectorsFromDB (PCAnalyzer ana)
{
Dictionary<int, Mirage.Vector> vectorMap = null;
lock (BansheeLibraryAnalyzer.Singleton.db_synch) {
vectorMap = BansheeLibraryAnalyzer.Singleton.db.GetMirageMaxVectors ();
}
if (vectorMap == null)
throw new DatabaseException ("vectorMap is null!");
foreach (int bid in vectorMap.Keys) {
try {
if (!ana.AddEntry (bid, BansheeLibraryAnalyzer.Singleton.
ConvertMirageVector (vectorMap [bid])))
throw new Exception("AddEntry failed!");
} catch (Exception e) {
Hyena.Log.Exception("NoNoise - PCA Problem", e);
}
}
}
/// <summary>
/// Test method to (manually) check if the pca computations are correct.
/// Prints the base vectors computed with a mirage generated covariance
/// matrix and with the classical self-implemented covariance computation.
/// </summary>
private void PcaEigenvectorBasisTest()
{
Mirage.Matrix m = new Mirage.Matrix (3, 5);
m.d [0, 0] = 0.2f;
m.d [1, 0] = 14.5f;
m.d [2, 0] = 166.0f;
m.d [0, 1] = 1.5f;
m.d [1, 1] = 20.5f;
m.d [2, 1] = 233.0f;
m.d [0, 2] = 0.8f;
m.d [1, 2] = 16.2f;
m.d [2, 2] = 189.0f;
m.d [0, 3] = 2.3f;
m.d [1, 3] = 11.7f;
m.d [2, 3] = 139.0f;
m.d [0, 4] = 1.7f;
m.d [1, 4] = 17.9f;
m.d [2, 4] = 206.0f;
PCAnalyzer pca = new PCAnalyzer (m.Covariance (m.Mean ()));
pca.PcaTest ();
int i = -1;
double[] data = { 0.2, 14.5, 166.0 };
PCAnalyzer ana = new PCAnalyzer ();
ana.AddEntry (i--, data);
data = new double[] { 1.5, 20.5, 233.0 };
ana.AddEntry (i--, data);
data = new double[] { 0.8, 16.2, 189.0 };
ana.AddEntry (i--, data);
data = new double[] { 2.3, 11.7, 139.0 };
ana.AddEntry (i--, data);
data = new double[] { 1.7, 17.9, 206.0 };
ana.AddEntry (i--, data);
ana.PerformPCA ();
}