marytts.signalproc.adaptation.outlier

Class GaussianOutlierEliminator

java.lang.Object

marytts.signalproc.adaptation.outlier.GaussianOutlierEliminator

public class GaussianOutlierEliminator
extends Object

Single Gaussian based outlier elimination. Looks at the difference distributions of aligned source-target LSF, f0, energy, and duration features. Eliminates outliers that fall outside user specified ranges. The ranges can be specified using GaussianOutlierEliminatorParams in the form of total standard deviations for each feature. It is also possible to eliminate too similar feature pairs to enforce a certain amount of dissimilarity in the training data. Reference: Türk, O., and Arslan, L. M., 2006, "Robust Processing Techniques for Voice Conversion", Computer Speech and Language 20 (2006), pp. 441-467.

Author:

Oytun Türk

Constructor Summary

Constructors

Constructor and Description
GaussianOutlierEliminator()

Method Summary

Methods

Modifier and Type	Method and Description
void	eliminate(GaussianOutlierEliminatorParams params, String codebookFileIn, String codebookFileOut)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

GaussianOutlierEliminator

public GaussianOutlierEliminator()

Method Detail

eliminate

public void eliminate(GaussianOutlierEliminatorParams params,
             String codebookFileIn,
             String codebookFileOut)