com.tagtraum.jipes.math

Class GriffinLim

java.lang.Object

com.tagtraum.jipes.math.GriffinLim

All Implemented Interfaces:

Transform

public class GriffinLim
extends Object
implements Transform

Transform for signal estimation from modified short-time fourier transform by Daniel W. Griffin and Jae and S. Lim.

Author:

Hendrik Schreiber

See Also:

Signal estimation from modified short-time fourier transform

Constructor Summary

Constructors

Constructor and Description
GriffinLim(float[] realStart, float[] imaginaryStart, int maxIterations)
GriffinLim(float[] realStart, float[] imaginaryStart, int maxIterations, float maxDistance)

Method Summary

Modifier and Type	Method and Description
float[][]	inverseTransform(float[] real, float[] imaginary) Performs an inverse transform.
float[][]	transform(float[] magnitudes) Transform the given magnitudes to a signal.
float[][]	transform(float[] real, float[] imaginary) Performs a complex transform.

Methods inherited from class java.lang.Object

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

Constructor Detail

GriffinLim

public GriffinLim(float[] realStart,
                  float[] imaginaryStart,
                  int maxIterations)

Parameters:

realStart - first real estimate for the signal

imaginaryStart - first imaginary estimate (may be zero-filled array or null)

maxIterations - max number of iterations to perform

GriffinLim

public GriffinLim(float[] realStart,
                  float[] imaginaryStart,
                  int maxIterations,
                  float maxDistance)

Parameters:

realStart - first real estimate for the signal

imaginaryStart - first imaginary estimate (may be zero-filled array or null)

maxIterations - max number of iterations to perform

maxDistance - stop, if the distance (sum of squared differences) to the previous magnitude spectrum is less than this value.

Method Detail

transform

public float[][] transform(float[] magnitudes)
                    throws UnsupportedOperationException

Transform the given magnitudes to a signal.

Specified by:

transform in interface Transform

Parameters:

magnitudes - magnitudes (must not be symmetric)

Returns:

real array, imaginary array

Throws:

UnsupportedOperationException

inverseTransform

public float[][] inverseTransform(float[] real,
                                  float[] imaginary)
                           throws UnsupportedOperationException

Description copied from interface: Transform

Performs an inverse transform.

Specified by:

inverseTransform in interface Transform

Parameters:

real - input array of floats

imaginary - input array of floats

Returns:

array with the real and the imaginary part of the transform

Throws:

UnsupportedOperationException - should the implementation not support this operation

transform

public float[][] transform(float[] real,
                           float[] imaginary)
                    throws UnsupportedOperationException

Description copied from interface: Transform

Performs a complex transform.

Specified by:

transform in interface Transform

Parameters:

real - input array of floats

imaginary - input array of floats

Returns:

array with the real and the imaginary part of the transform, the third float array contains the normalized frequencies, i.e. 1.0 is equal to the sample rate of the input

Throws:

UnsupportedOperationException - should the implementation not support this operation