marf.FeatureExtraction.FFT
Class FFT

java.lang.Object
  marf.FeatureExtraction.FeatureExtraction
      marf.FeatureExtraction.FFT.FFT

All Implemented Interfaces:

StorageManager

public class FFT

extends FeatureExtraction

Class FFT

Implements Fast Fourier Transform

$Header: /cvsroot/marf/marf/src/marf/FeatureExtraction/FFT/FFT.java,v 1.34 2004/04/20 16:38:35 mokhov Exp $

Author:

Stephen Sinclair

Field Summary

static int

DEFAULT_CHUNK_SIZE Default number (1024) of doubles per chunk in the window.

Fields inherited from class marf.FeatureExtraction.FeatureExtraction

adFeatures, oPreprocessing

Fields inherited from interface marf.Storage.StorageManager

DUMP_BINARY, DUMP_CSV_TEXT, DUMP_GZIP_BINARY, DUMP_XML

Constructor Summary

FFT(Preprocessing poPreprocessing)
FFT Constructor

Method Summary

static void	DoFFT(double[] InputReal, double[] InputImag, double[] OutputReal, double[] OutputImag, int direction) FFT algorithm, translated from "Numerical Recipes in C++" Implements the Fast Fourier Transform, which performs a discrete Fourier transform in O(n*log(n)).
void	dump() Not Implemented
boolean	extractFeatures() FFT Implementation of extractFeatures()
static void	NormalFFT(double[] sample, double[] magnitude) Performs a normal FFT, taking a real input (supposedly an audio sample) and returns the frequency analysis in terms of "magnitude".
static void	NormalFFT(double[] sample, double[] magnitude, double[] phaseAngle) Performs a normal FFT, taking a real input (supposedly an audio sample) and returns the frequency analysis in terms of "magnitude" and "phase angle".
void	restore() Not Implemented

Methods inherited from class marf.FeatureExtraction.FeatureExtraction

getFeaturesArray, hamming, hamming

Methods inherited from class java.lang.Object

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

Field Detail

DEFAULT_CHUNK_SIZE

public static final int DEFAULT_CHUNK_SIZE

Default number (1024) of doubles per chunk in the window. Feature vector will be half of the chunk size.

See Also:

Constant Field Values

Constructor Detail

FFT

public FFT(Preprocessing poPreprocessing)

FFT Constructor

Parameters:

poPreprocessing - Preprocessing module reference

Method Detail

extractFeatures

public final boolean extractFeatures()
                              throws FeatureExtractionException

FFT Implementation of extractFeatures()

Specified by:

extractFeatures in class FeatureExtraction

Returns:

true if there were features extracted, false otherwise

Throws:

FeatureExtractionException

dump

public void dump()
          throws java.io.IOException

Not Implemented

Throws:

java.io.IOException

restore

public void restore()
             throws java.io.IOException

Not Implemented

Throws:

java.io.IOException

DoFFT

public static final void DoFFT(double[] InputReal,
                               double[] InputImag,
                               double[] OutputReal,
                               double[] OutputImag,
                               int direction)
                        throws FeatureExtractionException

FFT algorithm, translated from "Numerical Recipes in C++" Implements the Fast Fourier Transform, which performs a discrete Fourier transform in O(n*log(n)).

Parameters:

InputReal - InputReal is real part of input array

InputImag - InputImag is imaginary part of input array

OutputReal - OutputReal is real part of output array

OutputImag - OutputImag is imaginary part of output array

direction - Direction is 1 for normal FFT, -1 for inverse FFT

Throws:

FeatureExtractionException

NormalFFT

public static final void NormalFFT(double[] sample,
                                   double[] magnitude,
                                   double[] phaseAngle)
                            throws FeatureExtractionException

Performs a normal FFT, taking a real input (supposedly an audio sample) and returns the frequency analysis in terms of "magnitude" and "phase angle".

Parameters:

sample - must be an array of size (2^k)

magnitude - must be half the size of "sample"

phaseAngle - must be half the size of "sample"

Throws:

FeatureExtractionException

NormalFFT

public static final void NormalFFT(double[] sample,
                                   double[] magnitude)
                            throws FeatureExtractionException

Performs a normal FFT, taking a real input (supposedly an audio sample) and returns the frequency analysis in terms of "magnitude".

Parameters:

sample - must be an array of size (2^k)

magnitude - must be half the size of "sample"

Throws:

FeatureExtractionException