001/**
002 * Copyright (c) 2011, The University of Southampton and the individual contributors.
003 * All rights reserved.
004 *
005 * Redistribution and use in source and binary forms, with or without modification,
006 * are permitted provided that the following conditions are met:
007 *
008 *   *  Redistributions of source code must retain the above copyright notice,
009 *      this list of conditions and the following disclaimer.
010 *
011 *   *  Redistributions in binary form must reproduce the above copyright notice,
012 *      this list of conditions and the following disclaimer in the documentation
013 *      and/or other materials provided with the distribution.
014 *
015 *   *  Neither the name of the University of Southampton nor the names of its
016 *      contributors may be used to endorse or promote products derived from this
017 *      software without specific prior written permission.
018 *
019 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
020 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
021 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
022 * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
023 * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
024 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
025 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
026 * ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
027 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
028 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
029 */
030/**
031 * 
032 */
033package org.openimaj.audio.analysis;
034
035import org.openimaj.audio.SampleChunk;
036import org.openimaj.audio.processor.AudioProcessor;
037
038import edu.emory.mathcs.jtransforms.fft.FloatFFT_1D;
039
040/**
041 *      An implementation of the power cepstrum of an audio signal. The
042 *      power cepstrum of an audio signal is the squared magnitude of the Fourier 
043 *      transform of the logarithm of the squared magnitude of the Fourier transform 
044 *      of a signal. Yeah, I know.
045 *
046 *      @author David Dupplaw (dpd@ecs.soton.ac.uk)
047 *  @created 18 Jul 2012
048 *      @version $Author$, $Revision$, $Date$
049 */
050public class PowerCepstrumTransform extends AudioProcessor
051{
052        /** The last generated cepstrum */
053        private float[][] lastCepstrum = null;
054
055        @Override
056    public SampleChunk process( final SampleChunk sample ) throws Exception
057    {
058                final FourierTransform fft = new FourierTransform();
059                
060                //
061                // The squared magnitude of the Fourier transform of the logarithm 
062                // of the squared magnitude of the Fourier transform of a signal...
063                //
064                // Working backwards...
065                // ... the FFT of a signal...
066                //
067                fft.process( sample );
068                final float[][] fftCoeffs = fft.getLastFFT();
069                
070                // ...the logarithm of the squared magnitude...
071                final float logMags[][] = new float[fftCoeffs.length][];
072                for( int c = 0; c < fftCoeffs.length; c++ )
073                {
074                        logMags[c] = new float[fftCoeffs[c].length/4];
075                        for( int i = 0; i < fftCoeffs[c].length/4; i++ )
076                        {
077                                // Calculate magnitude
078                                final float re = fftCoeffs[c][i*2];
079                                final float im = fftCoeffs[c][i*2+1];
080                                float mag = (float)Math.log(Math.sqrt( re*re + im*im )+1);
081                                
082                                // Square
083                                mag *= mag;
084                                
085                                // Logarithm
086                                final float logMag = (float)Math.log( mag );
087                                
088                                // Store
089                                logMags[c][i] = logMag;
090                        }
091                }
092                
093                // ... the Fast Fourier (of the log-squared-mags)
094                this.lastCepstrum  = new float[ logMags.length ][];
095                final FloatFFT_1D fft2 = new FloatFFT_1D( logMags[0].length/4 );
096                for( int c = 0; c < logMags.length; c++ )
097                {
098                        fft2.complexForward( logMags[c] );
099                        
100                        this.lastCepstrum[c] = new float[ logMags[c].length/4 ];
101                        
102                        // ...the squared magnitude of...
103                        for( int i = 0; i < logMags[c].length/4; i++ )
104                        {
105                                // Calculate magnitude
106                                final float re = logMags[c][i*2];
107                                final float im = logMags[c][i*2+1];
108                                float mag = (float)Math.log(Math.sqrt( re*re + im*im )+1);
109                                
110                                // Square
111                                mag *= mag;
112                                
113                                this.lastCepstrum[c][i] = mag;
114                        }
115                }
116                
117            return sample;
118    }
119        
120        /**
121         *      Returns the last generated cepstrum
122         *      @return The last generated cepstrum
123         */
124        public float[][] getLastCepstrum()
125        {
126                return this.lastCepstrum;
127        }
128}