Package com.tagtraum.audiokern.audioprocessor

Class AudioSpectrumFunctions

java.lang.Object

com.tagtraum.audiokern.audioprocessor.AudioSpectrumFunctions

public class AudioSpectrumFunctions
extends Object

AudioSpectrumFunctions.

Author:

Hendrik Schreiber

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static class	AudioSpectrumFunctions.AudioSpectrumAveraging	Effectively combining every three bins in a row into a single bins containing the weighted average.
static class	AudioSpectrumFunctions.LinearBackgroundRemoval	Implements a noise reduction/whitening scheme for linear-frequency spectra.
static class	AudioSpectrumFunctions.LogFrequencyBackgroundRemoval	Implements a noise reduction/whitening scheme for log-frequency spectra with an integer number of bins per semitone similar to the pre-processing defined in "Approximate note transcription for the improved identification of difficult chords" by M.
static class	AudioSpectrumFunctions.LogFrequencySpectrumToPitchAveraging	Effectively down-samples the bin resolution of a LogFrequencySpectrum to one bin per semitone by calculating a weighted average.
static class	AudioSpectrumFunctions.LogFrequencySpectrumToPitchDownsampling	Effectively down-samples the bin resolution of a LogFrequencySpectrum to one bin per semitone by picking one of multiple bins and discarding the others.

Field Summary

Fields

Modifier and Type	Field	Description
static MapFunction<AudioSpectrum>	PITCH_TO_CHROMA	Adds all corresponding semitones to a chroma vector.
static AggregateFunction<AudioSpectrum,Float>	SPECTRAL_CENTROID
static AggregateFunction<AudioSpectrum,Float>	SPECTRAL_SPREAD
static AggregateFunction<AudioSpectrum,Float>	STRONGEST_PARTIAL

Constructor Summary

Constructors

Constructor	Description
AudioSpectrumFunctions()

Method Summary

Modifier and Type	Method	Description
static AggregateFunction<AudioSpectrum,Float>	createBrightnessFunction(float cutoffFrequency)	Creates an AggregateFunction that can be used to compute brightness in an Aggregate.
static MapFunction<AudioSpectrum>	createGroupPowerSpectrumFunction()	Transforms a given spectrum into a LogFrequencySpectrum grouped according to rules similar to the ones defined in MPEG-7 for ASF (spectral flatness).
static MapFunction<AudioSpectrum>	createHighPassFilter(float minFrequency)	Creates a high pass filter that zeroes all real and imaginary values below a given frequency.
static MapFunction<AudioSpectrum>	createHighPassFilter(int minBin)	Creates a high pass filter that zeroes all real and imaginary values below a given sample/bin number.
static MapFunction<AudioSpectrum>	createLowPassFilter(float maxFrequency)	Creates a low pass filter that zeroes all real and imaginary values above a given frequency.
static MapFunction<AudioSpectrum>	createLowPassFilter(int maxBin)	Creates a low pass filter that zeroes all real and imaginary values above a given bin/sample number.
static AggregateFunction<AudioSpectrum,Integer>	createRollOffBinFunction(float percentThreshold)	Returns the bin below which the given percentage of this spectrum's magnitude resides.
static AggregateFunction<AudioSpectrum,Float>	createRollOffFunction(float percentThreshold)	Creates an AggregateFunction that can be used to compute roll-off in an Aggregate.
static MapFunction<AudioSpectrum>	createSpectralFlatnessFunction(int n, int bands)	Transforms a given spectrum into a MultiBandSpectrum containing spectral flatness values similar to MPEG-7 ASF.

Methods inherited from class java.lang.Object

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

Field Detail

STRONGEST_PARTIAL

public static AggregateFunction<AudioSpectrum,Float> STRONGEST_PARTIAL

SPECTRAL_CENTROID

public static AggregateFunction<AudioSpectrum,Float> SPECTRAL_CENTROID

SPECTRAL_SPREAD

public static AggregateFunction<AudioSpectrum,Float> SPECTRAL_SPREAD

PITCH_TO_CHROMA

public static MapFunction<AudioSpectrum> PITCH_TO_CHROMA

Adds all corresponding semitones to a chroma vector. Note that this function requires the spectrum to be in semitones already, i.e. one bin per semitone. The resulting LogFrequencySpectrum uses the first 12 frequencies of the source spectrum as its frequencies.

Constructor Detail

AudioSpectrumFunctions

public AudioSpectrumFunctions()

Method Detail

createSpectralFlatnessFunction

public static MapFunction<AudioSpectrum> createSpectralFlatnessFunction(int n,
                                                                        int bands)

Transforms a given spectrum into a MultiBandSpectrum containing spectral flatness values similar to MPEG-7 ASF. Values range between 0.0f and 1.0f, where 1.0f represents total flatness, i.e. random noise and 0.0f very pronounced tonality.

Parameters:

n - defines the lower frequency boundary with 2^(0.25*n) * 1kHz, e.g. -8

bands - number of bands, e.g. 24.

Returns:

map function

See Also:

Spectral Flatness on Wikipedia

createGroupPowerSpectrumFunction

public static MapFunction<AudioSpectrum> createGroupPowerSpectrumFunction()

Transforms a given spectrum into a LogFrequencySpectrum grouped according to rules similar to the ones defined in MPEG-7 for ASF (spectral flatness).

Returns:

map function

See Also:

Spectral Flatness on Wikipedia

createRollOffFunction

public static AggregateFunction<AudioSpectrum,Float> createRollOffFunction(float percentThreshold)

Creates an AggregateFunction that can be used to compute roll-off in an Aggregate.

Parameters:

percentThreshold - threshold in percent (i.e. 0 <= threshold <= 1), typically 0.85f or 0.95f

Returns:

a function object that computes spectral roll-off

createRollOffBinFunction

public static AggregateFunction<AudioSpectrum,Integer> createRollOffBinFunction(float percentThreshold)

Returns the bin below which the given percentage of this spectrum's magnitude resides. Note that roll-off values tend to be fairly random for spectra with very little overall energy.

Parameters:

percentThreshold - threshold in percent (i.e. 0 <= threshold <= 1), typical values are 0.85f (G. Tzanetakis, P. Cook. Musical genre classification of audio signals. IEEE Tr. Speech and Audio Processing, 10(5),293-302, 2002) or 0.95f (T. Pohle, E. Pampalk, G. Widmer. Evaluation of Frequently Used Audio Features for Classification of Music Into Perceptual Categories, Proceedings of the 4th International Workshop on Content-Based Multimedia Indexing, Riga, Latvia, June 2005.)

Returns:

a function for computing the RollOffBin

See Also:

createRollOffFunction(float)

createBrightnessFunction

public static AggregateFunction<AudioSpectrum,Float> createBrightnessFunction(float cutoffFrequency)

Creates an AggregateFunction that can be used to compute brightness in an Aggregate.

Brightness gives an indication of what percentage of the overall energy of a spectrum resides above a given cutoff frequency. Typical values are 1000Hz, 1500Hz, 3000Hz. 1500Hz can be considered a default value.

Parameters:

cutoffFrequency - cutoff frequency in Hz

Returns:

a function object that computes brightness values between 0.0f and 1.0f based on a spectrum

See Also:

Juslin, P. N. (2000). Cue utilization in communication of emotion in music performance: relating performance to perception. J Exp Psychol Hum Percept Perform. 2000 Dec;26(6):1797-813., Laukka, P., Juslin, P. N., and Bresin, R. (2005). A dimensional approach to vocal expression of emotion. Cognition and Emotion, 19, 633-653., createRollOffFunction(float)

createLowPassFilter

public static MapFunction<AudioSpectrum> createLowPassFilter(float maxFrequency)

Creates a low pass filter that zeroes all real and imaginary values above a given frequency.

Parameters:

maxFrequency - cut off frequency

Returns:

low pass filter function suitable for Mapping

createLowPassFilter

public static MapFunction<AudioSpectrum> createLowPassFilter(int maxBin)

Creates a low pass filter that zeroes all real and imaginary values above a given bin/sample number.

Parameters:

maxBin - cut off bin

Returns:

low pass filter function suitable for Mapping

createHighPassFilter

public static MapFunction<AudioSpectrum> createHighPassFilter(float minFrequency)

Creates a high pass filter that zeroes all real and imaginary values below a given frequency.

Parameters:

minFrequency - cut off frequency

Returns:

high pass filter function suitable for Mapping

createHighPassFilter

public static MapFunction<AudioSpectrum> createHighPassFilter(int minBin)

Creates a high pass filter that zeroes all real and imaginary values below a given sample/bin number.

Parameters:

minBin - cut off bin

Returns:

high pass filter function suitable for Mapping