marytts.signalproc.process

Class Blizzard09PostProcessor

java.lang.Object

marytts.signalproc.process.Blizzard09PostProcessor

public class Blizzard09PostProcessor
extends Object

This class implements post-processing of TTS output to make it sound more intelligible when used in a telephone channel. Several simple ideas are implemented: (Step 1) Increasing relative energy of consonants (requires labels along with TTS wav outputs, with the same filename and folder but with .lab extension) (Step 2) Sharpening formants explicitly using LSFs (Step 3) Increasing relative energy of higher formants implicitly by adding highpass filtered version to original (Step 4) Finally, the output is gain adjusted to prevent clipping

Author:

Oytun Türk

Field Summary

Fields

Modifier and Type	Field and Description
static boolean	APPLY_HIGHPASS_FILTER
static double	CONSONANT_MAX_GAIN_FACTOR
static double	CONSONANT_MAX_GAIN_RELATIVE_DURATION
static double	FORMANT_SHARPENING_END_FREQ
static double	FORMANT_SHARPENING_START_FREQ
static double	HIGHPASS_FILTER_CUTOFF
static double	HIGHPASS_FILTER_RELATIVE_GAIN
static boolean	INCREASE_CONSONANT_GAINS
static String	LABEL_FILE_EXTENSION
static boolean	LABELS_FROM_REALISED_DURATIONS_FILE
static double	MAX_LSF_PAIR_SEPARATION_IN_HZ
static boolean	REDUCE_VOWEL_GAINS
static double	RELATIVE_DECREASE_IN_LSF_PAIR_SEPARATION
static boolean	SHARPEN_FORMANTS
static double	SKIP_SIZE_IN_SECONDS_GAIN
static double	SKIP_SIZE_IN_SECONDS_LSF
static double	VOWEL_MIN_GAIN_FACTOR
static double	VOWEL_MIN_GAIN_RELATIVE_DURATION
static double	WINDOW_SIZE_IN_SECONDS_GAIN
static double	WINDOW_SIZE_IN_SECONDS_LSF

Constructor Summary

Constructors

Constructor and Description
Blizzard09PostProcessor()

Method Summary

Methods

Modifier and Type	Method and Description
static void	main(String[] args)
static void	mainSingleFile(String inputWavFile, String outputWavFile, Allophone[] allophones)
static double[]	process(double[] x, Labels labels, Allophone[] allophones, int samplingRateInHz, double absMaxOrig)
static double[]	processGains(double[] x, int samplingRateInHz, Labels labels, boolean[] toBeProcesseds, double extremumGainFactor, double extremumGainRelativeDuration)
static double[]	processHigherFormantGains(double[] x, int samplingRateInHz, Labels labels, boolean[] isPauses)
static double[]	processLSFs(double[] x, int samplingRateInHz, Labels labels, boolean[] isVowels, boolean[] isPauses)

Methods inherited from class java.lang.Object

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

Field Detail

LABELS_FROM_REALISED_DURATIONS_FILE

public static final boolean LABELS_FROM_REALISED_DURATIONS_FILE

See Also:

Constant Field Values

LABEL_FILE_EXTENSION

public static final String LABEL_FILE_EXTENSION

See Also:

Constant Field Values

WINDOW_SIZE_IN_SECONDS_LSF

public static final double WINDOW_SIZE_IN_SECONDS_LSF

See Also:

Constant Field Values

SKIP_SIZE_IN_SECONDS_LSF

public static final double SKIP_SIZE_IN_SECONDS_LSF

See Also:

Constant Field Values

SHARPEN_FORMANTS

public static final boolean SHARPEN_FORMANTS

See Also:

Constant Field Values

FORMANT_SHARPENING_START_FREQ

public static final double FORMANT_SHARPENING_START_FREQ

See Also:

Constant Field Values

FORMANT_SHARPENING_END_FREQ

public static final double FORMANT_SHARPENING_END_FREQ

See Also:

Constant Field Values

RELATIVE_DECREASE_IN_LSF_PAIR_SEPARATION

public static final double RELATIVE_DECREASE_IN_LSF_PAIR_SEPARATION

See Also:

Constant Field Values

MAX_LSF_PAIR_SEPARATION_IN_HZ

public static final double MAX_LSF_PAIR_SEPARATION_IN_HZ

See Also:

Constant Field Values

WINDOW_SIZE_IN_SECONDS_GAIN

public static final double WINDOW_SIZE_IN_SECONDS_GAIN

See Also:

Constant Field Values

SKIP_SIZE_IN_SECONDS_GAIN

public static final double SKIP_SIZE_IN_SECONDS_GAIN

See Also:

Constant Field Values

INCREASE_CONSONANT_GAINS

public static final boolean INCREASE_CONSONANT_GAINS

See Also:

Constant Field Values

CONSONANT_MAX_GAIN_FACTOR

public static final double CONSONANT_MAX_GAIN_FACTOR

See Also:

Constant Field Values

CONSONANT_MAX_GAIN_RELATIVE_DURATION

public static final double CONSONANT_MAX_GAIN_RELATIVE_DURATION

See Also:

Constant Field Values

REDUCE_VOWEL_GAINS

public static final boolean REDUCE_VOWEL_GAINS

See Also:

Constant Field Values

VOWEL_MIN_GAIN_FACTOR

public static final double VOWEL_MIN_GAIN_FACTOR

See Also:

Constant Field Values

VOWEL_MIN_GAIN_RELATIVE_DURATION

public static final double VOWEL_MIN_GAIN_RELATIVE_DURATION

See Also:

Constant Field Values

APPLY_HIGHPASS_FILTER

public static final boolean APPLY_HIGHPASS_FILTER

See Also:

Constant Field Values

HIGHPASS_FILTER_CUTOFF

public static final double HIGHPASS_FILTER_CUTOFF

See Also:

Constant Field Values

HIGHPASS_FILTER_RELATIVE_GAIN

public static final double HIGHPASS_FILTER_RELATIVE_GAIN

See Also:

Constant Field Values

Constructor Detail

Blizzard09PostProcessor

public Blizzard09PostProcessor()

Method Detail

process

public static double[] process(double[] x,
               Labels labels,
               Allophone[] allophones,
               int samplingRateInHz,
               double absMaxOrig)

processGains

public static double[] processGains(double[] x,
                    int samplingRateInHz,
                    Labels labels,
                    boolean[] toBeProcesseds,
                    double extremumGainFactor,
                    double extremumGainRelativeDuration)

processLSFs

public static double[] processLSFs(double[] x,
                   int samplingRateInHz,
                   Labels labels,
                   boolean[] isVowels,
                   boolean[] isPauses)

processHigherFormantGains

public static double[] processHigherFormantGains(double[] x,
                                 int samplingRateInHz,
                                 Labels labels,
                                 boolean[] isPauses)

mainSingleFile

public static void mainSingleFile(String inputWavFile,
                  String outputWavFile,
                  Allophone[] allophones)
                           throws UnsupportedAudioFileException,
                                  IOException

Throws:

UnsupportedAudioFileException

IOException

main

public static void main(String[] args)
                 throws UnsupportedAudioFileException,
                        IOException,
                        MaryConfigurationException

Throws:

UnsupportedAudioFileException

IOException

MaryConfigurationException