marytts.unitselection.analysis

Class VoiceDataDumper

java.lang.Object

marytts.unitselection.analysis.VoiceDataDumper

Direct Known Subclasses:

HnmVoiceDataDumper

public class VoiceDataDumper
extends Object

Convenience class to dump relevant data from a unit selection voice to a Praat TextGrid and a wav file for inspection of timeline data in external tools (e.g. Praat, WaveSurfer, etc.)

Author:

steiner

Field Summary

Fields

Modifier and Type	Field and Description
protected FeatureDefinition	featureDefinition
protected FeatureFileReader	featureFileReader
protected int	halfphoneLRFeatureIndex
protected long	numSamples
protected int	phoneFeatureIndex
protected UnitDatabase	unitDB

Constructor Summary

Constructors

Constructor and Description
VoiceDataDumper()

Method Summary

Methods

Modifier and Type	Method and Description
protected int	byteswap(int val)
protected short	byteswap(short val)
protected void	dumpAudio(String fileName) Adapted from WavWriter.export(String, int, byte[]) and WavWriter.doWrite(String, int)
protected void	dumpData(String voiceName) Get file names from voice config file.
protected void	dumpTextGrid(String fileName) Dump units to Praat TextGrid.
protected double	getDuration(Datagram[] datagrams) Get total duration of a Datagram array
protected byte[]	getSamples(Datagram[] datagrams) Get raw samples from all Datagrams in an array
protected TimelineReader	loadAudioTimeline(String fileName) Load audio timeline from file
protected void	loadFeatureFile(String fileName) Load unit feature file from file
protected void	loadUnitDatabase(String audioTimelineFileName, String basenameTimelineFileName, String unitFileName) Load unit database from various relevant files
static void	main(String[] args) Main method.

Methods inherited from class java.lang.Object

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

Field Detail

unitDB

protected UnitDatabase unitDB

featureFileReader

protected FeatureFileReader featureFileReader

numSamples

protected long numSamples

featureDefinition

protected FeatureDefinition featureDefinition

phoneFeatureIndex

protected int phoneFeatureIndex

halfphoneLRFeatureIndex

protected int halfphoneLRFeatureIndex

Constructor Detail

VoiceDataDumper

public VoiceDataDumper()

Method Detail

byteswap

protected int byteswap(int val)

Parameters:

val - val

Returns:

(((val & 0xff000000) >>> 24) + ((val & 0x00ff0000) >>> 8) + ((val & 0x0000ff00) << 8) + ((val & 0x000000ff) << 24))

See Also:

WavWriter.byteswap(int)

byteswap

protected short byteswap(short val)

Parameters:

val - val

Returns:

((short) ((((int) (val) & 0xff00) >>> 8) + (((int) (val) & 0x00ff) << 8)))

See Also:

WavWriter.byteswap(short)

loadAudioTimeline

protected TimelineReader loadAudioTimeline(String fileName)
                                    throws IOException,
                                           MaryConfigurationException

Load audio timeline from file

Parameters:

fileName - to load

Returns:

TimelineReader

Throws:

IOException - IOException

MaryConfigurationException - MaryConfigurationException

loadUnitDatabase

protected void loadUnitDatabase(String audioTimelineFileName,
                    String basenameTimelineFileName,
                    String unitFileName)
                         throws IOException,
                                MaryConfigurationException

Load unit database from various relevant files

Parameters:

audioTimelineFileName - to load

basenameTimelineFileName - to load

unitFileName - to load

Throws:

IOException - IOException

MaryConfigurationException - MaryConfigurationException

loadFeatureFile

protected void loadFeatureFile(String fileName)
                        throws IOException,
                               MaryConfigurationException

Load unit feature file from file

Parameters:

fileName - to load

Throws:

IOException - IOException

MaryConfigurationException - MaryConfigurationException

getDuration

protected double getDuration(Datagram[] datagrams)

Get total duration of a Datagram array

Parameters:

datagrams - whose duration to get

Returns:

total duration in seconds

getSamples

protected byte[] getSamples(Datagram[] datagrams)
                     throws IOException

Get raw samples from all Datagrams in an array

Parameters:

datagrams - whose samples to get

Returns:

raw samples as stored in the Datagrams

Throws:

IOException - IOException

dumpTextGrid

protected void dumpTextGrid(String fileName)
                     throws IOException

Dump units to Praat TextGrid. This will have three tiers:

1. halfphone units, labeled with unit indices;
2. phone units, labeled with allophones;
3. basenames, labeled with basename of original utterance.

Parameters:

fileName - of new TextGrid

Throws:

IOException - if data files cannot be read, or TextGrid cannot be written

dumpAudio

protected void dumpAudio(String fileName)
                  throws IOException

Adapted from WavWriter.export(String, int, byte[]) and WavWriter.doWrite(String, int)

Parameters:

fileName - fileName

Throws:

IOException - IOException

dumpData

protected void dumpData(String voiceName)
                 throws Exception

Get file names from voice config file. Dump relevant data from audio timeline, unit file, etc. to Praat TextGrid and wav file.

Parameters:

voiceName - for config file to read (e.g. "bits3")

Throws:

Exception - Exception

main

public static void main(String[] args)
                 throws Exception

Main method. Add VOICE jar to classpath, then call with

 -ea -Xmx1gb -Dmary.base=$MARYBASE VOICE
 

or something similar

Parameters:

args - voice name (without the Locale) of voice to dump data from

Throws:

Exception - Exception