segSpData.c File Reference

Functions used for the input/output of data. More...

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "SEGRec.h"
#include "segStdLib.h"
#include "segV3Compat.h"

Defines
#define	isSWAP   1
	Read feature data will be byte-swapped if 1.
Functions
int	segCheckMasterList (SEG_UTT_LIST *uL, int detail)
	Display details of the utterance list for debugging.
int	segTime2FrameNum (SEG_UTT_LIST *uL, int winSize, int numVecs)
	Convert annotation times.
int	segFreeBreakPoints (int uttVecs, int **breakPt)
	Free memory allocated to break points.
int	segFreeMasterList (int totLabs, SEG_UTT_LIST *masterList)
	Free memory space allocated to the master utterance list.
int	segReadSpData (double ***spData, FILE *curSpFile, int numVecs, int vecSize)
	Read in the acoustic feature vectors.
int	segFreeSpData (int numVecs, double ***spData)
	Free memory space allocated to the current speech data.
int	segCheckSpData (int sT, int eT, int vecSize, double **spData)
	Display the speech vectors for debugging.
int	segCountLabels (FILE *labFile, int numVecs, int winSize)
	Count the number of labels in an annotation file.
int	segJoinLabels (int *startT, int *endT, char ***labList, FILE *curLabFile, int nLabs)
	Extract overall start and end times of the utterance and loads label names.
int	segGetBreakPoints (int **breakPt, int startT, int endT, FILE *curLabFile, int emb, int winSize, int nLabs, int verbose)
	Identify the frames that correspond to the start of a new utterance.
int	segWriteMLFData (SEG_ANN_DATA *recOp, int nLabs, int winSize, FILE *mlfFile, char *labFileName)
	Generate and write new annotations from the classification/recognition output.

---

Detailed Description

Functions used for the input/output of data.

Date:

05.Jul.2007

Author:

Yoshinori Shiga @ CVSSP, University of Surrey

Note:

This code was originally created on the basis of Martin Russell & P Jackson's SEGVit Ver.3.4.

Revision

---

Function Documentation

int segCheckMasterList	(	SEG_UTT_LIST *	uL,
		int	detail
	)

Display details of the utterance list for debugging.

The index of the utterance, its start time, end time and the number of labels in the utterance are displayed on the screen.

Parameters:

[in]	uL	Utterance list
[in]	detail	Level of debugging

Return values:

NO_ERROR

Successful execution

Author:

Yoshinori Shiga @ CVSSP, University of Surrey

int segCheckSpData	(	int	sT,
		int	eT,
		int	vecSize,
		double **	spData
	)

Display the speech vectors for debugging.

Parameters:

[in]	sT	Start time of the utterance
[in]	eT	End time of the utterance
[in]	vecSize	Length of acoustic feature vector in bytes
[in]	spData	Acoustic feature vectors of speech

Return values:

NO_ERROR

Successful execution

Author:

Yoshinori Shiga @ CVSSP, University of Surrey

int segCountLabels	(	FILE *	labFile,
		int	numVecs,
		int	winSize
	)

Count the number of labels in an annotation file.

This function goes through the file for the first time to count the number of labels.

Parameters:

[in]	labFile	File pointer to the annotation file
[in]	numVecs	Number of frames
[in]	winSize	Frame shift size

Returns:

Number of labels in the utterance

Author:

Yoshinori Shiga @ CVSSP, University of Surrey

int segFreeBreakPoints	(	int	uttVecs,
		int **	breakPt
	)

Free memory allocated to break points.

Parameters:

[in]	uttVecs	Number of utterances
[in]	breakPt	Pointer to the break point array

Return values:

NO_ERROR

Successful execution

Author:

Yoshinori Shiga @ CVSSP, University of Surrey

int segFreeMasterList	(	int	totLabs,
		SEG_UTT_LIST *	masterList
	)

Free memory space allocated to the master utterance list.

Parameters:

[in]	totLabs	Number of labels in the utterance
[in]	masterList	Master utterance list

Return values:

NO_ERROR

Successful execution

Author:

Yoshinori Shiga @ CVSSP, University of Surrey

int segFreeSpData	(	int	numVecs,
		double ***	spData
	)

Free memory space allocated to the current speech data.

The memory allocated to the speech data is freed for all frames.

Parameters:

[in]	numVecs	Number of frames in the utterance
[in]	spData	Pointer to memory storing speech data

Return values:

NO_ERROR

Author:

Yoshinori Shiga @ CVSSP, University of Surrey

int segGetBreakPoints	(	int **	breakPt,
		int	startT,
		int	endT,
		FILE *	curLabFile,
		int	emb,
		int	winSize,
		int	nLabs,
		int	verbose
	)

Identify the frames that correspond to the start of a new utterance.

This function is used to create an array of break points for all frames in the utterance.

Parameters:

[in,out]	breakPt	Array for the break points
[in]	startT	Start time of the utterance
[in]	endT	End time of the utterance
[in]	curLabFile	File pointer to the current label file
[in]	emb	Training type (supervised/embedded)
[in]	winSize	Frame shift size
[in]	nLabs	Number of labels int the utterance
[in]	verbose	Level of reporting during execution

Return values:

	RTN_ERROR	Memory allocation error / File read error
	otherwise	Number of elements in the break point array

Author:

Yoshinori Shiga @ CVSSP, University of Surrey

Note:

It will be the start of each label for phone-level processing, for example.

A 'breakPoint' is the start time of a TIMIT label. This function creates an int array breakPt such that breakPt[t] = 1 if t is a breakPoint and 0 otherwise.

int segJoinLabels	(	int *	startT,
		int *	endT,
		char ***	labList,
		FILE *	curLabFile,
		int	nLabs
	)

Extract overall start and end times of the utterance and loads label names.

This function goes through the label file for the second time and process the labels.

Parameters:

[in,out]	startT	Start time of the utterance
[in,out]	endT	End time of the utterance
[in,out]	labList	Pointer to the label name list
[in]	curLabFile	File pointer to the current label file
[in]	nLabs	Number of labels in the utterance

Return values:

	RTN_ERROR	Memory allocation error / File read error
	otherwise	Index of the last label in the file

Author:

Yoshinori Shiga @ CVSSP, University of Surrey

int segReadSpData	(	double ***	spData,
		FILE *	curSpFile,
		int	numVecs,
		int	vecSize
	)

Read in the acoustic feature vectors.

Parameters:

[in,out]	spData	Pointer to a 2D array used to store the vectors
[in,out]	curSpFile	File pointer to the speech file
[in]	numVecs	Number of frames in the utterance
[in]	vecSize	Sample size

Return values:

	RTN_ERROR	Memory allocation error
	totRead	Total number of items in the speech file that are read

Author:

Yoshinori Shiga @ CVSSP, University of Surrey

Note:

This function returns a two dimensional array spData, where spData[t][f] contains coefficient number f of frame number t. 0 <= t < numVecs, 0 <= f < numCoefs

Function segReadAVector() is called internally.

See also:

segReadAVector()

int segTime2FrameNum	(	SEG_UTT_LIST *	uL,
		int	winSize,
		int	numVecs
	)

Convert annotation times.

Convert the start and end times in nano seconds to appropriate frame indices.

Parameters:

[in,out]	uL	Utterance list
[in]	winSize	Size of window (ms) used in processing speech data
[in]	numVecs	Number of frames in the utterance

Return values:

NO_ERROR

Successful execution

Author:

Yoshinori Shiga @ CVSSP, University of Surrey

Note:

This function converts annotation times from 100ns units (as used by HTK) to frame indeces that (rather quirkily) start from one, and therefore lie in the range of 1 to numVecs, which is the number of frames in the current speechfile.

int segWriteMLFData	(	SEG_ANN_DATA *	recOp,
		int	nLabs,
		int	winSize,
		FILE *	mlfFile,
		char *	labFileName
	)

Generate and write new annotations from the classification/recognition output.

Parameters:

[in]	recOp	Array of struct annData variables
[in]	nLabs	Number of labels in the utterance
[in]	winSize	Size of the window (ms) used in processing speech data
[in]	mlfFile	File pointer to a MLF file
[in]	labFileName	Name of the label file

Return values:

	RTN_ERROR	File write error
	NO_ERROR	Successful execution

Author:

Yoshinori Shiga @ CVSSP, University of Surrey

---