s3_align.h File Reference

data structure for alignment More...

#include <logmath.h>
#include <s3types.h>

Go to the source code of this file.

Classes
struct	align_stseg_s
struct	align_phseg_s
struct	align_wdseg_s

Typedefs
typedef struct align_stseg_s	align_stseg_t
typedef struct align_phseg_s	align_phseg_t
typedef struct align_wdseg_s	align_wdseg_t

Functions
int32	align_init (mdef_t *_mdef, tmat_t *_tmat, dict_t *_dict, cmd_ln_t *_config, logmath_t *_logmath)
void	align_free (void)
int32	align_build_sent_hmm (char *transcript, int insert_sil)
int32	align_destroy_sent_hmm (void)
int32	align_start_utt (char *uttid)
void	align_sen_active (uint8 *senlist, int32 n_sen)
int32	align_frame (int32 *senscr)
int32	align_end_utt (align_stseg_t **stseg, align_phseg_t **phseg, align_wdseg_t **wdseg)

Detailed Description

data structure for alignment

Typedef Documentation

align_phseg_t

typedef struct align_phseg_s align_phseg_t

Phone level segmentation/alignment information

align_stseg_t

typedef struct align_stseg_s align_stseg_t

State level segmentation/alignment; one entry per frame

align_wdseg_t

typedef struct align_wdseg_s align_wdseg_t

Word level segmentation/alignment information

Function Documentation

align_build_sent_hmm()

int32 align_build_sent_hmm	(	char *	wordstr,
		int	insert_sil
	)

Build a sentence HMM for the given transcription (wordstr). A two-level DAG is built: phone-level and state-level.

• <s> and </s> always added at the beginning and end of sentence to form an augmented transcription.
• Optional <sil> and noise words added between words in the augmented transcription. wordstr must contain only the transcript; no extraneous stuff such as utterance-id. Phone-level HMM structure has replicated nodes to allow for different left and right context CI phones; hence, each pnode corresponds to a unique triphone in the sentence HMM. Return 0 if successful, <0 if any error (eg, OOV word encountered).

Parameters

wordstr	In: Word transcript
insert_sil	In: Whether to insert silences/fillers

References slink_s::node.

align_destroy_sent_hmm()

int32 align_destroy_sent_hmm

(

void

)

align_end_utt()

int32 align_end_utt	(	align_stseg_t **	stseg_out,
		align_phseg_t **	phseg_out,
		align_wdseg_t **	wdseg_out
	)

All frames consumed. Trace back best Viterbi state sequence and dump it out.

Parameters

stseg_out	Out: list of state segmentation
phseg_out	Out: list of phone segmentation
wdseg_out	Out: list of word segmentation

align_frame()

int32 align_frame

(

int32 *

senscr

)

Step time aligner one frame forward
Wind up utterance and return final result (READ-ONLY). Results only valid until the next utterance is begun.

One frame of Viterbi time alignment.

Parameters

senscr

In: array of senone scores this frame

align_free()

void align_free

(

void

)

align_init()

int32 align_init	(	mdef_t *	_mdef,
		tmat_t *	_tmat,
		dict_t *	_dict,
		cmd_ln_t *	_config,
		logmath_t *	_logmath
	)

align_sen_active()

void align_sen_active	(	uint8 *	senlist,
		int32	n_sen
	)

Called at the beginning of a frame to flag the active senones (any senone used by active HMMs) in that frame.

Flag the active senones.

Parameters

senlist	Out: senlist[s] TRUE iff active in frame
n_sen	In: Size of senlist[] array

align_start_utt()

int32 align_start_utt

(

char *

uttid

)

Start Viterbi alignment using the sentence HMM previously built. Assumes that each utterance will only be aligned once; state member variables initialized during sentence HMM building.