libTLK  1.3.1
Data Structures | Defines | Functions
Initialization from transcriptions of HMMs using linear segmentation
Algorithms

Data Structures

struct  tLHMMInitializer
 HMM initializer. More...

Defines

#define tL_hmm_initializer_free(INIT)   (INIT)->meths->free ( INIT )
 Frees memory.
#define tL_init_hmm(INIT, COUNTS, C, FEA)   tL_init_hmm_seg ( INIT, COUNTS, C, FEA, 0, (FEA)->cols )
 Initializes an HMM using linear segmentation of sample.

Functions

 tL_hmm_initializer_new (const tLAModel *amodel, char **err)
 Gets a HMM initializer.
 tL_init_hmm_seg (tLHMMInitializer *init, tLAMiCounts *counts, const int c, const tLFea *fea, const int begin, const int nframes)
 Initializes an HMM using linear segmentation of a sample segment.

Define Documentation

#define tL_hmm_initializer_free (   INIT)    (INIT)->meths->free ( INIT )

Frees memory.

Frees the memory allocated for the HMM initializer.

Parameters:
INITThe HMM initializer.
#define tL_init_hmm (   INIT,
  COUNTS,
  C,
  FEA 
)    tL_init_hmm_seg ( INIT, COUNTS, C, FEA, 0, (FEA)->cols )

Initializes an HMM using linear segmentation of sample.

This macro is equivalent to tL_init_hmm_seg but using all frames contained in fea.

Parameters:
INITThe HMM initializer.
COUNTSInteger counts related to AMODEL used by INIT.
CThe index of the HMM to be initialized.
FEAA feature vector manager, which must contain a sequence of feature vectors of an adequate type for the AMODEL.

Function Documentation

tL_hmm_initializer_new ( const tLAModel amodel,
char **  err 
)

Gets a HMM initializer.

This function creates a new HMM initializer used to initialize HMMs using linear segmentation. Currently supported types are: Bernoulli and DGaussian.

Parameters:
amodelThe acoustic model.
errPointer to string variable. If not NULL, an error message is allocated in the variable in case of error.
Returns:
HMM initializer, or NULL in case of error.
tL_init_hmm_seg ( tLHMMInitializer *  init,
tLAMiCounts counts,
const int  c,
const tLFea fea,
const int  begin,
const int  nframes 
)

Initializes an HMM using linear segmentation of a sample segment.

This function linearly segments the input frames, which are a segment of the given feature vector sequence, according to the number of states of the specified HMM. According to this segmentation, input frames are assigned to states, and then, state counts are consequently increased.

Parameters:
initHMM initializer.
countsInteger counts related to amodel used by the HMM initializer.
cThe index of the HMM to be initialized.
feaA feature vector manager, which must contain a sequence of feature vectors of an adequate type for the amodel.
beginThe first input frame.
nframesThe number of input frames.
 All Data Structures Variables