ms_gauden.h File Reference

(Sphinx 3.0 specific) Gaussian density module. More...

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

Go to the source code of this file.

Classes
struct	gauden_t
	Multivariate gaussian mixture density parameters. More...
struct	gauden_dist_t
	Structure to store distance (density) values for a given input observation wrt density values in some given codebook. More...
Functions
gauden_t *	gauden_init (const char *meanfile, const char *varfile, float32 varfloor, int32 precompute, logmath_t *logmath)
void	gauden_free (gauden_t *g)
int32	gauden_mean_reload (gauden_t *g, const char *meanfile)
int32	gauden_var_reload (gauden_t *g, const char *meanfile)
int32	gauden_dist_precompute (gauden_t *g, float32 varfloor)
int32	gauden_dist (gauden_t *g, s3mgauid_t mgau, int32 n_top, vector_t *obs, gauden_dist_t **out_dist)
int32	gauden_dist_norm (gauden_t *g, int32 n_top, gauden_dist_t ***dist, uint8 *active)
void	gauden_dump (const gauden_t *g)
void	gauden_dump_ind (const gauden_t *g, int senidx)

---

Detailed Description

(Sphinx 3.0 specific) Gaussian density module.

Gaussian density distribution implementation. There are two major difference bettwen ms_gauden and cont_mgau. One is the fact that ms_gauden only take cares of the Gaussian computation part where cont_mgau actually take care of senone computation as well. The other is the fact that ms_gauden is a multi-stream implementation of GMM computation.

---

Function Documentation

int32 gauden_dist (  gauden_t *  g, s3mgauid_t  mgau, int32  n_top, vector_t *  obs, gauden_dist_t **  out_dist )

Compute gaussian density values for the given input observation vector wrt the specified mixture gaussian codebook (which may consist of several feature streams). Density values are left UNnormalized. Returns: 0 if successful, -1 otherwise. Parameters: g  In: handle to entire ensemble of codebooks mgau  In: codebook for which density values to be evaluated (g->{mean,var}[mgau]) n_top  In: #top densities to be evaluated obs  In: Observation vector; obs[f] = for feature f out_dist  Out: n_top best codewords and density values, in worsening order, for each feature stream. out_dist[f][i] = i-th best density for feature f. Caller must allocate memory for this output

int32 gauden_dist_norm (  gauden_t *  g, int32  n_top, gauden_dist_t ***  dist, uint8 *  active )

Normalize density values (previously computed by gauden_dist). Two cases: If (g->n_mgau == 1), normalize such that the sum of the n_top codeword scores for each feature in dist sums to 1 (in prob domain). Otherwise, normalize by dividing the density value (subtracting, in logprob domain) for each codeword by the best one. Returns: scaling applied to every senone score as a result of the normalization. Parameters: g  In: handle to all collection of codebooks n_top  In: #density values computed per feature dist  In/Out: n_top density indices and values for each feature. On return, density values are normalized. active  In: active[gid] is non-0 iff codebook gid is active. If NULL, all codebooks active

int32 gauden_dist_precompute (  gauden_t *  g, float32  varfloor )

Re-precompute variance related sufficient statistics. You must call this after gauden_var_reload().

void gauden_dump (  const gauden_t *  g  )

Dump the definitionn of Gaussian distribution. Parameters: g  In: Gaussian distribution g

void gauden_dump_ind (  const gauden_t *  g, int  senidx )

Dump the definition of Gaussian distribution of a particular index to the standard output stream Parameters: g  In: Gaussian distribution g senidx  In: The senone index of the Gaussian

void gauden_free (  gauden_t *  g  )

In: The gauden_t to free

gauden_t* gauden_init (  const char *  meanfile, const char *  varfile, float32  varfloor, int32  precompute, logmath_t *  logmath )

Read mixture gaussian codebooks from the given files. Allocate memory space needed for them. Apply the specified variance floor value. Return value: ptr to the model created; NULL if error. (See Sphinx3 model file-format documentation.) Parameters: meanfile  Input: File containing means of mixture gaussians varfile  Input: File containing variances of mixture gaussians varfloor  Input: Floor value to be applied to variances precompute  Input: Whether we should precompute

int32 gauden_mean_reload (  gauden_t *  g, const char *  meanfile )

Reload mixture Gaussian means from the given file. The means must have already been loaded at least once (using gauden_init). Returns: 0 if successful, -1 otherwise. Parameters: g  In/Out: g->mean to be reloaded meanfile  In: File to reload means from

int32 gauden_var_reload (  gauden_t *  g, const char *  meanfile )

Reload mixture Gaussian variances from the given file. The variances must have already been loaded at least once (using gauden_init). Precomputation is not done. Returns: 0 if successful, -1 otherwise. Parameters: g  In/Out: g->mean to be reloaded meanfile  In: File to reload means from

---