mllr.h File Reference

(Sphinx 3.X specific) Single Stream MLLR. More...

#include <s3types.h>
#include "cont_mgau.h"

Go to the source code of this file.

Functions
void	mllr_dump (float32 ***A, float32 **B, float32 **H, int32 veclen, int32 nclass)
int32	mllr_read_regmat (const char *regmatfile, float32 ****A, float32 ***B, float32 ***H, int32 *nclass, int32 ceplen)
int32	mllr_free_regmat (float32 ***A, float32 **B, float32 **H)
int32	mllr_norm_mgau (mgau_model_t *mgauset, float32 ***A, float32 **B, float32 **H, int32 nclass, int32 *cb2mllr)

---

Detailed Description

(Sphinx 3.X specific) Single Stream MLLR.

---

Function Documentation

void mllr_dump (  float32 ***  A, float32 **  B, float32 **  H, int32  veclen, int32  nclass )

Dump the regression matrix from a given file Parameters: A  In: [nclass][streamlen][streamlen] B  In: [nclass][streamlen] H  In: [nclass][streamlen] veclen  In: vector length nclass  In: Number of classes

int32 mllr_free_regmat (  float32 ***  A, float32 **  B, float32 **  H )

Free a regression matrix previously read in by mllr_read_regmat. Returns: 0 if successful, -1 otherwise. Parameters: A  In: A[streamlen][streamlen] B  In: B[streamlen] H  In: H[streamlen]

int32 mllr_norm_mgau (  mgau_model_t *  mgauset, float32 ***  A, float32 **  B, float32 **  H, int32  nclass, int32 *  cb2mllr )

Transform a mixture density mean matrix according to the given regression matrix. Returns: 0 if successful, -1 otherwise. Parameters: mgauset  In/Out: The gaussian distribution needs to be transformed A  In: "matrix" portion of regression matrix B  In: "vector" portion of regression matrix H  In: diagonal "matrix" for variance adaptation nclass  In: number of classes cb2mllr  In: class to senone mapping

int32 mllr_read_regmat (  const char *  regmatfile, float32 ****  A, float32 ***  B, float32 ***  H, int32 *  nclass, int32  ceplen )

In: vector length Parameters: regmatfile  In: File to be read A  Out: [*A][nclass][streamlen][streamlen] B  Out: [*B][nclass][streamlen] H  Out: [*H][nclass][streamlen] nclass  Out: number of classes

---