classify.c

speech signal process tools

/*----------------------------------------------------------------------+
|									|
|   This material contains proprietary software of Entropic Speech,	|
|   Inc.  Any reproduction, distribution, or publication without the	|
|   prior written permission of Entropic Speech, Inc. is strictly	|
|   prohibited.  Any public distribution of copies of this work		|
|   authorized in writing by Entropic Speech, Inc. must bear the	|
|   notice								|
|									|
|    "Copyright (c) 1987 Entropic Speech, Inc. All rights reserved."	|
|									|
+-----------------------------------------------------------------------+
| 									|
|  Program:	main.c							|
| 									|
|  Written by:  Rodney Johnson.						|
|  Checked by:								|
| 									|
|  Main module of classify.						|
|									|
|  Classifies records in an ESPS feature file by the maximum-likeli-	|
|  hood method, according to information in a statistics file.		|
|									|
+----------------------------------------------------------------------*/
#ifndef lint
static char *sccs_id = "@(#)classify.c	1.1 6/5/87";
#endif
/*
 * system include files
 */
#include <stdio.h>
/*
 * ESPS include files
 */
#include <sps/sps.h>
#include <sps/fea.h>
#include <sps/feastat.h>
/*
 * defines
 */
#define DEBUG(n) if (debug_level >= n) Fprintf
#define Fprintf (void)fprintf
#define PROG Fprintf(stderr, "%s: ", ProgName)
#define EXIT Fprintf(stderr, "\n"); exit(1);
#define ERROR_EXIT(text) {PROG; Fprintf(stderr, (text)); EXIT}
#define ERROR_EXIT1(fmt,a) {PROG; Fprintf(stderr, (fmt), (a)); EXIT}
#define ERROR_EXIT2(fmt,a,b) {PROG; Fprintf(stderr, (fmt), (a), (b)); EXIT}
#define TRYALLOC(type,num,var,msg) { \
    if (((var) = (type *) calloc((unsigned)(num),sizeof(type))) == NULL) \
    ERROR_EXIT1("Can't allocate memory--%s", (msg))}
#define SYNTAX USAGE("classify [-x debug_level][-f field][-e elements]\
[-d][-C field][-L field]\n\
[-h file.his] input.stat input.fea output.fea")
/*
 * system functions and variables
 */
int     getopt();
extern int  optind;
extern char *optarg;
int	atoi();
long	time();
char	*ctime();
int	strcmp();
char    *strcpy();
void    exit();
char    *calloc();
void	perror();
/*
 * external ESPS functions
 */
char	*get_cmd_line();
void	write_header();
struct header
	*read_header();
int	lin_search();
int	add_comment();
char	*fea_decode();
long	*grange_switch();
char	**get_fea_deriv();
long	get_fea_siz();
short	get_fea_type();
char	*get_deriv_vec();
void	copy_fea_rec();
/*
 * global function declarations
 */
void	pr_farray();
int	classify();
int	num_enums();
/*
 * global variable declarations
 */
int     debug_level = 0;       
char    *ProgName = "classify";
char    *Version = "1.1";
char	*Date = "6/5/87";
char	*cmd_line;
/*
 * main program
 */
main(argc, argv)
    int	    argc;
    char    **argv;
{
    char    *statfield;		    /* Field name.  Get features from this */
				    /* field or its source fields. */
    long    feasize;		    /* Size of named field in input. */
    long    statsize;		    /* Field size from statistics file. */
    long    *elements;		    /* Indices specified with -e option. */
    long    num_elems;		    /* Number of indices specified. */
    char    *e_arg;		    /* Argument of -e option. */
    short   d_flag = NO,	    /* -d option specified? */
	    e_flag = NO,	    /* -e option specified? */
	    f_flag = NO;	    /* -f option specified? */

    char    *hist_name = NULL;	    /* History-file name. */
    FILE    *hist_strm;		    /* History file. */
    long    tloc;		    /* Current time. */

    char    *stat_name;		    /* Statistics-file name. */
    FILE    *stat_strm;		    /* Statistics file. */
    struct header   *stat_ih;	    /* Statistics-file header. */
    struct feastat  *stat_rec,	    /* Statistics-file record. */
	    **stat_recs;	    /* All statistics-file records. */
    float   **means;		    /* Mean vectors from statistics file. */
    float   ***invcovars;	    /* Inverse covariance matrices from */
				    /* statistics file. */
    int	    max_clas,		    /* Upper bound on number of classes. */
	    num_clas;		    /* Actual number of classes. */
    char    **fields;		    /* Source-field definitions. */

    char    *infea_name;	    /* Name of input feature file. */
    FILE    *infea_strm;	    /* Input feature file. */
    struct header   *infea_ih;	    /* Input feature-file header. */
    struct fea_data *infea_rec;	    /* Input feature-file record. */
    char    *fea_ptr;		    /* Pointer to data in input record. */
    float   *feavec = NULL;	    /* Features from input record. */
    long    length;		    /* Length of feavec. */

    char    *outfea_name;	    /* Output feature-file name. */
    FILE    *outfea_strm;	    /* Output feature-file.  */
    struct header   *outfea_oh;	    /* Output feature-file header. */
    struct fea_data *outfea_rec;    /* Output feature-file record. */
    char    **enums;		    /* Class names for output header. */
    char    *clasfield = "class";   /* Output field name for class. */
    short   *class;		    /* Pointer to output field for class. */
    char    *postfield = "posteriors";
				    /* Output field name for likelihoods. */
    float   *posteriors;	    /* Output vector for likelihoods. */
    long    *count;		    /* Count records in each class. */
    long    nrec;		    /* Number of records. */

    int	    c;			    /* Input option letter. */
    int	    i, j;		    /* Loop indices. */

    cmd_line = get_cmd_line(argc, argv);
/*
 * process command line options
 */
    while ((c = getopt(argc, argv, "x:f:e:dC:L:h:")) != EOF)
    {
	switch (c)
	{
	case 'x': 
	    debug_level = atoi(optarg);
	    break;
	case 'f':
	    f_flag = YES;
	    statfield = optarg;
	    break;
	case 'e':
	    e_flag = YES;
	    elements = grange_switch(e_arg = optarg, &num_elems);
	    break;
	case 'd':
	    d_flag = YES;
	    break;
	case 'C':
	    clasfield = optarg;
	    break;
	case 'L':
	    postfield = optarg;
	    break;
	case 'h':
	    hist_name = optarg;
	    break;
	default:
	    SYNTAX
	    break;
	}
    }
    if (d_flag && (e_flag || f_flag))
	ERROR_EXIT("The -d option is incompatible with -e and -f.")
/*
 * process file arguments
 */
    if (optind != argc - 3) SYNTAX

    stat_name = argv[optind++];
    if (strcmp(stat_name, "-") == 0)
    {
	stat_name = "<stdin>";
	stat_strm = stdin;
    }
    else
	TRYOPEN(ProgName, stat_name, "r", stat_strm);
    if ((stat_ih = read_header(stat_strm)) == NULL)
	NOTSPS(ProgName, stat_name)
    if (stat_ih->common.type != FT_FEA
	    || stat_ih->hd.fea->fea_type != FEA_STAT)
	ERROR_EXIT1("%s is not an ESPS statistics file.", stat_name)

    infea_name = argv[optind++];
    if (strcmp(infea_name, "-") == 0)
    {
	if (stat_strm == stdin)
	    ERROR_EXIT("Input files can't both be standard input.")
	infea_name = "<stdin>";
	infea_strm = stdin;
    }
    else
	TRYOPEN(ProgName, infea_name, "r", infea_strm);
    if ((infea_ih = read_header(infea_strm)) == NULL)
	NOTSPS(ProgName, infea_name)
    if (infea_ih->common.type != FT_FEA)
	ERROR_EXIT1("%s is not an ESPS feature file.", infea_name)

    outfea_name = argv[optind++];
    if (strcmp(outfea_name, "-") == 0)
    {
	outfea_name = "<stdout>";
	outfea_strm = stdout;
    }
    else
	TRYOPEN(ProgName, outfea_name, "w", outfea_strm);
/*
 * open optional history file and initial output
 */
    if (hist_name != NULL)
    {
	TRYOPEN(ProgName, hist_name, "w", hist_strm);
	tloc = time((long *) NULL);
	Fprintf(hist_strm, "Classify history output on %s", ctime(&tloc));
	Fprintf(hist_strm, "Classify version %s of %s\n", Version, Date);
	Fprintf(hist_strm, "Command line:\n%s", cmd_line);
    }
/*
 * read statistics records
 */
    if (get_fea_siz("invcovar", stat_ih, (short *) NULL, (long **) NULL)
	<= 0) ERROR_EXIT(
	    "Statistics file does not have inverse covariance matrices.")
    max_clas = num_enums("class", stat_ih);
    TRYALLOC(struct feastat *, max_clas, stat_recs, 
	    "statistics record pointers.")
    stat_rec = allo_feastat_rec(stat_ih);
    num_clas = 0;
    while (get_feastat_rec(stat_rec, stat_ih, stat_strm) != EOF
	    && num_clas < max_clas)
    {
	stat_recs[num_clas++] = stat_rec;
	stat_rec = allo_feastat_rec(stat_ih);
    }
    if (num_clas == 0) ERROR_EXIT("Empty statistics file.")

    DEBUG(1)(stderr, "Read %d statistics records.\n", num_clas);

    for (i = 0; i < num_clas; i++)