gc_read_1.cc

这是处理语音信号的程序

// file: gc_read_1.cc
// 

// isip include files
//
#include "grammar_compiler.h"
#include "grammar_compiler_constants.h"

// system include files
//
#include <string.h>

// method: preprocessing_grammar_cc
//  
// arguments:
//  char_1** symbols: (input/output) the array of symbols
//  int_4& size_a: (input/output) number of symbols in array
//
// return: a logical_1 indicating status
//
// this method preprocess the grammar from array of symbols
// distribute the probability value from sub-expression to terminals
//

logical_1 preprocessing_grammar_cc(char_1** symbols_a, int_4& size_a) {
  
  int_4 i_pointer = size_a-1; // loop variable begining at the end of grammar
  int_4 i_end=0, i_start=0; //start and end position of matched "()"
  int_4 i_brace_indicate=0; // matched brace indicate  =0 mean matched brace

  // store the probability value for each terminal or sub-expression
  //
  float_8 probability_a[GC_GRAM_MAX_SYMBOL_SIZE];

  // store the boolean for if probability value has been given
  //
  logical_1 prob_bool_a[GC_GRAM_MAX_SYMBOL_SIZE];
  logical_1 alternative_bool=ISIP_FALSE; // default no alternative operation

  // sub total probability for all given probability in a sub-expression  
  //
  float_8 sub_total_prob=0.0;

  // sub total number for all no given probability in a sub-expression
  //
  int_4 sub_total_num_no_prob=0;
  int_4 offset; //offset for add ~ and probability value
  int_4 sub_total=0; //total terminals or sub-expressions in
  float_8 distribution_prob=0.0;
  float_8 target_prob=0.0;

  int_4 i_loop=size_a-1; //loop variable to control whether finished

  // check if the grammar has matched brace
  //
  while(i_pointer > 0)
    {
      if(strcmp((char *)symbols_a[i_pointer-1], (char *)BRACE_START)==0)
	  i_brace_indicate++;
      else if(strcmp((char *)symbols_a[i_pointer-1], (char *)BRACE_CLOSE)==0)
	  i_brace_indicate--;

      i_pointer--;
    }

  if(i_brace_indicate==0)
    {}
  else
    {
      printf("Unmatched brace found!!!\n");
      exit(ISIP_PROTO_ERROR);
    }
  
  i_pointer=size_a-1;
  i_brace_indicate=0;
  
  // Check whether the format in the grammar is correct
  // look for ]~  >~  }~
  // if found 
  // print error message, then exit
  // else 
  // continue
  //
  for (int_4 j = size_a-1; j >0 ; j--) {
    
    if((strcmp((char *)symbols_a[j], (char *)PROBABILITY_SYMBOL)==0
	&& strcmp((char *)symbols_a[j-1],
		  (char *)ONE_OR_MORE_REPEAT_CLOSE)==0)
       ||(strcmp((char *)symbols_a[j], (char *)PROBABILITY_SYMBOL)==0
	  && strcmp((char *)symbols_a[j-1],
		    (char *)ZERO_OR_MORE_REPEAT_CLOSE)==0)
       ||(strcmp((char *)symbols_a[j], (char *)PROBABILITY_SYMBOL)==0
	  && strcmp((char *)symbols_a[j-1], (char *)OPTIONAL_CLOSE)==0)
       )
      {
	printf("Illegal probability assignment after expansion of subexpression:   %s%s\n", symbols_a[j-1], symbols_a[j]);
	exit(ISIP_PROTO_ERROR);
      }
  }
  
  
  // find the ")~" , search from the end of grammar
  //
  while(i_pointer > 0)
    {
      if((strcmp((char *)symbols_a[i_pointer], (char *)PROBABILITY_SYMBOL)==0
	  && strcmp((char *)symbols_a[i_pointer-1], (char *)BRACE_CLOSE)==0) )
	{
	  target_prob=(float_8) atof((char*)symbols_a[i_pointer+1]);
	  i_end=i_pointer-1;
	  i_brace_indicate++;
	  break;
	}
      else
	i_pointer--;
    }

  i_loop=i_end;
  i_brace_indicate=0;

  // i_loop=0 mean no more ")~" combination in grammar
  //
  while(i_loop>0) {
    
    // Check whether the format in the grammar is correct
    // look for ]~  >~  }~
    // if found 
    // print error message, then exit
    // else 
    // continue
    //
    for (int_4 j = size_a-1; j >0 ; j--) {
      
      if((strcmp((char *)symbols_a[j], (char *)PROBABILITY_SYMBOL)==0
	  && strcmp((char *)symbols_a[j-1],
		    (char *)ONE_OR_MORE_REPEAT_CLOSE)==0)
	 ||(strcmp((char *)symbols_a[j], (char *)PROBABILITY_SYMBOL)==0
	    && strcmp((char *)symbols_a[j-1],
		      (char *)ZERO_OR_MORE_REPEAT_CLOSE)==0)
	 ||(strcmp((char *)symbols_a[j], (char *)PROBABILITY_SYMBOL)==0
	    && strcmp((char *)symbols_a[j-1], (char *)OPTIONAL_CLOSE)==0)) {

	printf("Illegal probability assignment after expansion of subexpression:   %s%s\n", symbols_a[j-1], symbols_a[j]);
	exit(ISIP_PROTO_ERROR);
      }
    }
    
    // Apply the "()" sub-expression probability distribution rules
    //  to grammar.
    // Inside the sub-expression, apply alternative probability
    //  distribution rules to grammar if applicable.
    //
    i_pointer=size_a-1;
    
    // get the position of the first appear string ")~"
    // beginning from the end of grammar
    //
    while(i_pointer > 0) {
	  if((strcmp((char *)symbols_a[i_pointer],
		     (char *)PROBABILITY_SYMBOL)==0
	      && strcmp((char *)symbols_a[i_pointer-1],
			(char *)BRACE_CLOSE)==0) ) {
	    target_prob=(float_8) atof((char*)symbols_a[i_pointer+1]);
	    i_end=i_pointer-1;
	    i_brace_indicate++;
	    break;
	  }
	  else
	    i_pointer--;
    }
    
    i_pointer=i_end;
    i_loop=i_pointer;
    
    // initialization probability value and indicate
    //
    for (int_4 j = 0; j < GC_GRAM_MAX_SYMBOL_SIZE; j++) {
      probability_a[j] = 0.0;
      prob_bool_a[j]=ISIP_FALSE;
    }  
    
    // get the position of the string "(" which matched the ')~'
    // at the position i_end of the grammar
    //
    while(i_loop > 0) {
      if(strcmp((char *)symbols_a[i_loop], (char *)BRACE_START)==0) {
	i_brace_indicate--;
	if(i_brace_indicate==0) {
	  i_start=i_loop;
	  break;
	}
      }
      else if(strcmp((char *)symbols_a[i_loop-1], (char *)BRACE_CLOSE)==0) 
	i_brace_indicate++;
      i_loop--;
    } // end of while i_loop>0
    
    i_pointer=i_start+1;
    i_brace_indicate=0;
    sub_total=0;
    sub_total_prob=0.0;
    sub_total_num_no_prob=0;
    
    // find all terminals or sub-expressions in the sub-expression
    // between i_start and i_end
    //
      while(i_pointer < i_end) {
	
	// if it is sub-expression, move all the way
	// till the end of sub-expression
	//
	if(strcmp((char *)symbols_a[i_pointer], (char *)BRACE_START)==0) {
	  i_brace_indicate++;
	  i_pointer++; //move to next symbol
	  
	  while (i_brace_indicate!=0) {
	    if(strcmp((char *)symbols_a[i_pointer], (char *)BRACE_START)==0)
	      i_brace_indicate++;
	    else if(strcmp((char *)symbols_a[i_pointer],
			   (char *)BRACE_CLOSE)==0)
	      i_brace_indicate--;
	    i_pointer++; //move to next symbol
	  } //end of while i_brace_indicate
	  
	  // if it is "~", ignore the probability symbol
	  //
	  if(strcmp((char *)symbols_a[i_pointer],
		    (char *)PROBABILITY_SYMBOL)==0) {
	    prob_bool_a[sub_total]=ISIP_TRUE;
	    i_pointer++;
	    probability_a[sub_total]=
	      (float_8) atof((char*)symbols_a[i_pointer]);
		}
	  else
	    i_pointer--; //push back the symbol
	  
	  sub_total++;
	} //end of if BRACE_START
	
	// if it is alternative symbol, ignore it
	//
	else if(strcmp((char *)symbols_a[i_pointer],
		       (char *)ALTERNATIVE_SYMBOL)==0) {
	  i_pointer++;
	  alternative_bool=ISIP_TRUE;
	  continue;
	}
	
	// if move to here this is terminal
	//
	else {
	  i_pointer++;
	  if(strcmp((char *)symbols_a[i_pointer],
		    (char *)PROBABILITY_SYMBOL)==0) {
	    prob_bool_a[sub_total]=ISIP_TRUE;
	    i_pointer++;
	    probability_a[sub_total]=
	      (float_8) atof((char*)symbols_a[i_pointer]);
	  }
	  else
	    i_pointer--;
	  
	  sub_total++;
	}
	i_pointer++; //move to next symbol
      }
      
      // calculate the total given probability in this sub-expression
      //
      for (int_4 j = 0; j < sub_total; j++) {
	sub_total_prob+=probability_a[j];
	if(prob_bool_a[j]==ISIP_FALSE) sub_total_num_no_prob++;
      }
      
      // offset to add ~ and probability value
      // probability value need to distribute to each no-given terminals
      //
      offset = 2*sub_total_num_no_prob;
      distribution_prob=(1-sub_total_prob)/sub_total_num_no_prob;
      
      if(alternative_bool==ISIP_TRUE) {  
	for (int_4 j = 0; j < sub_total; j++) {
	  if(prob_bool_a[j]==ISIP_FALSE) probability_a[j]=distribution_prob;
	  probability_a[j] *=target_prob;
	}
      }
      else {
	if(prob_bool_a[0]==ISIP_FALSE)   //no probability value
	  probability_a[0]=target_prob;
	else                      //has probability value