/* syllable.c
   Simple Mandarin tone recognizer: syllable routines

   Jim Gilsinan IV
   gilsinan@post.harvard.edu
   http://www.fas.harvard.edu/~gilsinan/

   Senior Thesis
   Harvard University
   2000-2001

*/

#include <stdio.h>
#include <stdlib.h>

#include "tone.h"
#include "syllable.h"

// detect the number of syllables in a pitch track file,
// given pointer to file, address to store count
// returns array of offsets to syllables

int * detect_syllables(FILE * in, int * syllable_count)
{

  int count = 0,
      current_offset = 0,
      i;
  double current_pitch = 0;
  char line[MAXLINE];
  int * offsets;

  rewind(in);

  // first count syllables

  while(!feof(in)) {

    // skip leading zeroes

    while(!current_pitch) {
      fgets(line, MAXLINE, in);
      sscanf(line, "%lf", &current_pitch);
      if (feof(in)) break;
    }

    count++;
    
    while(current_pitch) {
      fgets(line, MAXLINE, in);
      sscanf(line, "%lf", &current_pitch); 
    }
  }

  // now get offsets

  count--;
  rewind(in);

  offsets = malloc(count * sizeof(int));

  for(i = 0; i < count; i++) {

    // skip leading zeroes

    while(!current_pitch) {
      fgets(line, MAXLINE, in);
      sscanf(line, "%lf", &current_pitch);
      if (feof(in)) break;
      current_offset++;
    }

    offsets[i] = current_offset;
    
    while(current_pitch) {
      fgets(line, MAXLINE, in);
      sscanf(line, "%lf", &current_pitch); 
      current_offset++;
    }
  }

  *syllable_count = count;
  return offsets;
} 

// jump to the specified line in the pitch track file

void syllable_seek (FILE * in, int offset)
{
  int i;
  char line[MAXLINE];

  rewind(in);

  for (i = 0; i < offset; i++)
    fgets(line, MAXLINE, in);

  return;
}

// get the derivative of a syllable pitch track,
// given pitch track array, current length, address to store
// derivative syllable

void syllable_derivative (double * in, int len, double * out)
{
  int i;

  for (i = 0; i < len - 1; i++) {
    out[i] = in[i+1] - in[i];
  }

  out[len-1] = 0;

  return;
}

// make the syllable the standard length,
// and normalize by dividing by standard deviation
// given pitch track array, current length, address to store
// standardized syllable

void standardize_syllable (double * in, int len, double * out)
{
  int i, int_part;
  double position, f_part;

  // standardize in time

  for (i = 0; i < STANDARD_LEN; i++) {
    position = ((double)i/STANDARD_LEN)*len;
	int_part = (int)position;
	f_part = position - int_part;
    if (int_part + 1 == len)
	  out[i] = in[int_part];
	else
	  out[i] = in[int_part] * (1-f_part) +
	  	   in[int_part + 1] * (f_part);
  }
  
/*
  // normalize in space

  s = 0;

  for (i = 0; i < STANDARD_LEN; i++)
    // first get std deviation
	s += (fabsf(out[i]) * fabsf(out[i]));

  s /= STANDARD_LEN;

  for ( i = 0; i < STANDARD_LEN; i++)
  	// divide by std dev
	out[i] /= s;
*/

  return;
}

// normalize a syllable by subtracting the mean
// pitch from each sample in the pitch track

// arguments: pointer to pitch track array, length of array

void normalize_syllable (double * in, int len)
{
  double sample_mean = get_mean_pitch(in, len);
  int i;

  for (i = 0; i < len; i++)
    in[i] -= sample_mean;

  return;
}

// get the mean pitch of a syllable

// arguments: pointer to pitch track array, length of array

double get_mean_pitch ( double * in, int len )
{
  int i;
  double mean = 0;

  for ( i = 0; i < len; i++ )
    mean += in[i];

  mean /= len;

  return ( mean );
}

// read a syllable from the specified file

// arguments: file pointer, pointer in which to store read length,
// syllable start offset

// returns pitch track array

double * get_syllable (FILE * in, int * len, int syllable_offset)
{
  int i,
      sample_count = 0;
  double * out;
  double current_pitch = 1;
  char line[MAXLINE];

  // seek to syllable

  syllable_seek(in, syllable_offset); 

  // first get number of samples in pitch track

  while ( current_pitch ) {
    sample_count++;
    fgets(line, MAXLINE, in);
    sscanf(line, "%lf", &current_pitch);
  }

  sample_count--;

  // allocate an array for 'em

  out = malloc(sample_count * sizeof(double));

  // go back to the beginning of syllable
  // and actually read the pitches this time

  syllable_seek(in, syllable_offset);

  for (i = 0; i < sample_count; i++) {
    fgets(line, MAXLINE, in);
    sscanf(line, "%lf", &out[i]);
  }

  *len = sample_count;

  return (out);
}