adin_file.c

about sound recognition.i want to downlod

/**
 * @file   adin_file.c
 * @author Akinobu LEE
 * @date   Sun Feb 13 13:31:20 2005
 *
 * <JA>
 * @brief  ファイル掐蜗¨WAV/RAWファイルおよび筛洁掐蜗からの粕み哈み
 *
 * 不兰ファイルからの掐蜗を乖なう簇眶ですˉサポ〖トしているファイル妨及は
 * Microsoft WAVE妨及の不兰ファイル·およびヘッダ痰し∈RAW∷ファイルですˉ
 * オ〖ディオ妨及は·痰暗教PCM·16bit·モノラルに嘎られていますˉ
 * RAWファイルの眷圭·デ〖タのバイトオ〖ダ〖が big endian であることを
 * 涟捏としていますので·庙罢してくださいˉ
 * 
 * ファイルのサンプリングレ〖トはシステムの妥滇するサンプリングレ〖ト
 * ∈adin_standby() で回年される猛∷と办米する涩妥がありますˉ
 * WAVE妨及のファイルの眷圭は·
 * ファイルのサンプリングレ〖トがこの回年猛と办米しなければエラ〖となりますˉ
 * RAWファイル掐蜗の眷圭は·ファイルにヘッダ攫鼠が痰く峡不箕の
 * サンプリングレ〖トが稍汤なため·チェック痰しでファイルの
 * サンプリングレ〖トが adin_standby() で回年された猛である
 * と簿年して借妄されますˉ
 *
 * 掐蜗ファイル叹は·筛洁掐蜗から粕み哈まれますˉ
 * ファイル叹を误刁したファイルリストファイルが回年された眷圭·
 * そのファイルから掐蜗ファイル叹が界肌粕み哈まれますˉ
 *
 * libsndfile を蝗脱する眷圭·adin_sndfile.c 柒の簇眶が蝗脱されますˉ
 * この眷圭·このファイルの簇眶は蝗脱されませんˉ
 *
 * このファイル柒では int を 4byte, short を 2byte と簿年していますˉ
 * </JA>
 * <EN>
 * @brief  Audio input from file or stdin
 *
 * Functions to get input from wave file or standard input.
 * Two file formats are supported, Microsoft WAVE format and RAW (no header)
 * format.  The audio format should be uncompressed PCM, 16bit, monoral.
 * On RAW file input, the data byte order must be in big endian.
 *
 * The sampling rate of input file must be equal to the system requirement
 * value which is specified by adin_standby() . For WAVE format file,
 * the sampling rate of the input file described in its header is checked
 * against the system value, and rejected if not matched.  But for RAW format
 * file, no check will be applied since it has no header information about
 * the recording sampling rate, so be careful of the sampling rate setting.
 *
 * When file input mode, the file name will be read from standard input.
 * If a filelist file is specified, the file names are read from the file
 * sequencially instead.
 *
 * When compiled with libsndfile support, the functions in adin_sndfile.c
 * is used for file input instead of functions below.
 *
 * In this file, assume sizeof(int)=4, sizeof(short)=2
 * </EN>
 *
 * $Revision: 1.3 $
 * 
 */
/*
 * Copyright (c) 1991-2006 Kawahara Lab., Kyoto University
 * Copyright (c) 2000-2005 Shikano Lab., Nara Institute of Science and Technology
 * Copyright (c) 2005-2006 Julius project team, Nagoya Institute of Technology
 * All rights reserved
 */


#include <sent/stddefs.h>
#include <sent/speech.h>
#include <sent/adin.h>

static FILE *gfp;		///< File pointer of current input file
static boolean wav_p;		///< TRUE if input is WAVE file, FALSE if RAW file
static int maxlen;		///< Number of samples, described in the header of WAVE file
static int nowlen;		///< Current number of read samples
/**
 * When file input, the first 4 bytes of the file are read at first to
 * identify whether it is WAVE file format.  This work area is used to
 * keep back the 4 bytes if the input is actually a RAW format.
 */
static SP16 pre_data[2];
static boolean has_pre;		///< TRUE if pre_data is available

static unsigned int sfreq;	///< Sampling frequency in Hz, specified by adin_standby()

/* read .wav data with endian conversion */
/* (all .wav datas are in little endian) */

/** 
 * Read a header value from WAVE file with endian conversion.
 * If required number of data has not been read, it produces error.
 * 
 * @param buf [out] Pointer to store read data
 * @param unitbyte [in] Unit size
 * @param unitnum [in] Required number of units to read
 * @param fp [in] File pointer
 * 
 * @return TRUE on success, FALSE if required number of data has not been read.
 */
static boolean
myread(void *buf, size_t unitbyte, int unitnum, FILE *fp)
{
  int tmp;
  if ((tmp = myfread(buf, unitbyte, unitnum, fp)) < unitnum) {
    return(FALSE);
  }
#ifdef WORDS_BIGENDIAN
  swap_bytes(buf, unitbyte, unitnum);
#endif
  return(TRUE);
}
/// Abbreviation for header reading
#define MYREAD(A,B,C,D)  if (!myread(A, B, C, D)) {j_printerr("adin_file: file is corrupted\n"); return -1;}

/** 
 * @brief  Parse header part of a WAVE file to prepare for data reading
 *
 * The audio format will be checked here, and data length is also read from
 * the header.  Then the pointer is moved to the start point of data part.
 *
 * When called, the file pointer should be located just after 
 * the first 4 bytes, "RIFF".  It also sets @a maxlen and @a nowlen .
 * 
 * @param fp [in] File pointer
 * 
 * @return TRUE if check successfully passed, FALSE if an error occured.
 */
static boolean
setup_wav(FILE *fp)
{
  char dummy[9];
  unsigned int i, len;
  unsigned short s;

  /* 4 byte: byte num of rest ( = filesize - 8) */
  /* --- just skip them */
  MYREAD(dummy, 1, 4, fp);
  /* first part: WAVE format specifications */
  /* 4 byte: "WAVE" */
  MYREAD(dummy, 1, 4, fp);
  if (dummy[0] != 'W' ||
      dummy[1] != 'A' ||
      dummy[2] != 'V' ||
      dummy[3] != 'E') {
    j_printerr("adin_file: WAVE header not found, file corrupted?\n");
    fclose_readfile(fp); return FALSE;
  }
  /* format chunk: "fmt " */
  MYREAD(dummy, 1, 4, fp);
  if (dummy[0] != 'f' ||
      dummy[1] != 'm' ||
      dummy[2] != 't' ||
      dummy[3] != ' ') {
    j_printerr("adin_file: fmt chunk not found, file corrupted?\n");
    fclose_readfile(fp); return FALSE;
  }
  /* 4byte: byte size of this part */
  MYREAD(&len, 4, 1, fp);

  /* 2byte: data format */
  MYREAD(&s, 2, 1, fp);
  if (s != 1) {
    j_printerr("adin_file: data format != PCM (id=%d)\n", s);
    fclose_readfile(fp); return FALSE;
  }
  /* 2byte: channel num */
  MYREAD(&s, 2, 1, fp);
  if (s >= 2) {
    j_printerr("adin_file: channel num != 1 (%d)\n", s);
    fclose_readfile(fp); return FALSE;
  }
  /* 4byte: sampling rate */
  MYREAD(&i, 4, 1, fp);
  if (i != sfreq) {
    j_printerr("adin_file: sampling rate != %d (%d)\n", sfreq, i);
    fclose_readfile(fp); return FALSE;
  }
  /* 4byte: bytes per second */
  MYREAD(&i, 4, 1, fp);
  if (i != sfreq * sizeof(SP16)) {
    j_printerr("adin_file: bytes per second != %d (%d)\n", sfreq * sizeof(SP16), i);
    fclose_readfile(fp); return FALSE;
  }
  /* 2bytes: bytes per frame ( = (bytes per sample) x channel ) */
  MYREAD(&s, 2, 1, fp);
  if (s != 2) {
    j_printerr("adin_file: (bytes per sample) x channel != 2 (%d)\n", s);
    fclose_readfile(fp); return FALSE;
  }
  /* 2bytes: bits per sample */
  MYREAD(&s, 2, 1, fp);
  if (s != 16) {
    j_printerr("adin_file: bits per sample != 16 (%d)\n", s);
    fclose_readfile(fp); return FALSE;
  }
  /* skip rest */
  if (len > 16) {
    len -= 16;
    while (len > 0) {
      if (len > 8) {
	MYREAD(dummy, 1, 8, fp);
	len -= 8;
      } else {
	MYREAD(dummy, 1, len, fp);
	len = 0;
      }
    }
  }
  /* end of fmt part */

  /* seek for 'data' part */
  while (myread(dummy, 1, 4, fp)) {
    MYREAD(&len, 4, 1, fp);
    if (dummy[0] == 'd' &&
	dummy[1] == 'a' &&
	dummy[2] == 't' &&
	dummy[3] == 'a') {
      break;
    }
    for (i=0;i<len;i++) myread(dummy, 1, 1, fp);
  }
  /* ready to read in "data" part --- this is speech data */
  maxlen = len / sizeof(SP16);
  nowlen = 0;
  return TRUE;
}

/** 
 * @brief  Open input file
 *
 * Open the file, check the file format, and set the file pointer to @a gfp .
 * 
 * @param filename [in] file name, or NULL for standard input
 * 
 * @return TRUE on success, FALSE on failure.
 */
static boolean
adin_file_open(char *filename)	/* NULL for standard input */
{
  FILE *fp;
  char dummy[4];

  if (filename != NULL) {
    if ((fp = fopen_readfile(filename)) == NULL) {
      j_printerr("failed to open %s\n",filename);
      return(FALSE);
    }
  } else {
    fp = stdin;
  }
  /* check first 4 byte to detect Microsoft WAVE format */
  if (myfread(dummy, 1, 4, fp) < 4) {
    j_printerr("Error: size less than 4 bytes?\n",filename);
    fclose_readfile(fp);
    return(FALSE);
  }
  if (dummy[0] == 'R' &&
      dummy[1] == 'I' &&
      dummy[2] == 'F' &&
      dummy[3] == 'F') {
    /* it's a WAVE file */
    wav_p = TRUE;
    has_pre = FALSE;
    if (setup_wav(fp) == FALSE) {
      j_printerr("Error: failed to read %s as a wav file\n",filename);
      fclose_readfile(fp);
      return(FALSE);
    }
  } else {
    /* read as raw format file */
    wav_p = FALSE;
    memcpy(pre_data, dummy, 4);    /* already read (4/sizeof(SP)) samples */
    has_pre = TRUE;
  }

  gfp = fp;

  return(TRUE);
}