src/htk.c

   1 /* libSoX file format: HTK   (c) 2008 robs@users.sourceforge.net
   2  *
   3  * See http://labrosa.ee.columbia.edu/doc/HTKBook21/HTKBook.html
   4  *
   5  * This library is free software; you can redistribute it and/or modify it
   6  * under the terms of the GNU Lesser General Public License as published by
   7  * the Free Software Foundation; either version 2.1 of the License, or (at
   8  * your option) any later version.
   9  *
  10  * This library is distributed in the hope that it will be useful, but
  11  * WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU Lesser
  13  * General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU Lesser General Public License
  16  * along with this library; if not, write to the Free Software Foundation,
  17  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301  USA
  18  */
  19
  20 #include "sox_i.h"
  21
  22 typedef enum {
  23   Waveform, Lpc, Lprefc, Lpcepstra, Lpdelcep, Irefc,
  24   Mfcc, Fbank, Melspec, User, Discrete, Unknown} kind_t;
  25 static char const * const str[] = {
  26   "Sampled waveform", "Linear prediction filter", "Linear prediction",
  27   "LPC cepstral", "LPC cepstra plus delta", "LPC reflection coef in",
  28   "Mel-frequency cepstral", "Log mel-filter bank", "Linear mel-filter bank",
  29   "User defined sample", "Vector quantised data", "Unknown"};
  30
  31 static int start_read(sox_format_t * ft)
  32 {
  33   uint32_t period_100ns, num_samples;
  34   uint16_t bytes_per_sample, parmKind;
  35
  36   if (lsx_readdw(ft, &num_samples     ) ||
  37       lsx_readdw(ft, &period_100ns    ) ||
  38       lsx_readw (ft, &bytes_per_sample) ||
  39       lsx_readw (ft, &parmKind        )) return SOX_EOF;
  40   if (parmKind != Waveform) {
  41     int n = min(parmKind & 077, Unknown);
  42     lsx_fail_errno(ft, SOX_EFMT, "unsupported HTK type `%s' (0%o)", str[n], parmKind);
  43     return SOX_EOF;
  44   }
  45   return lsx_check_read_params(ft, 1, 1e7 / period_100ns, SOX_ENCODING_SIGN2,
  46       (unsigned)bytes_per_sample << 3, (uint64_t)num_samples, sox_true);
  47 }
  48
  49 static int write_header(sox_format_t * ft)
  50 {
  51   double period_100ns = 1e7 / ft->signal.rate;
  52   uint64_t len = ft->olength? ft->olength:ft->signal.length;
  53
  54   if (len > UINT_MAX)
  55   {
  56     lsx_warn("length greater than 32 bits - cannot fit actual length in header");
  57     len = UINT_MAX;
  58   }
  59   if (!ft->olength && floor(period_100ns) != period_100ns)
  60     lsx_warn("rounding sample period %f (x 100ns) to nearest integer", period_100ns);
  61   return lsx_writedw(ft, (unsigned)len)
  62       || lsx_writedw(ft, (unsigned)(period_100ns + .5))
  63       || lsx_writew(ft, ft->encoding.bits_per_sample >> 3)
  64       || lsx_writew(ft, Waveform) ? SOX_EOF : SOX_SUCCESS;
  65 }
  66
  67 LSX_FORMAT_HANDLER(htk)
  68 {
  69   static char const * const names[] = {"htk", NULL};
  70   static unsigned const write_encodings[] = {SOX_ENCODING_SIGN2, 16, 0, 0};
  71   static sox_format_handler_t handler = {
  72     SOX_LIB_VERSION_CODE,
  73     "PCM format used for Hidden Markov Model speech processing",
  74     names, SOX_FILE_BIG_END | SOX_FILE_MONO | SOX_FILE_REWIND,
  75     start_read, lsx_rawread, NULL,
  76     write_header, lsx_rawwrite, NULL,
  77     lsx_rawseek, write_encodings, NULL, 0
  78   };
  79   return &handler;
  80 }