Alexandre Lision | 7c6f4a6 | 2013-09-05 13:27:01 -0400 | [diff] [blame] | 1 | /* |
| 2 | ** Copyright (C) 2002-2011 Erik de Castro Lopo <erikd@mega-nerd.com> |
| 3 | ** |
| 4 | ** This program is free software; you can redistribute it and/or modify |
| 5 | ** it under the terms of the GNU Lesser General Public License as published by |
| 6 | ** the Free Software Foundation; either version 2.1 of the License, or |
| 7 | ** (at your option) any later version. |
| 8 | ** |
| 9 | ** This program is distributed in the hope that it will be useful, |
| 10 | ** but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 11 | ** MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 12 | ** GNU Lesser General Public License for more details. |
| 13 | ** |
| 14 | ** You should have received a copy of the GNU Lesser General Public License |
| 15 | ** along with this program; if not, write to the Free Software |
| 16 | ** Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. |
| 17 | */ |
| 18 | |
| 19 | #include "sfconfig.h" |
| 20 | |
| 21 | #include <stdio.h> |
| 22 | #include <fcntl.h> |
| 23 | #include <string.h> |
| 24 | #include <ctype.h> |
| 25 | |
| 26 | #include "sndfile.h" |
| 27 | #include "sfendian.h" |
| 28 | #include "common.h" |
| 29 | |
| 30 | /*------------------------------------------------------------------------------ |
| 31 | ** Macros to handle big/little endian issues. |
| 32 | */ |
| 33 | |
| 34 | #define SFE_HTK_BAD_FILE_LEN 1666 |
| 35 | #define SFE_HTK_NOT_WAVEFORM 1667 |
| 36 | |
| 37 | /*------------------------------------------------------------------------------ |
| 38 | ** Private static functions. |
| 39 | */ |
| 40 | |
| 41 | static int htk_close (SF_PRIVATE *psf) ; |
| 42 | |
| 43 | static int htk_write_header (SF_PRIVATE *psf, int calc_length) ; |
| 44 | static int htk_read_header (SF_PRIVATE *psf) ; |
| 45 | |
| 46 | /*------------------------------------------------------------------------------ |
| 47 | ** Public function. |
| 48 | */ |
| 49 | |
| 50 | int |
| 51 | htk_open (SF_PRIVATE *psf) |
| 52 | { int subformat ; |
| 53 | int error = 0 ; |
| 54 | |
| 55 | if (psf->is_pipe) |
| 56 | return SFE_HTK_NO_PIPE ; |
| 57 | |
| 58 | if (psf->file.mode == SFM_READ || (psf->file.mode == SFM_RDWR && psf->filelength > 0)) |
| 59 | { if ((error = htk_read_header (psf))) |
| 60 | return error ; |
| 61 | } ; |
| 62 | |
| 63 | subformat = SF_CODEC (psf->sf.format) ; |
| 64 | |
| 65 | if (psf->file.mode == SFM_WRITE || psf->file.mode == SFM_RDWR) |
| 66 | { if ((SF_CONTAINER (psf->sf.format)) != SF_FORMAT_HTK) |
| 67 | return SFE_BAD_OPEN_FORMAT ; |
| 68 | |
| 69 | psf->endian = SF_ENDIAN_BIG ; |
| 70 | |
| 71 | if (htk_write_header (psf, SF_FALSE)) |
| 72 | return psf->error ; |
| 73 | |
| 74 | psf->write_header = htk_write_header ; |
| 75 | } ; |
| 76 | |
| 77 | psf->container_close = htk_close ; |
| 78 | |
| 79 | psf->blockwidth = psf->bytewidth * psf->sf.channels ; |
| 80 | |
| 81 | switch (subformat) |
| 82 | { case SF_FORMAT_PCM_16 : /* 16-bit linear PCM. */ |
| 83 | error = pcm_init (psf) ; |
| 84 | break ; |
| 85 | |
| 86 | default : break ; |
| 87 | } ; |
| 88 | |
| 89 | return error ; |
| 90 | } /* htk_open */ |
| 91 | |
| 92 | /*------------------------------------------------------------------------------ |
| 93 | */ |
| 94 | |
| 95 | static int |
| 96 | htk_close (SF_PRIVATE *psf) |
| 97 | { |
| 98 | if (psf->file.mode == SFM_WRITE || psf->file.mode == SFM_RDWR) |
| 99 | htk_write_header (psf, SF_TRUE) ; |
| 100 | |
| 101 | return 0 ; |
| 102 | } /* htk_close */ |
| 103 | |
| 104 | static int |
| 105 | htk_write_header (SF_PRIVATE *psf, int calc_length) |
| 106 | { sf_count_t current ; |
| 107 | int sample_count, sample_period ; |
| 108 | |
| 109 | current = psf_ftell (psf) ; |
| 110 | |
| 111 | if (calc_length) |
| 112 | psf->filelength = psf_get_filelen (psf) ; |
| 113 | |
| 114 | /* Reset the current header length to zero. */ |
| 115 | psf->header [0] = 0 ; |
| 116 | psf->headindex = 0 ; |
| 117 | psf_fseek (psf, 0, SEEK_SET) ; |
| 118 | |
| 119 | if (psf->filelength > 12) |
| 120 | sample_count = (psf->filelength - 12) / 2 ; |
| 121 | else |
| 122 | sample_count = 0 ; |
| 123 | |
| 124 | sample_period = 10000000 / psf->sf.samplerate ; |
| 125 | |
| 126 | psf_binheader_writef (psf, "E444", sample_count, sample_period, 0x20000) ; |
| 127 | |
| 128 | /* Header construction complete so write it out. */ |
| 129 | psf_fwrite (psf->header, psf->headindex, 1, psf) ; |
| 130 | |
| 131 | if (psf->error) |
| 132 | return psf->error ; |
| 133 | |
| 134 | psf->dataoffset = psf->headindex ; |
| 135 | |
| 136 | if (current > 0) |
| 137 | psf_fseek (psf, current, SEEK_SET) ; |
| 138 | |
| 139 | return psf->error ; |
| 140 | } /* htk_write_header */ |
| 141 | |
| 142 | /* |
| 143 | ** Found the following info in a comment block within Bill Schottstaedt's |
| 144 | ** sndlib library. |
| 145 | ** |
| 146 | ** HTK format files consist of a contiguous sequence of samples preceded by a |
| 147 | ** header. Each sample is a vector of either 2-byte integers or 4-byte floats. |
| 148 | ** 2-byte integers are used for compressed forms as described below and for |
| 149 | ** vector quantised data as described later in section 5.11. HTK format data |
| 150 | ** files can also be used to store speech waveforms as described in section 5.8. |
| 151 | ** |
| 152 | ** The HTK file format header is 12 bytes long and contains the following data |
| 153 | ** nSamples -- number of samples in file (4-byte integer) |
| 154 | ** sampPeriod -- sample period in 100ns units (4-byte integer) |
| 155 | ** sampSize -- number of bytes per sample (2-byte integer) |
| 156 | ** parmKind -- a code indicating the sample kind (2-byte integer) |
| 157 | ** |
| 158 | ** The parameter kind consists of a 6 bit code representing the basic |
| 159 | ** parameter kind plus additional bits for each of the possible qualifiers. |
| 160 | ** The basic parameter kind codes are |
| 161 | ** |
| 162 | ** 0 WAVEFORM sampled waveform |
| 163 | ** 1 LPC linear prediction filter coefficients |
| 164 | ** 2 LPREFC linear prediction reflection coefficients |
| 165 | ** 3 LPCEPSTRA LPC cepstral coefficients |
| 166 | ** 4 LPDELCEP LPC cepstra plus delta coefficients |
| 167 | ** 5 IREFC LPC reflection coef in 16 bit integer format |
| 168 | ** 6 MFCC mel-frequency cepstral coefficients |
| 169 | ** 7 FBANK log mel-filter bank channel outputs |
| 170 | ** 8 MELSPEC linear mel-filter bank channel outputs |
| 171 | ** 9 USER user defined sample kind |
| 172 | ** 10 DISCRETE vector quantised data |
| 173 | ** |
| 174 | ** and the bit-encoding for the qualifiers (in octal) is |
| 175 | ** _E 000100 has energy |
| 176 | ** _N 000200 absolute energy suppressed |
| 177 | ** _D 000400 has delta coefficients |
| 178 | ** _A 001000 has acceleration coefficients |
| 179 | ** _C 002000 is compressed |
| 180 | ** _Z 004000 has zero mean static coef. |
| 181 | ** _K 010000 has CRC checksum |
| 182 | ** _O 020000 has 0'th cepstral coef. |
| 183 | */ |
| 184 | |
| 185 | static int |
| 186 | htk_read_header (SF_PRIVATE *psf) |
| 187 | { int sample_count, sample_period, marker ; |
| 188 | |
| 189 | psf_binheader_readf (psf, "pE444", 0, &sample_count, &sample_period, &marker) ; |
| 190 | |
| 191 | if (2 * sample_count + 12 != psf->filelength) |
| 192 | return SFE_HTK_BAD_FILE_LEN ; |
| 193 | |
| 194 | if (marker != 0x20000) |
| 195 | return SFE_HTK_NOT_WAVEFORM ; |
| 196 | |
| 197 | psf->sf.channels = 1 ; |
| 198 | |
| 199 | if (sample_period > 0) |
| 200 | { psf->sf.samplerate = 10000000 / sample_period ; |
| 201 | psf_log_printf (psf, "HTK Waveform file\n Sample Count : %d\n Sample Period : %d => %d Hz\n", |
| 202 | sample_count, sample_period, psf->sf.samplerate) ; |
| 203 | } |
| 204 | else |
| 205 | { psf->sf.samplerate = 16000 ; |
| 206 | psf_log_printf (psf, "HTK Waveform file\n Sample Count : %d\n Sample Period : %d (should be > 0) => Guessed sample rate %d Hz\n", |
| 207 | sample_count, sample_period, psf->sf.samplerate) ; |
| 208 | } ; |
| 209 | |
| 210 | psf->sf.format = SF_FORMAT_HTK | SF_FORMAT_PCM_16 ; |
| 211 | psf->bytewidth = 2 ; |
| 212 | |
| 213 | /* HTK always has a 12 byte header. */ |
| 214 | psf->dataoffset = 12 ; |
| 215 | psf->endian = SF_ENDIAN_BIG ; |
| 216 | |
| 217 | psf->datalength = psf->filelength - psf->dataoffset ; |
| 218 | |
| 219 | psf->blockwidth = psf->sf.channels * psf->bytewidth ; |
| 220 | |
| 221 | if (! psf->sf.frames && psf->blockwidth) |
| 222 | psf->sf.frames = (psf->filelength - psf->dataoffset) / psf->blockwidth ; |
| 223 | |
| 224 | return 0 ; |
| 225 | } /* htk_read_header */ |
| 226 | |