• Main Page
  • Related Pages
  • Modules
  • Data Structures
  • Files
  • Examples
  • File List
  • Globals

libavcodec/truespeech.c

Go to the documentation of this file.
00001 /*
00002  * DSP Group TrueSpeech compatible decoder
00003  * Copyright (c) 2005 Konstantin Shishkov
00004  *
00005  * This file is part of FFmpeg.
00006  *
00007  * FFmpeg is free software; you can redistribute it and/or
00008  * modify it under the terms of the GNU Lesser General Public
00009  * License as published by the Free Software Foundation; either
00010  * version 2.1 of the License, or (at your option) any later version.
00011  *
00012  * FFmpeg is distributed in the hope that it will be useful,
00013  * but WITHOUT ANY WARRANTY; without even the implied warranty of
00014  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
00015  * Lesser General Public License for more details.
00016  *
00017  * You should have received a copy of the GNU Lesser General Public
00018  * License along with FFmpeg; if not, write to the Free Software
00019  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
00020  */
00021 
00022 #include "libavutil/intreadwrite.h"
00023 #include "avcodec.h"
00024 
00025 #include "truespeech_data.h"
00034 typedef struct {
00035     /* input data */
00036     int16_t vector[8];  //< input vector: 5/5/4/4/4/3/3/3
00037     int offset1[2];     //< 8-bit value, used in one copying offset
00038     int offset2[4];     //< 7-bit value, encodes offsets for copying and for two-point filter
00039     int pulseoff[4];    //< 4-bit offset of pulse values block
00040     int pulsepos[4];    //< 27-bit variable, encodes 7 pulse positions
00041     int pulseval[4];    //< 7x2-bit pulse values
00042     int flag;           //< 1-bit flag, shows how to choose filters
00043     /* temporary data */
00044     int filtbuf[146];   // some big vector used for storing filters
00045     int prevfilt[8];    // filter from previous frame
00046     int16_t tmp1[8];    // coefficients for adding to out
00047     int16_t tmp2[8];    // coefficients for adding to out
00048     int16_t tmp3[8];    // coefficients for adding to out
00049     int16_t cvector[8]; // correlated input vector
00050     int filtval;        // gain value for one function
00051     int16_t newvec[60]; // tmp vector
00052     int16_t filters[32]; // filters for every subframe
00053 } TSContext;
00054 
00055 static av_cold int truespeech_decode_init(AVCodecContext * avctx)
00056 {
00057 //    TSContext *c = avctx->priv_data;
00058 
00059     if (avctx->channels != 1) {
00060         av_log_ask_for_sample(avctx, "Unsupported channel count: %d\n", avctx->channels);
00061         return AVERROR(EINVAL);
00062     }
00063 
00064     avctx->sample_fmt = AV_SAMPLE_FMT_S16;
00065     return 0;
00066 }
00067 
00068 static void truespeech_read_frame(TSContext *dec, const uint8_t *input)
00069 {
00070     uint32_t t;
00071 
00072     /* first dword */
00073     t = AV_RL32(input);
00074     input += 4;
00075 
00076     dec->flag = t & 1;
00077 
00078     dec->vector[0] = ts_codebook[0][(t >>  1) & 0x1F];
00079     dec->vector[1] = ts_codebook[1][(t >>  6) & 0x1F];
00080     dec->vector[2] = ts_codebook[2][(t >> 11) &  0xF];
00081     dec->vector[3] = ts_codebook[3][(t >> 15) &  0xF];
00082     dec->vector[4] = ts_codebook[4][(t >> 19) &  0xF];
00083     dec->vector[5] = ts_codebook[5][(t >> 23) &  0x7];
00084     dec->vector[6] = ts_codebook[6][(t >> 26) &  0x7];
00085     dec->vector[7] = ts_codebook[7][(t >> 29) &  0x7];
00086 
00087     /* second dword */
00088     t = AV_RL32(input);
00089     input += 4;
00090 
00091     dec->offset2[0] = (t >>  0) & 0x7F;
00092     dec->offset2[1] = (t >>  7) & 0x7F;
00093     dec->offset2[2] = (t >> 14) & 0x7F;
00094     dec->offset2[3] = (t >> 21) & 0x7F;
00095 
00096     dec->offset1[0] = ((t >> 28) & 0xF) << 4;
00097 
00098     /* third dword */
00099     t = AV_RL32(input);
00100     input += 4;
00101 
00102     dec->pulseval[0] = (t >>  0) & 0x3FFF;
00103     dec->pulseval[1] = (t >> 14) & 0x3FFF;
00104 
00105     dec->offset1[1] = (t >> 28) & 0x0F;
00106 
00107     /* fourth dword */
00108     t = AV_RL32(input);
00109     input += 4;
00110 
00111     dec->pulseval[2] = (t >>  0) & 0x3FFF;
00112     dec->pulseval[3] = (t >> 14) & 0x3FFF;
00113 
00114     dec->offset1[1] |= ((t >> 28) & 0x0F) << 4;
00115 
00116     /* fifth dword */
00117     t = AV_RL32(input);
00118     input += 4;
00119 
00120     dec->pulsepos[0] = (t >> 4) & 0x7FFFFFF;
00121 
00122     dec->pulseoff[0] = (t >> 0) & 0xF;
00123 
00124     dec->offset1[0] |= (t >> 31) & 1;
00125 
00126     /* sixth dword */
00127     t = AV_RL32(input);
00128     input += 4;
00129 
00130     dec->pulsepos[1] = (t >> 4) & 0x7FFFFFF;
00131 
00132     dec->pulseoff[1] = (t >> 0) & 0xF;
00133 
00134     dec->offset1[0] |= ((t >> 31) & 1) << 1;
00135 
00136     /* seventh dword */
00137     t = AV_RL32(input);
00138     input += 4;
00139 
00140     dec->pulsepos[2] = (t >> 4) & 0x7FFFFFF;
00141 
00142     dec->pulseoff[2] = (t >> 0) & 0xF;
00143 
00144     dec->offset1[0] |= ((t >> 31) & 1) << 2;
00145 
00146     /* eighth dword */
00147     t = AV_RL32(input);
00148     input += 4;
00149 
00150     dec->pulsepos[3] = (t >> 4) & 0x7FFFFFF;
00151 
00152     dec->pulseoff[3] = (t >> 0) & 0xF;
00153 
00154     dec->offset1[0] |= ((t >> 31) & 1) << 3;
00155 
00156 }
00157 
00158 static void truespeech_correlate_filter(TSContext *dec)
00159 {
00160     int16_t tmp[8];
00161     int i, j;
00162 
00163     for(i = 0; i < 8; i++){
00164         if(i > 0){
00165             memcpy(tmp, dec->cvector, i * 2);
00166             for(j = 0; j < i; j++)
00167                 dec->cvector[j] = ((tmp[i - j - 1] * dec->vector[i]) +
00168                                    (dec->cvector[j] << 15) + 0x4000) >> 15;
00169         }
00170         dec->cvector[i] = (8 - dec->vector[i]) >> 3;
00171     }
00172     for(i = 0; i < 8; i++)
00173         dec->cvector[i] = (dec->cvector[i] * ts_decay_994_1000[i]) >> 15;
00174 
00175     dec->filtval = dec->vector[0];
00176 }
00177 
00178 static void truespeech_filters_merge(TSContext *dec)
00179 {
00180     int i;
00181 
00182     if(!dec->flag){
00183         for(i = 0; i < 8; i++){
00184             dec->filters[i + 0] = dec->prevfilt[i];
00185             dec->filters[i + 8] = dec->prevfilt[i];
00186         }
00187     }else{
00188         for(i = 0; i < 8; i++){
00189             dec->filters[i + 0]=(dec->cvector[i] * 21846 + dec->prevfilt[i] * 10923 + 16384) >> 15;
00190             dec->filters[i + 8]=(dec->cvector[i] * 10923 + dec->prevfilt[i] * 21846 + 16384) >> 15;
00191         }
00192     }
00193     for(i = 0; i < 8; i++){
00194         dec->filters[i + 16] = dec->cvector[i];
00195         dec->filters[i + 24] = dec->cvector[i];
00196     }
00197 }
00198 
00199 static void truespeech_apply_twopoint_filter(TSContext *dec, int quart)
00200 {
00201     int16_t tmp[146 + 60], *ptr0, *ptr1;
00202     const int16_t *filter;
00203     int i, t, off;
00204 
00205     t = dec->offset2[quart];
00206     if(t == 127){
00207         memset(dec->newvec, 0, 60 * 2);
00208         return;
00209     }
00210     for(i = 0; i < 146; i++)
00211         tmp[i] = dec->filtbuf[i];
00212     off = (t / 25) + dec->offset1[quart >> 1] + 18;
00213     ptr0 = tmp + 145 - off;
00214     ptr1 = tmp + 146;
00215     filter = (const int16_t*)ts_order2_coeffs + (t % 25) * 2;
00216     for(i = 0; i < 60; i++){
00217         t = (ptr0[0] * filter[0] + ptr0[1] * filter[1] + 0x2000) >> 14;
00218         ptr0++;
00219         dec->newvec[i] = t;
00220         ptr1[i] = t;
00221     }
00222 }
00223 
00224 static void truespeech_place_pulses(TSContext *dec, int16_t *out, int quart)
00225 {
00226     int16_t tmp[7];
00227     int i, j, t;
00228     const int16_t *ptr1;
00229     int16_t *ptr2;
00230     int coef;
00231 
00232     memset(out, 0, 60 * 2);
00233     for(i = 0; i < 7; i++) {
00234         t = dec->pulseval[quart] & 3;
00235         dec->pulseval[quart] >>= 2;
00236         tmp[6 - i] = ts_pulse_scales[dec->pulseoff[quart] * 4 + t];
00237     }
00238 
00239     coef = dec->pulsepos[quart] >> 15;
00240     ptr1 = (const int16_t*)ts_pulse_values + 30;
00241     ptr2 = tmp;
00242     for(i = 0, j = 3; (i < 30) && (j > 0); i++){
00243         t = *ptr1++;
00244         if(coef >= t)
00245             coef -= t;
00246         else{
00247             out[i] = *ptr2++;
00248             ptr1 += 30;
00249             j--;
00250         }
00251     }
00252     coef = dec->pulsepos[quart] & 0x7FFF;
00253     ptr1 = (const int16_t*)ts_pulse_values;
00254     for(i = 30, j = 4; (i < 60) && (j > 0); i++){
00255         t = *ptr1++;
00256         if(coef >= t)
00257             coef -= t;
00258         else{
00259             out[i] = *ptr2++;
00260             ptr1 += 30;
00261             j--;
00262         }
00263     }
00264 
00265 }
00266 
00267 static void truespeech_update_filters(TSContext *dec, int16_t *out, int quart)
00268 {
00269     int i;
00270 
00271     for(i = 0; i < 86; i++)
00272         dec->filtbuf[i] = dec->filtbuf[i + 60];
00273     for(i = 0; i < 60; i++){
00274         dec->filtbuf[i + 86] = out[i] + dec->newvec[i] - (dec->newvec[i] >> 3);
00275         out[i] += dec->newvec[i];
00276     }
00277 }
00278 
00279 static void truespeech_synth(TSContext *dec, int16_t *out, int quart)
00280 {
00281     int i,k;
00282     int t[8];
00283     int16_t *ptr0, *ptr1;
00284 
00285     ptr0 = dec->tmp1;
00286     ptr1 = dec->filters + quart * 8;
00287     for(i = 0; i < 60; i++){
00288         int sum = 0;
00289         for(k = 0; k < 8; k++)
00290             sum += ptr0[k] * ptr1[k];
00291         sum = (sum + (out[i] << 12) + 0x800) >> 12;
00292         out[i] = av_clip(sum, -0x7FFE, 0x7FFE);
00293         for(k = 7; k > 0; k--)
00294             ptr0[k] = ptr0[k - 1];
00295         ptr0[0] = out[i];
00296     }
00297 
00298     for(i = 0; i < 8; i++)
00299         t[i] = (ts_decay_35_64[i] * ptr1[i]) >> 15;
00300 
00301     ptr0 = dec->tmp2;
00302     for(i = 0; i < 60; i++){
00303         int sum = 0;
00304         for(k = 0; k < 8; k++)
00305             sum += ptr0[k] * t[k];
00306         for(k = 7; k > 0; k--)
00307             ptr0[k] = ptr0[k - 1];
00308         ptr0[0] = out[i];
00309         out[i] = ((out[i] << 12) - sum) >> 12;
00310     }
00311 
00312     for(i = 0; i < 8; i++)
00313         t[i] = (ts_decay_3_4[i] * ptr1[i]) >> 15;
00314 
00315     ptr0 = dec->tmp3;
00316     for(i = 0; i < 60; i++){
00317         int sum = out[i] << 12;
00318         for(k = 0; k < 8; k++)
00319             sum += ptr0[k] * t[k];
00320         for(k = 7; k > 0; k--)
00321             ptr0[k] = ptr0[k - 1];
00322         ptr0[0] = av_clip((sum + 0x800) >> 12, -0x7FFE, 0x7FFE);
00323 
00324         sum = ((ptr0[1] * (dec->filtval - (dec->filtval >> 2))) >> 4) + sum;
00325         sum = sum - (sum >> 3);
00326         out[i] = av_clip((sum + 0x800) >> 12, -0x7FFE, 0x7FFE);
00327     }
00328 }
00329 
00330 static void truespeech_save_prevvec(TSContext *c)
00331 {
00332     int i;
00333 
00334     for(i = 0; i < 8; i++)
00335         c->prevfilt[i] = c->cvector[i];
00336 }
00337 
00338 static int truespeech_decode_frame(AVCodecContext *avctx,
00339                 void *data, int *data_size,
00340                 AVPacket *avpkt)
00341 {
00342     const uint8_t *buf = avpkt->data;
00343     int buf_size = avpkt->size;
00344     TSContext *c = avctx->priv_data;
00345 
00346     int i, j;
00347     short *samples = data;
00348     int consumed = 0;
00349     int16_t out_buf[240];
00350     int iterations;
00351 
00352     if (!buf_size)
00353         return 0;
00354 
00355     if (buf_size < 32) {
00356         av_log(avctx, AV_LOG_ERROR,
00357                "Too small input buffer (%d bytes), need at least 32 bytes\n", buf_size);
00358         return -1;
00359     }
00360     iterations = FFMIN(buf_size / 32, *data_size / 480);
00361     for(j = 0; j < iterations; j++) {
00362         truespeech_read_frame(c, buf + consumed);
00363         consumed += 32;
00364 
00365         truespeech_correlate_filter(c);
00366         truespeech_filters_merge(c);
00367 
00368         memset(out_buf, 0, 240 * 2);
00369         for(i = 0; i < 4; i++) {
00370             truespeech_apply_twopoint_filter(c, i);
00371             truespeech_place_pulses(c, out_buf + i * 60, i);
00372             truespeech_update_filters(c, out_buf + i * 60, i);
00373             truespeech_synth(c, out_buf + i * 60, i);
00374         }
00375 
00376         truespeech_save_prevvec(c);
00377 
00378         /* finally output decoded frame */
00379         for(i = 0; i < 240; i++)
00380             *samples++ = out_buf[i];
00381 
00382     }
00383 
00384     *data_size = consumed * 15;
00385 
00386     return consumed;
00387 }
00388 
00389 AVCodec ff_truespeech_decoder = {
00390     "truespeech",
00391     AVMEDIA_TYPE_AUDIO,
00392     CODEC_ID_TRUESPEECH,
00393     sizeof(TSContext),
00394     truespeech_decode_init,
00395     NULL,
00396     NULL,
00397     truespeech_decode_frame,
00398     .long_name = NULL_IF_CONFIG_SMALL("DSP Group TrueSpeech"),
00399 };

Generated on Wed Apr 11 2012 07:31:35 for FFmpeg by  doxygen 1.7.1