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

libavcodec/mpegaudioenc.c

Go to the documentation of this file.
00001 /*
00002  * The simplest mpeg audio layer 2 encoder
00003  * Copyright (c) 2000, 2001 Fabrice Bellard
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 
00027 #include "avcodec.h"
00028 #include "bitstream.h"
00029 
00030 #undef  CONFIG_MPEGAUDIO_HP
00031 #define CONFIG_MPEGAUDIO_HP 0
00032 #include "mpegaudio.h"
00033 
00034 /* currently, cannot change these constants (need to modify
00035    quantization stage) */
00036 #define MUL(a,b) (((int64_t)(a) * (int64_t)(b)) >> FRAC_BITS)
00037 
00038 #define SAMPLES_BUF_SIZE 4096
00039 
00040 typedef struct MpegAudioContext {
00041     PutBitContext pb;
00042     int nb_channels;
00043     int freq, bit_rate;
00044     int lsf;           /* 1 if mpeg2 low bitrate selected */
00045     int bitrate_index; /* bit rate */
00046     int freq_index;
00047     int frame_size; /* frame size, in bits, without padding */
00048     int64_t nb_samples; /* total number of samples encoded */
00049     /* padding computation */
00050     int frame_frac, frame_frac_incr, do_padding;
00051     short samples_buf[MPA_MAX_CHANNELS][SAMPLES_BUF_SIZE]; /* buffer for filter */
00052     int samples_offset[MPA_MAX_CHANNELS];       /* offset in samples_buf */
00053     int sb_samples[MPA_MAX_CHANNELS][3][12][SBLIMIT];
00054     unsigned char scale_factors[MPA_MAX_CHANNELS][SBLIMIT][3]; /* scale factors */
00055     /* code to group 3 scale factors */
00056     unsigned char scale_code[MPA_MAX_CHANNELS][SBLIMIT];
00057     int sblimit; /* number of used subbands */
00058     const unsigned char *alloc_table;
00059 } MpegAudioContext;
00060 
00061 /* define it to use floats in quantization (I don't like floats !) */
00062 //#define USE_FLOATS
00063 
00064 #include "mpegaudiodata.h"
00065 #include "mpegaudiotab.h"
00066 
00067 static av_cold int MPA_encode_init(AVCodecContext *avctx)
00068 {
00069     MpegAudioContext *s = avctx->priv_data;
00070     int freq = avctx->sample_rate;
00071     int bitrate = avctx->bit_rate;
00072     int channels = avctx->channels;
00073     int i, v, table;
00074     float a;
00075 
00076     if (channels <= 0 || channels > 2){
00077         av_log(avctx, AV_LOG_ERROR, "encoding %d channel(s) is not allowed in mp2\n", channels);
00078         return -1;
00079     }
00080     bitrate = bitrate / 1000;
00081     s->nb_channels = channels;
00082     s->freq = freq;
00083     s->bit_rate = bitrate * 1000;
00084     avctx->frame_size = MPA_FRAME_SIZE;
00085 
00086     /* encoding freq */
00087     s->lsf = 0;
00088     for(i=0;i<3;i++) {
00089         if (ff_mpa_freq_tab[i] == freq)
00090             break;
00091         if ((ff_mpa_freq_tab[i] / 2) == freq) {
00092             s->lsf = 1;
00093             break;
00094         }
00095     }
00096     if (i == 3){
00097         av_log(avctx, AV_LOG_ERROR, "Sampling rate %d is not allowed in mp2\n", freq);
00098         return -1;
00099     }
00100     s->freq_index = i;
00101 
00102     /* encoding bitrate & frequency */
00103     for(i=0;i<15;i++) {
00104         if (ff_mpa_bitrate_tab[s->lsf][1][i] == bitrate)
00105             break;
00106     }
00107     if (i == 15){
00108         av_log(avctx, AV_LOG_ERROR, "bitrate %d is not allowed in mp2\n", bitrate);
00109         return -1;
00110     }
00111     s->bitrate_index = i;
00112 
00113     /* compute total header size & pad bit */
00114 
00115     a = (float)(bitrate * 1000 * MPA_FRAME_SIZE) / (freq * 8.0);
00116     s->frame_size = ((int)a) * 8;
00117 
00118     /* frame fractional size to compute padding */
00119     s->frame_frac = 0;
00120     s->frame_frac_incr = (int)((a - floor(a)) * 65536.0);
00121 
00122     /* select the right allocation table */
00123     table = ff_mpa_l2_select_table(bitrate, s->nb_channels, freq, s->lsf);
00124 
00125     /* number of used subbands */
00126     s->sblimit = ff_mpa_sblimit_table[table];
00127     s->alloc_table = ff_mpa_alloc_tables[table];
00128 
00129 #ifdef DEBUG
00130     av_log(avctx, AV_LOG_DEBUG, "%d kb/s, %d Hz, frame_size=%d bits, table=%d, padincr=%x\n",
00131            bitrate, freq, s->frame_size, table, s->frame_frac_incr);
00132 #endif
00133 
00134     for(i=0;i<s->nb_channels;i++)
00135         s->samples_offset[i] = 0;
00136 
00137     for(i=0;i<257;i++) {
00138         int v;
00139         v = ff_mpa_enwindow[i];
00140 #if WFRAC_BITS != 16
00141         v = (v + (1 << (16 - WFRAC_BITS - 1))) >> (16 - WFRAC_BITS);
00142 #endif
00143         filter_bank[i] = v;
00144         if ((i & 63) != 0)
00145             v = -v;
00146         if (i != 0)
00147             filter_bank[512 - i] = v;
00148     }
00149 
00150     for(i=0;i<64;i++) {
00151         v = (int)(pow(2.0, (3 - i) / 3.0) * (1 << 20));
00152         if (v <= 0)
00153             v = 1;
00154         scale_factor_table[i] = v;
00155 #ifdef USE_FLOATS
00156         scale_factor_inv_table[i] = pow(2.0, -(3 - i) / 3.0) / (float)(1 << 20);
00157 #else
00158 #define P 15
00159         scale_factor_shift[i] = 21 - P - (i / 3);
00160         scale_factor_mult[i] = (1 << P) * pow(2.0, (i % 3) / 3.0);
00161 #endif
00162     }
00163     for(i=0;i<128;i++) {
00164         v = i - 64;
00165         if (v <= -3)
00166             v = 0;
00167         else if (v < 0)
00168             v = 1;
00169         else if (v == 0)
00170             v = 2;
00171         else if (v < 3)
00172             v = 3;
00173         else
00174             v = 4;
00175         scale_diff_table[i] = v;
00176     }
00177 
00178     for(i=0;i<17;i++) {
00179         v = ff_mpa_quant_bits[i];
00180         if (v < 0)
00181             v = -v;
00182         else
00183             v = v * 3;
00184         total_quant_bits[i] = 12 * v;
00185     }
00186 
00187     avctx->coded_frame= avcodec_alloc_frame();
00188     avctx->coded_frame->key_frame= 1;
00189 
00190     return 0;
00191 }
00192 
00193 /* 32 point floating point IDCT without 1/sqrt(2) coef zero scaling */
00194 static void idct32(int *out, int *tab)
00195 {
00196     int i, j;
00197     int *t, *t1, xr;
00198     const int *xp = costab32;
00199 
00200     for(j=31;j>=3;j-=2) tab[j] += tab[j - 2];
00201 
00202     t = tab + 30;
00203     t1 = tab + 2;
00204     do {
00205         t[0] += t[-4];
00206         t[1] += t[1 - 4];
00207         t -= 4;
00208     } while (t != t1);
00209 
00210     t = tab + 28;
00211     t1 = tab + 4;
00212     do {
00213         t[0] += t[-8];
00214         t[1] += t[1-8];
00215         t[2] += t[2-8];
00216         t[3] += t[3-8];
00217         t -= 8;
00218     } while (t != t1);
00219 
00220     t = tab;
00221     t1 = tab + 32;
00222     do {
00223         t[ 3] = -t[ 3];
00224         t[ 6] = -t[ 6];
00225 
00226         t[11] = -t[11];
00227         t[12] = -t[12];
00228         t[13] = -t[13];
00229         t[15] = -t[15];
00230         t += 16;
00231     } while (t != t1);
00232 
00233 
00234     t = tab;
00235     t1 = tab + 8;
00236     do {
00237         int x1, x2, x3, x4;
00238 
00239         x3 = MUL(t[16], FIX(SQRT2*0.5));
00240         x4 = t[0] - x3;
00241         x3 = t[0] + x3;
00242 
00243         x2 = MUL(-(t[24] + t[8]), FIX(SQRT2*0.5));
00244         x1 = MUL((t[8] - x2), xp[0]);
00245         x2 = MUL((t[8] + x2), xp[1]);
00246 
00247         t[ 0] = x3 + x1;
00248         t[ 8] = x4 - x2;
00249         t[16] = x4 + x2;
00250         t[24] = x3 - x1;
00251         t++;
00252     } while (t != t1);
00253 
00254     xp += 2;
00255     t = tab;
00256     t1 = tab + 4;
00257     do {
00258         xr = MUL(t[28],xp[0]);
00259         t[28] = (t[0] - xr);
00260         t[0] = (t[0] + xr);
00261 
00262         xr = MUL(t[4],xp[1]);
00263         t[ 4] = (t[24] - xr);
00264         t[24] = (t[24] + xr);
00265 
00266         xr = MUL(t[20],xp[2]);
00267         t[20] = (t[8] - xr);
00268         t[ 8] = (t[8] + xr);
00269 
00270         xr = MUL(t[12],xp[3]);
00271         t[12] = (t[16] - xr);
00272         t[16] = (t[16] + xr);
00273         t++;
00274     } while (t != t1);
00275     xp += 4;
00276 
00277     for (i = 0; i < 4; i++) {
00278         xr = MUL(tab[30-i*4],xp[0]);
00279         tab[30-i*4] = (tab[i*4] - xr);
00280         tab[   i*4] = (tab[i*4] + xr);
00281 
00282         xr = MUL(tab[ 2+i*4],xp[1]);
00283         tab[ 2+i*4] = (tab[28-i*4] - xr);
00284         tab[28-i*4] = (tab[28-i*4] + xr);
00285 
00286         xr = MUL(tab[31-i*4],xp[0]);
00287         tab[31-i*4] = (tab[1+i*4] - xr);
00288         tab[ 1+i*4] = (tab[1+i*4] + xr);
00289 
00290         xr = MUL(tab[ 3+i*4],xp[1]);
00291         tab[ 3+i*4] = (tab[29-i*4] - xr);
00292         tab[29-i*4] = (tab[29-i*4] + xr);
00293 
00294         xp += 2;
00295     }
00296 
00297     t = tab + 30;
00298     t1 = tab + 1;
00299     do {
00300         xr = MUL(t1[0], *xp);
00301         t1[0] = (t[0] - xr);
00302         t[0] = (t[0] + xr);
00303         t -= 2;
00304         t1 += 2;
00305         xp++;
00306     } while (t >= tab);
00307 
00308     for(i=0;i<32;i++) {
00309         out[i] = tab[bitinv32[i]];
00310     }
00311 }
00312 
00313 #define WSHIFT (WFRAC_BITS + 15 - FRAC_BITS)
00314 
00315 static void filter(MpegAudioContext *s, int ch, short *samples, int incr)
00316 {
00317     short *p, *q;
00318     int sum, offset, i, j;
00319     int tmp[64];
00320     int tmp1[32];
00321     int *out;
00322 
00323     //    print_pow1(samples, 1152);
00324 
00325     offset = s->samples_offset[ch];
00326     out = &s->sb_samples[ch][0][0][0];
00327     for(j=0;j<36;j++) {
00328         /* 32 samples at once */
00329         for(i=0;i<32;i++) {
00330             s->samples_buf[ch][offset + (31 - i)] = samples[0];
00331             samples += incr;
00332         }
00333 
00334         /* filter */
00335         p = s->samples_buf[ch] + offset;
00336         q = filter_bank;
00337         /* maxsum = 23169 */
00338         for(i=0;i<64;i++) {
00339             sum = p[0*64] * q[0*64];
00340             sum += p[1*64] * q[1*64];
00341             sum += p[2*64] * q[2*64];
00342             sum += p[3*64] * q[3*64];
00343             sum += p[4*64] * q[4*64];
00344             sum += p[5*64] * q[5*64];
00345             sum += p[6*64] * q[6*64];
00346             sum += p[7*64] * q[7*64];
00347             tmp[i] = sum;
00348             p++;
00349             q++;
00350         }
00351         tmp1[0] = tmp[16] >> WSHIFT;
00352         for( i=1; i<=16; i++ ) tmp1[i] = (tmp[i+16]+tmp[16-i]) >> WSHIFT;
00353         for( i=17; i<=31; i++ ) tmp1[i] = (tmp[i+16]-tmp[80-i]) >> WSHIFT;
00354 
00355         idct32(out, tmp1);
00356 
00357         /* advance of 32 samples */
00358         offset -= 32;
00359         out += 32;
00360         /* handle the wrap around */
00361         if (offset < 0) {
00362             memmove(s->samples_buf[ch] + SAMPLES_BUF_SIZE - (512 - 32),
00363                     s->samples_buf[ch], (512 - 32) * 2);
00364             offset = SAMPLES_BUF_SIZE - 512;
00365         }
00366     }
00367     s->samples_offset[ch] = offset;
00368 
00369     //    print_pow(s->sb_samples, 1152);
00370 }
00371 
00372 static void compute_scale_factors(unsigned char scale_code[SBLIMIT],
00373                                   unsigned char scale_factors[SBLIMIT][3],
00374                                   int sb_samples[3][12][SBLIMIT],
00375                                   int sblimit)
00376 {
00377     int *p, vmax, v, n, i, j, k, code;
00378     int index, d1, d2;
00379     unsigned char *sf = &scale_factors[0][0];
00380 
00381     for(j=0;j<sblimit;j++) {
00382         for(i=0;i<3;i++) {
00383             /* find the max absolute value */
00384             p = &sb_samples[i][0][j];
00385             vmax = abs(*p);
00386             for(k=1;k<12;k++) {
00387                 p += SBLIMIT;
00388                 v = abs(*p);
00389                 if (v > vmax)
00390                     vmax = v;
00391             }
00392             /* compute the scale factor index using log 2 computations */
00393             if (vmax > 1) {
00394                 n = av_log2(vmax);
00395                 /* n is the position of the MSB of vmax. now
00396                    use at most 2 compares to find the index */
00397                 index = (21 - n) * 3 - 3;
00398                 if (index >= 0) {
00399                     while (vmax <= scale_factor_table[index+1])
00400                         index++;
00401                 } else {
00402                     index = 0; /* very unlikely case of overflow */
00403                 }
00404             } else {
00405                 index = 62; /* value 63 is not allowed */
00406             }
00407 
00408 #if 0
00409             printf("%2d:%d in=%x %x %d\n",
00410                    j, i, vmax, scale_factor_table[index], index);
00411 #endif
00412             /* store the scale factor */
00413             assert(index >=0 && index <= 63);
00414             sf[i] = index;
00415         }
00416 
00417         /* compute the transmission factor : look if the scale factors
00418            are close enough to each other */
00419         d1 = scale_diff_table[sf[0] - sf[1] + 64];
00420         d2 = scale_diff_table[sf[1] - sf[2] + 64];
00421 
00422         /* handle the 25 cases */
00423         switch(d1 * 5 + d2) {
00424         case 0*5+0:
00425         case 0*5+4:
00426         case 3*5+4:
00427         case 4*5+0:
00428         case 4*5+4:
00429             code = 0;
00430             break;
00431         case 0*5+1:
00432         case 0*5+2:
00433         case 4*5+1:
00434         case 4*5+2:
00435             code = 3;
00436             sf[2] = sf[1];
00437             break;
00438         case 0*5+3:
00439         case 4*5+3:
00440             code = 3;
00441             sf[1] = sf[2];
00442             break;
00443         case 1*5+0:
00444         case 1*5+4:
00445         case 2*5+4:
00446             code = 1;
00447             sf[1] = sf[0];
00448             break;
00449         case 1*5+1:
00450         case 1*5+2:
00451         case 2*5+0:
00452         case 2*5+1:
00453         case 2*5+2:
00454             code = 2;
00455             sf[1] = sf[2] = sf[0];
00456             break;
00457         case 2*5+3:
00458         case 3*5+3:
00459             code = 2;
00460             sf[0] = sf[1] = sf[2];
00461             break;
00462         case 3*5+0:
00463         case 3*5+1:
00464         case 3*5+2:
00465             code = 2;
00466             sf[0] = sf[2] = sf[1];
00467             break;
00468         case 1*5+3:
00469             code = 2;
00470             if (sf[0] > sf[2])
00471               sf[0] = sf[2];
00472             sf[1] = sf[2] = sf[0];
00473             break;
00474         default:
00475             assert(0); //cannot happen
00476             code = 0;           /* kill warning */
00477         }
00478 
00479 #if 0
00480         printf("%d: %2d %2d %2d %d %d -> %d\n", j,
00481                sf[0], sf[1], sf[2], d1, d2, code);
00482 #endif
00483         scale_code[j] = code;
00484         sf += 3;
00485     }
00486 }
00487 
00488 /* The most important function : psycho acoustic module. In this
00489    encoder there is basically none, so this is the worst you can do,
00490    but also this is the simpler. */
00491 static void psycho_acoustic_model(MpegAudioContext *s, short smr[SBLIMIT])
00492 {
00493     int i;
00494 
00495     for(i=0;i<s->sblimit;i++) {
00496         smr[i] = (int)(fixed_smr[i] * 10);
00497     }
00498 }
00499 
00500 
00501 #define SB_NOTALLOCATED  0
00502 #define SB_ALLOCATED     1
00503 #define SB_NOMORE        2
00504 
00505 /* Try to maximize the smr while using a number of bits inferior to
00506    the frame size. I tried to make the code simpler, faster and
00507    smaller than other encoders :-) */
00508 static void compute_bit_allocation(MpegAudioContext *s,
00509                                    short smr1[MPA_MAX_CHANNELS][SBLIMIT],
00510                                    unsigned char bit_alloc[MPA_MAX_CHANNELS][SBLIMIT],
00511                                    int *padding)
00512 {
00513     int i, ch, b, max_smr, max_ch, max_sb, current_frame_size, max_frame_size;
00514     int incr;
00515     short smr[MPA_MAX_CHANNELS][SBLIMIT];
00516     unsigned char subband_status[MPA_MAX_CHANNELS][SBLIMIT];
00517     const unsigned char *alloc;
00518 
00519     memcpy(smr, smr1, s->nb_channels * sizeof(short) * SBLIMIT);
00520     memset(subband_status, SB_NOTALLOCATED, s->nb_channels * SBLIMIT);
00521     memset(bit_alloc, 0, s->nb_channels * SBLIMIT);
00522 
00523     /* compute frame size and padding */
00524     max_frame_size = s->frame_size;
00525     s->frame_frac += s->frame_frac_incr;
00526     if (s->frame_frac >= 65536) {
00527         s->frame_frac -= 65536;
00528         s->do_padding = 1;
00529         max_frame_size += 8;
00530     } else {
00531         s->do_padding = 0;
00532     }
00533 
00534     /* compute the header + bit alloc size */
00535     current_frame_size = 32;
00536     alloc = s->alloc_table;
00537     for(i=0;i<s->sblimit;i++) {
00538         incr = alloc[0];
00539         current_frame_size += incr * s->nb_channels;
00540         alloc += 1 << incr;
00541     }
00542     for(;;) {
00543         /* look for the subband with the largest signal to mask ratio */
00544         max_sb = -1;
00545         max_ch = -1;
00546         max_smr = INT_MIN;
00547         for(ch=0;ch<s->nb_channels;ch++) {
00548             for(i=0;i<s->sblimit;i++) {
00549                 if (smr[ch][i] > max_smr && subband_status[ch][i] != SB_NOMORE) {
00550                     max_smr = smr[ch][i];
00551                     max_sb = i;
00552                     max_ch = ch;
00553                 }
00554             }
00555         }
00556 #if 0
00557         printf("current=%d max=%d max_sb=%d alloc=%d\n",
00558                current_frame_size, max_frame_size, max_sb,
00559                bit_alloc[max_sb]);
00560 #endif
00561         if (max_sb < 0)
00562             break;
00563 
00564         /* find alloc table entry (XXX: not optimal, should use
00565            pointer table) */
00566         alloc = s->alloc_table;
00567         for(i=0;i<max_sb;i++) {
00568             alloc += 1 << alloc[0];
00569         }
00570 
00571         if (subband_status[max_ch][max_sb] == SB_NOTALLOCATED) {
00572             /* nothing was coded for this band: add the necessary bits */
00573             incr = 2 + nb_scale_factors[s->scale_code[max_ch][max_sb]] * 6;
00574             incr += total_quant_bits[alloc[1]];
00575         } else {
00576             /* increments bit allocation */
00577             b = bit_alloc[max_ch][max_sb];
00578             incr = total_quant_bits[alloc[b + 1]] -
00579                 total_quant_bits[alloc[b]];
00580         }
00581 
00582         if (current_frame_size + incr <= max_frame_size) {
00583             /* can increase size */
00584             b = ++bit_alloc[max_ch][max_sb];
00585             current_frame_size += incr;
00586             /* decrease smr by the resolution we added */
00587             smr[max_ch][max_sb] = smr1[max_ch][max_sb] - quant_snr[alloc[b]];
00588             /* max allocation size reached ? */
00589             if (b == ((1 << alloc[0]) - 1))
00590                 subband_status[max_ch][max_sb] = SB_NOMORE;
00591             else
00592                 subband_status[max_ch][max_sb] = SB_ALLOCATED;
00593         } else {
00594             /* cannot increase the size of this subband */
00595             subband_status[max_ch][max_sb] = SB_NOMORE;
00596         }
00597     }
00598     *padding = max_frame_size - current_frame_size;
00599     assert(*padding >= 0);
00600 
00601 #if 0
00602     for(i=0;i<s->sblimit;i++) {
00603         printf("%d ", bit_alloc[i]);
00604     }
00605     printf("\n");
00606 #endif
00607 }
00608 
00609 /*
00610  * Output the mpeg audio layer 2 frame. Note how the code is small
00611  * compared to other encoders :-)
00612  */
00613 static void encode_frame(MpegAudioContext *s,
00614                          unsigned char bit_alloc[MPA_MAX_CHANNELS][SBLIMIT],
00615                          int padding)
00616 {
00617     int i, j, k, l, bit_alloc_bits, b, ch;
00618     unsigned char *sf;
00619     int q[3];
00620     PutBitContext *p = &s->pb;
00621 
00622     /* header */
00623 
00624     put_bits(p, 12, 0xfff);
00625     put_bits(p, 1, 1 - s->lsf); /* 1 = mpeg1 ID, 0 = mpeg2 lsf ID */
00626     put_bits(p, 2, 4-2);  /* layer 2 */
00627     put_bits(p, 1, 1); /* no error protection */
00628     put_bits(p, 4, s->bitrate_index);
00629     put_bits(p, 2, s->freq_index);
00630     put_bits(p, 1, s->do_padding); /* use padding */
00631     put_bits(p, 1, 0);             /* private_bit */
00632     put_bits(p, 2, s->nb_channels == 2 ? MPA_STEREO : MPA_MONO);
00633     put_bits(p, 2, 0); /* mode_ext */
00634     put_bits(p, 1, 0); /* no copyright */
00635     put_bits(p, 1, 1); /* original */
00636     put_bits(p, 2, 0); /* no emphasis */
00637 
00638     /* bit allocation */
00639     j = 0;
00640     for(i=0;i<s->sblimit;i++) {
00641         bit_alloc_bits = s->alloc_table[j];
00642         for(ch=0;ch<s->nb_channels;ch++) {
00643             put_bits(p, bit_alloc_bits, bit_alloc[ch][i]);
00644         }
00645         j += 1 << bit_alloc_bits;
00646     }
00647 
00648     /* scale codes */
00649     for(i=0;i<s->sblimit;i++) {
00650         for(ch=0;ch<s->nb_channels;ch++) {
00651             if (bit_alloc[ch][i])
00652                 put_bits(p, 2, s->scale_code[ch][i]);
00653         }
00654     }
00655 
00656     /* scale factors */
00657     for(i=0;i<s->sblimit;i++) {
00658         for(ch=0;ch<s->nb_channels;ch++) {
00659             if (bit_alloc[ch][i]) {
00660                 sf = &s->scale_factors[ch][i][0];
00661                 switch(s->scale_code[ch][i]) {
00662                 case 0:
00663                     put_bits(p, 6, sf[0]);
00664                     put_bits(p, 6, sf[1]);
00665                     put_bits(p, 6, sf[2]);
00666                     break;
00667                 case 3:
00668                 case 1:
00669                     put_bits(p, 6, sf[0]);
00670                     put_bits(p, 6, sf[2]);
00671                     break;
00672                 case 2:
00673                     put_bits(p, 6, sf[0]);
00674                     break;
00675                 }
00676             }
00677         }
00678     }
00679 
00680     /* quantization & write sub band samples */
00681 
00682     for(k=0;k<3;k++) {
00683         for(l=0;l<12;l+=3) {
00684             j = 0;
00685             for(i=0;i<s->sblimit;i++) {
00686                 bit_alloc_bits = s->alloc_table[j];
00687                 for(ch=0;ch<s->nb_channels;ch++) {
00688                     b = bit_alloc[ch][i];
00689                     if (b) {
00690                         int qindex, steps, m, sample, bits;
00691                         /* we encode 3 sub band samples of the same sub band at a time */
00692                         qindex = s->alloc_table[j+b];
00693                         steps = ff_mpa_quant_steps[qindex];
00694                         for(m=0;m<3;m++) {
00695                             sample = s->sb_samples[ch][k][l + m][i];
00696                             /* divide by scale factor */
00697 #ifdef USE_FLOATS
00698                             {
00699                                 float a;
00700                                 a = (float)sample * scale_factor_inv_table[s->scale_factors[ch][i][k]];
00701                                 q[m] = (int)((a + 1.0) * steps * 0.5);
00702                             }
00703 #else
00704                             {
00705                                 int q1, e, shift, mult;
00706                                 e = s->scale_factors[ch][i][k];
00707                                 shift = scale_factor_shift[e];
00708                                 mult = scale_factor_mult[e];
00709 
00710                                 /* normalize to P bits */
00711                                 if (shift < 0)
00712                                     q1 = sample << (-shift);
00713                                 else
00714                                     q1 = sample >> shift;
00715                                 q1 = (q1 * mult) >> P;
00716                                 q[m] = ((q1 + (1 << P)) * steps) >> (P + 1);
00717                             }
00718 #endif
00719                             if (q[m] >= steps)
00720                                 q[m] = steps - 1;
00721                             assert(q[m] >= 0 && q[m] < steps);
00722                         }
00723                         bits = ff_mpa_quant_bits[qindex];
00724                         if (bits < 0) {
00725                             /* group the 3 values to save bits */
00726                             put_bits(p, -bits,
00727                                      q[0] + steps * (q[1] + steps * q[2]));
00728 #if 0
00729                             printf("%d: gr1 %d\n",
00730                                    i, q[0] + steps * (q[1] + steps * q[2]));
00731 #endif
00732                         } else {
00733 #if 0
00734                             printf("%d: gr3 %d %d %d\n",
00735                                    i, q[0], q[1], q[2]);
00736 #endif
00737                             put_bits(p, bits, q[0]);
00738                             put_bits(p, bits, q[1]);
00739                             put_bits(p, bits, q[2]);
00740                         }
00741                     }
00742                 }
00743                 /* next subband in alloc table */
00744                 j += 1 << bit_alloc_bits;
00745             }
00746         }
00747     }
00748 
00749     /* padding */
00750     for(i=0;i<padding;i++)
00751         put_bits(p, 1, 0);
00752 
00753     /* flush */
00754     flush_put_bits(p);
00755 }
00756 
00757 static int MPA_encode_frame(AVCodecContext *avctx,
00758                             unsigned char *frame, int buf_size, void *data)
00759 {
00760     MpegAudioContext *s = avctx->priv_data;
00761     short *samples = data;
00762     short smr[MPA_MAX_CHANNELS][SBLIMIT];
00763     unsigned char bit_alloc[MPA_MAX_CHANNELS][SBLIMIT];
00764     int padding, i;
00765 
00766     for(i=0;i<s->nb_channels;i++) {
00767         filter(s, i, samples + i, s->nb_channels);
00768     }
00769 
00770     for(i=0;i<s->nb_channels;i++) {
00771         compute_scale_factors(s->scale_code[i], s->scale_factors[i],
00772                               s->sb_samples[i], s->sblimit);
00773     }
00774     for(i=0;i<s->nb_channels;i++) {
00775         psycho_acoustic_model(s, smr[i]);
00776     }
00777     compute_bit_allocation(s, smr, bit_alloc, &padding);
00778 
00779     init_put_bits(&s->pb, frame, MPA_MAX_CODED_FRAME_SIZE);
00780 
00781     encode_frame(s, bit_alloc, padding);
00782 
00783     s->nb_samples += MPA_FRAME_SIZE;
00784     return pbBufPtr(&s->pb) - s->pb.buf;
00785 }
00786 
00787 static av_cold int MPA_encode_close(AVCodecContext *avctx)
00788 {
00789     av_freep(&avctx->coded_frame);
00790     return 0;
00791 }
00792 
00793 AVCodec mp2_encoder = {
00794     "mp2",
00795     CODEC_TYPE_AUDIO,
00796     CODEC_ID_MP2,
00797     sizeof(MpegAudioContext),
00798     MPA_encode_init,
00799     MPA_encode_frame,
00800     MPA_encode_close,
00801     NULL,
00802     .sample_fmts = (enum SampleFormat[]){SAMPLE_FMT_S16,SAMPLE_FMT_NONE},
00803     .long_name = NULL_IF_CONFIG_SMALL("MP2 (MPEG audio layer 2)"),
00804 };
00805 
00806 #undef FIX

Generated on Sat Feb 16 2013 09:23:13 for ffmpeg by  doxygen 1.7.1