Libav
|
00001 00022 #include "avcodec.h" 00023 #include "put_bits.h" 00024 #include "dsputil.h" 00025 #include "lpc.h" 00026 #include "mathops.h" 00027 00028 #define DEFAULT_FRAME_SIZE 4096 00029 #define DEFAULT_SAMPLE_SIZE 16 00030 #define MAX_CHANNELS 8 00031 #define ALAC_EXTRADATA_SIZE 36 00032 #define ALAC_FRAME_HEADER_SIZE 55 00033 #define ALAC_FRAME_FOOTER_SIZE 3 00034 00035 #define ALAC_ESCAPE_CODE 0x1FF 00036 #define ALAC_MAX_LPC_ORDER 30 00037 #define DEFAULT_MAX_PRED_ORDER 6 00038 #define DEFAULT_MIN_PRED_ORDER 4 00039 #define ALAC_MAX_LPC_PRECISION 9 00040 #define ALAC_MAX_LPC_SHIFT 9 00041 00042 #define ALAC_CHMODE_LEFT_RIGHT 0 00043 #define ALAC_CHMODE_LEFT_SIDE 1 00044 #define ALAC_CHMODE_RIGHT_SIDE 2 00045 #define ALAC_CHMODE_MID_SIDE 3 00046 00047 typedef struct RiceContext { 00048 int history_mult; 00049 int initial_history; 00050 int k_modifier; 00051 int rice_modifier; 00052 } RiceContext; 00053 00054 typedef struct LPCContext { 00055 int lpc_order; 00056 int lpc_coeff[ALAC_MAX_LPC_ORDER+1]; 00057 int lpc_quant; 00058 } LPCContext; 00059 00060 typedef struct AlacEncodeContext { 00061 int compression_level; 00062 int min_prediction_order; 00063 int max_prediction_order; 00064 int max_coded_frame_size; 00065 int write_sample_size; 00066 int32_t sample_buf[MAX_CHANNELS][DEFAULT_FRAME_SIZE]; 00067 int32_t predictor_buf[DEFAULT_FRAME_SIZE]; 00068 int interlacing_shift; 00069 int interlacing_leftweight; 00070 PutBitContext pbctx; 00071 RiceContext rc; 00072 LPCContext lpc[MAX_CHANNELS]; 00073 DSPContext dspctx; 00074 AVCodecContext *avctx; 00075 } AlacEncodeContext; 00076 00077 00078 static void init_sample_buffers(AlacEncodeContext *s, int16_t *input_samples) 00079 { 00080 int ch, i; 00081 00082 for(ch=0;ch<s->avctx->channels;ch++) { 00083 int16_t *sptr = input_samples + ch; 00084 for(i=0;i<s->avctx->frame_size;i++) { 00085 s->sample_buf[ch][i] = *sptr; 00086 sptr += s->avctx->channels; 00087 } 00088 } 00089 } 00090 00091 static void encode_scalar(AlacEncodeContext *s, int x, int k, int write_sample_size) 00092 { 00093 int divisor, q, r; 00094 00095 k = FFMIN(k, s->rc.k_modifier); 00096 divisor = (1<<k) - 1; 00097 q = x / divisor; 00098 r = x % divisor; 00099 00100 if(q > 8) { 00101 // write escape code and sample value directly 00102 put_bits(&s->pbctx, 9, ALAC_ESCAPE_CODE); 00103 put_bits(&s->pbctx, write_sample_size, x); 00104 } else { 00105 if(q) 00106 put_bits(&s->pbctx, q, (1<<q) - 1); 00107 put_bits(&s->pbctx, 1, 0); 00108 00109 if(k != 1) { 00110 if(r > 0) 00111 put_bits(&s->pbctx, k, r+1); 00112 else 00113 put_bits(&s->pbctx, k-1, 0); 00114 } 00115 } 00116 } 00117 00118 static void write_frame_header(AlacEncodeContext *s, int is_verbatim) 00119 { 00120 put_bits(&s->pbctx, 3, s->avctx->channels-1); // No. of channels -1 00121 put_bits(&s->pbctx, 16, 0); // Seems to be zero 00122 put_bits(&s->pbctx, 1, 1); // Sample count is in the header 00123 put_bits(&s->pbctx, 2, 0); // FIXME: Wasted bytes field 00124 put_bits(&s->pbctx, 1, is_verbatim); // Audio block is verbatim 00125 put_bits32(&s->pbctx, s->avctx->frame_size); // No. of samples in the frame 00126 } 00127 00128 static void calc_predictor_params(AlacEncodeContext *s, int ch) 00129 { 00130 int32_t coefs[MAX_LPC_ORDER][MAX_LPC_ORDER]; 00131 int shift[MAX_LPC_ORDER]; 00132 int opt_order; 00133 00134 if (s->compression_level == 1) { 00135 s->lpc[ch].lpc_order = 6; 00136 s->lpc[ch].lpc_quant = 6; 00137 s->lpc[ch].lpc_coeff[0] = 160; 00138 s->lpc[ch].lpc_coeff[1] = -190; 00139 s->lpc[ch].lpc_coeff[2] = 170; 00140 s->lpc[ch].lpc_coeff[3] = -130; 00141 s->lpc[ch].lpc_coeff[4] = 80; 00142 s->lpc[ch].lpc_coeff[5] = -25; 00143 } else { 00144 opt_order = ff_lpc_calc_coefs(&s->dspctx, s->sample_buf[ch], 00145 s->avctx->frame_size, 00146 s->min_prediction_order, 00147 s->max_prediction_order, 00148 ALAC_MAX_LPC_PRECISION, coefs, shift, 1, 00149 ORDER_METHOD_EST, ALAC_MAX_LPC_SHIFT, 1); 00150 00151 s->lpc[ch].lpc_order = opt_order; 00152 s->lpc[ch].lpc_quant = shift[opt_order-1]; 00153 memcpy(s->lpc[ch].lpc_coeff, coefs[opt_order-1], opt_order*sizeof(int)); 00154 } 00155 } 00156 00157 static int estimate_stereo_mode(int32_t *left_ch, int32_t *right_ch, int n) 00158 { 00159 int i, best; 00160 int32_t lt, rt; 00161 uint64_t sum[4]; 00162 uint64_t score[4]; 00163 00164 /* calculate sum of 2nd order residual for each channel */ 00165 sum[0] = sum[1] = sum[2] = sum[3] = 0; 00166 for(i=2; i<n; i++) { 00167 lt = left_ch[i] - 2*left_ch[i-1] + left_ch[i-2]; 00168 rt = right_ch[i] - 2*right_ch[i-1] + right_ch[i-2]; 00169 sum[2] += FFABS((lt + rt) >> 1); 00170 sum[3] += FFABS(lt - rt); 00171 sum[0] += FFABS(lt); 00172 sum[1] += FFABS(rt); 00173 } 00174 00175 /* calculate score for each mode */ 00176 score[0] = sum[0] + sum[1]; 00177 score[1] = sum[0] + sum[3]; 00178 score[2] = sum[1] + sum[3]; 00179 score[3] = sum[2] + sum[3]; 00180 00181 /* return mode with lowest score */ 00182 best = 0; 00183 for(i=1; i<4; i++) { 00184 if(score[i] < score[best]) { 00185 best = i; 00186 } 00187 } 00188 return best; 00189 } 00190 00191 static void alac_stereo_decorrelation(AlacEncodeContext *s) 00192 { 00193 int32_t *left = s->sample_buf[0], *right = s->sample_buf[1]; 00194 int i, mode, n = s->avctx->frame_size; 00195 int32_t tmp; 00196 00197 mode = estimate_stereo_mode(left, right, n); 00198 00199 switch(mode) 00200 { 00201 case ALAC_CHMODE_LEFT_RIGHT: 00202 s->interlacing_leftweight = 0; 00203 s->interlacing_shift = 0; 00204 break; 00205 00206 case ALAC_CHMODE_LEFT_SIDE: 00207 for(i=0; i<n; i++) { 00208 right[i] = left[i] - right[i]; 00209 } 00210 s->interlacing_leftweight = 1; 00211 s->interlacing_shift = 0; 00212 break; 00213 00214 case ALAC_CHMODE_RIGHT_SIDE: 00215 for(i=0; i<n; i++) { 00216 tmp = right[i]; 00217 right[i] = left[i] - right[i]; 00218 left[i] = tmp + (right[i] >> 31); 00219 } 00220 s->interlacing_leftweight = 1; 00221 s->interlacing_shift = 31; 00222 break; 00223 00224 default: 00225 for(i=0; i<n; i++) { 00226 tmp = left[i]; 00227 left[i] = (tmp + right[i]) >> 1; 00228 right[i] = tmp - right[i]; 00229 } 00230 s->interlacing_leftweight = 1; 00231 s->interlacing_shift = 1; 00232 break; 00233 } 00234 } 00235 00236 static void alac_linear_predictor(AlacEncodeContext *s, int ch) 00237 { 00238 int i; 00239 LPCContext lpc = s->lpc[ch]; 00240 00241 if(lpc.lpc_order == 31) { 00242 s->predictor_buf[0] = s->sample_buf[ch][0]; 00243 00244 for(i=1; i<s->avctx->frame_size; i++) 00245 s->predictor_buf[i] = s->sample_buf[ch][i] - s->sample_buf[ch][i-1]; 00246 00247 return; 00248 } 00249 00250 // generalised linear predictor 00251 00252 if(lpc.lpc_order > 0) { 00253 int32_t *samples = s->sample_buf[ch]; 00254 int32_t *residual = s->predictor_buf; 00255 00256 // generate warm-up samples 00257 residual[0] = samples[0]; 00258 for(i=1;i<=lpc.lpc_order;i++) 00259 residual[i] = samples[i] - samples[i-1]; 00260 00261 // perform lpc on remaining samples 00262 for(i = lpc.lpc_order + 1; i < s->avctx->frame_size; i++) { 00263 int sum = 1 << (lpc.lpc_quant - 1), res_val, j; 00264 00265 for (j = 0; j < lpc.lpc_order; j++) { 00266 sum += (samples[lpc.lpc_order-j] - samples[0]) * 00267 lpc.lpc_coeff[j]; 00268 } 00269 00270 sum >>= lpc.lpc_quant; 00271 sum += samples[0]; 00272 residual[i] = sign_extend(samples[lpc.lpc_order+1] - sum, 00273 s->write_sample_size); 00274 res_val = residual[i]; 00275 00276 if(res_val) { 00277 int index = lpc.lpc_order - 1; 00278 int neg = (res_val < 0); 00279 00280 while(index >= 0 && (neg ? (res_val < 0):(res_val > 0))) { 00281 int val = samples[0] - samples[lpc.lpc_order - index]; 00282 int sign = (val ? FFSIGN(val) : 0); 00283 00284 if(neg) 00285 sign*=-1; 00286 00287 lpc.lpc_coeff[index] -= sign; 00288 val *= sign; 00289 res_val -= ((val >> lpc.lpc_quant) * 00290 (lpc.lpc_order - index)); 00291 index--; 00292 } 00293 } 00294 samples++; 00295 } 00296 } 00297 } 00298 00299 static void alac_entropy_coder(AlacEncodeContext *s) 00300 { 00301 unsigned int history = s->rc.initial_history; 00302 int sign_modifier = 0, i, k; 00303 int32_t *samples = s->predictor_buf; 00304 00305 for(i=0;i < s->avctx->frame_size;) { 00306 int x; 00307 00308 k = av_log2((history >> 9) + 3); 00309 00310 x = -2*(*samples)-1; 00311 x ^= (x>>31); 00312 00313 samples++; 00314 i++; 00315 00316 encode_scalar(s, x - sign_modifier, k, s->write_sample_size); 00317 00318 history += x * s->rc.history_mult 00319 - ((history * s->rc.history_mult) >> 9); 00320 00321 sign_modifier = 0; 00322 if(x > 0xFFFF) 00323 history = 0xFFFF; 00324 00325 if((history < 128) && (i < s->avctx->frame_size)) { 00326 unsigned int block_size = 0; 00327 00328 k = 7 - av_log2(history) + ((history + 16) >> 6); 00329 00330 while((*samples == 0) && (i < s->avctx->frame_size)) { 00331 samples++; 00332 i++; 00333 block_size++; 00334 } 00335 encode_scalar(s, block_size, k, 16); 00336 00337 sign_modifier = (block_size <= 0xFFFF); 00338 00339 history = 0; 00340 } 00341 00342 } 00343 } 00344 00345 static void write_compressed_frame(AlacEncodeContext *s) 00346 { 00347 int i, j; 00348 00349 if(s->avctx->channels == 2) 00350 alac_stereo_decorrelation(s); 00351 put_bits(&s->pbctx, 8, s->interlacing_shift); 00352 put_bits(&s->pbctx, 8, s->interlacing_leftweight); 00353 00354 for(i=0;i<s->avctx->channels;i++) { 00355 00356 calc_predictor_params(s, i); 00357 00358 put_bits(&s->pbctx, 4, 0); // prediction type : currently only type 0 has been RE'd 00359 put_bits(&s->pbctx, 4, s->lpc[i].lpc_quant); 00360 00361 put_bits(&s->pbctx, 3, s->rc.rice_modifier); 00362 put_bits(&s->pbctx, 5, s->lpc[i].lpc_order); 00363 // predictor coeff. table 00364 for(j=0;j<s->lpc[i].lpc_order;j++) { 00365 put_sbits(&s->pbctx, 16, s->lpc[i].lpc_coeff[j]); 00366 } 00367 } 00368 00369 // apply lpc and entropy coding to audio samples 00370 00371 for(i=0;i<s->avctx->channels;i++) { 00372 alac_linear_predictor(s, i); 00373 alac_entropy_coder(s); 00374 } 00375 } 00376 00377 static av_cold int alac_encode_init(AVCodecContext *avctx) 00378 { 00379 AlacEncodeContext *s = avctx->priv_data; 00380 uint8_t *alac_extradata = av_mallocz(ALAC_EXTRADATA_SIZE+1); 00381 00382 avctx->frame_size = DEFAULT_FRAME_SIZE; 00383 avctx->bits_per_coded_sample = DEFAULT_SAMPLE_SIZE; 00384 00385 if(avctx->sample_fmt != SAMPLE_FMT_S16) { 00386 av_log(avctx, AV_LOG_ERROR, "only pcm_s16 input samples are supported\n"); 00387 return -1; 00388 } 00389 00390 // Set default compression level 00391 if(avctx->compression_level == FF_COMPRESSION_DEFAULT) 00392 s->compression_level = 2; 00393 else 00394 s->compression_level = av_clip(avctx->compression_level, 0, 2); 00395 00396 // Initialize default Rice parameters 00397 s->rc.history_mult = 40; 00398 s->rc.initial_history = 10; 00399 s->rc.k_modifier = 14; 00400 s->rc.rice_modifier = 4; 00401 00402 s->max_coded_frame_size = 8 + (avctx->frame_size*avctx->channels*avctx->bits_per_coded_sample>>3); 00403 00404 s->write_sample_size = avctx->bits_per_coded_sample + avctx->channels - 1; // FIXME: consider wasted_bytes 00405 00406 AV_WB32(alac_extradata, ALAC_EXTRADATA_SIZE); 00407 AV_WB32(alac_extradata+4, MKBETAG('a','l','a','c')); 00408 AV_WB32(alac_extradata+12, avctx->frame_size); 00409 AV_WB8 (alac_extradata+17, avctx->bits_per_coded_sample); 00410 AV_WB8 (alac_extradata+21, avctx->channels); 00411 AV_WB32(alac_extradata+24, s->max_coded_frame_size); 00412 AV_WB32(alac_extradata+28, avctx->sample_rate*avctx->channels*avctx->bits_per_coded_sample); // average bitrate 00413 AV_WB32(alac_extradata+32, avctx->sample_rate); 00414 00415 // Set relevant extradata fields 00416 if(s->compression_level > 0) { 00417 AV_WB8(alac_extradata+18, s->rc.history_mult); 00418 AV_WB8(alac_extradata+19, s->rc.initial_history); 00419 AV_WB8(alac_extradata+20, s->rc.k_modifier); 00420 } 00421 00422 s->min_prediction_order = DEFAULT_MIN_PRED_ORDER; 00423 if(avctx->min_prediction_order >= 0) { 00424 if(avctx->min_prediction_order < MIN_LPC_ORDER || 00425 avctx->min_prediction_order > ALAC_MAX_LPC_ORDER) { 00426 av_log(avctx, AV_LOG_ERROR, "invalid min prediction order: %d\n", avctx->min_prediction_order); 00427 return -1; 00428 } 00429 00430 s->min_prediction_order = avctx->min_prediction_order; 00431 } 00432 00433 s->max_prediction_order = DEFAULT_MAX_PRED_ORDER; 00434 if(avctx->max_prediction_order >= 0) { 00435 if(avctx->max_prediction_order < MIN_LPC_ORDER || 00436 avctx->max_prediction_order > ALAC_MAX_LPC_ORDER) { 00437 av_log(avctx, AV_LOG_ERROR, "invalid max prediction order: %d\n", avctx->max_prediction_order); 00438 return -1; 00439 } 00440 00441 s->max_prediction_order = avctx->max_prediction_order; 00442 } 00443 00444 if(s->max_prediction_order < s->min_prediction_order) { 00445 av_log(avctx, AV_LOG_ERROR, "invalid prediction orders: min=%d max=%d\n", 00446 s->min_prediction_order, s->max_prediction_order); 00447 return -1; 00448 } 00449 00450 avctx->extradata = alac_extradata; 00451 avctx->extradata_size = ALAC_EXTRADATA_SIZE; 00452 00453 avctx->coded_frame = avcodec_alloc_frame(); 00454 avctx->coded_frame->key_frame = 1; 00455 00456 s->avctx = avctx; 00457 dsputil_init(&s->dspctx, avctx); 00458 00459 return 0; 00460 } 00461 00462 static int alac_encode_frame(AVCodecContext *avctx, uint8_t *frame, 00463 int buf_size, void *data) 00464 { 00465 AlacEncodeContext *s = avctx->priv_data; 00466 PutBitContext *pb = &s->pbctx; 00467 int i, out_bytes, verbatim_flag = 0; 00468 00469 if(avctx->frame_size > DEFAULT_FRAME_SIZE) { 00470 av_log(avctx, AV_LOG_ERROR, "input frame size exceeded\n"); 00471 return -1; 00472 } 00473 00474 if(buf_size < 2*s->max_coded_frame_size) { 00475 av_log(avctx, AV_LOG_ERROR, "buffer size is too small\n"); 00476 return -1; 00477 } 00478 00479 verbatim: 00480 init_put_bits(pb, frame, buf_size); 00481 00482 if((s->compression_level == 0) || verbatim_flag) { 00483 // Verbatim mode 00484 int16_t *samples = data; 00485 write_frame_header(s, 1); 00486 for(i=0; i<avctx->frame_size*avctx->channels; i++) { 00487 put_sbits(pb, 16, *samples++); 00488 } 00489 } else { 00490 init_sample_buffers(s, data); 00491 write_frame_header(s, 0); 00492 write_compressed_frame(s); 00493 } 00494 00495 put_bits(pb, 3, 7); 00496 flush_put_bits(pb); 00497 out_bytes = put_bits_count(pb) >> 3; 00498 00499 if(out_bytes > s->max_coded_frame_size) { 00500 /* frame too large. use verbatim mode */ 00501 if(verbatim_flag || (s->compression_level == 0)) { 00502 /* still too large. must be an error. */ 00503 av_log(avctx, AV_LOG_ERROR, "error encoding frame\n"); 00504 return -1; 00505 } 00506 verbatim_flag = 1; 00507 goto verbatim; 00508 } 00509 00510 return out_bytes; 00511 } 00512 00513 static av_cold int alac_encode_close(AVCodecContext *avctx) 00514 { 00515 av_freep(&avctx->extradata); 00516 avctx->extradata_size = 0; 00517 av_freep(&avctx->coded_frame); 00518 return 0; 00519 } 00520 00521 AVCodec alac_encoder = { 00522 "alac", 00523 AVMEDIA_TYPE_AUDIO, 00524 CODEC_ID_ALAC, 00525 sizeof(AlacEncodeContext), 00526 alac_encode_init, 00527 alac_encode_frame, 00528 alac_encode_close, 00529 .capabilities = CODEC_CAP_SMALL_LAST_FRAME, 00530 .sample_fmts = (const enum SampleFormat[]){ SAMPLE_FMT_S16, SAMPLE_FMT_NONE}, 00531 .long_name = NULL_IF_CONFIG_SMALL("ALAC (Apple Lossless Audio Codec)"), 00532 };