X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libavcodec%2Faac.c;h=35c2945b7428fdc29b5213a4bb64af7ac752eadf;hb=fbdae895f25b8f4b672d09823bcc17250871a145;hp=7dd0c5ea8b147df1d3e52d08c4a52e5935412fbf;hpb=9ffd5c1cee8d46128bf6b3faad8befb886763ae3;p=ffmpeg diff --git a/libavcodec/aac.c b/libavcodec/aac.c index 7dd0c5ea8b1..35c2945b742 100644 --- a/libavcodec/aac.c +++ b/libavcodec/aac.c @@ -79,6 +79,7 @@ #include "avcodec.h" #include "bitstream.h" #include "dsputil.h" +#include "lpc.h" #include "aac.h" #include "aactab.h" @@ -118,7 +119,7 @@ static int output_configure(AACContext *ac, enum ChannelPosition che_pos[4][MAX_ * Set up default 1:1 output mapping. * * For a 5.1 stream the output order will be: - * [ Front Left ] [ Front Right ] [ Center ] [ LFE ] [ Surround Left ] [ Surround Right ] + * [ Center ] [ Front Left ] [ Front Right ] [ LFE ] [ Surround Left ] [ Surround Right ] */ for(i = 0; i < MAX_ELEM_ID; i++) { @@ -593,15 +594,63 @@ static int decode_scalefactors(AACContext * ac, float sf[120], GetBitContext * g /** * Decode pulse data; reference: table 4.7. */ -static void decode_pulses(Pulse * pulse, GetBitContext * gb, const uint16_t * swb_offset) { - int i; +static int decode_pulses(Pulse * pulse, GetBitContext * gb, const uint16_t * swb_offset, int num_swb) { + int i, pulse_swb; pulse->num_pulse = get_bits(gb, 2) + 1; - pulse->pos[0] = get_bits(gb, 5) + swb_offset[get_bits(gb, 6)]; + pulse_swb = get_bits(gb, 6); + if (pulse_swb >= num_swb) + return -1; + pulse->pos[0] = swb_offset[pulse_swb]; + pulse->pos[0] += get_bits(gb, 5); + if (pulse->pos[0] > 1023) + return -1; pulse->amp[0] = get_bits(gb, 4); for (i = 1; i < pulse->num_pulse; i++) { pulse->pos[i] = get_bits(gb, 5) + pulse->pos[i-1]; + if (pulse->pos[i] > 1023) + return -1; pulse->amp[i] = get_bits(gb, 4); } + return 0; +} + +/** + * Decode Temporal Noise Shaping data; reference: table 4.48. + * + * @return Returns error status. 0 - OK, !0 - error + */ +static int decode_tns(AACContext * ac, TemporalNoiseShaping * tns, + GetBitContext * gb, const IndividualChannelStream * ics) { + int w, filt, i, coef_len, coef_res, coef_compress; + const int is8 = ics->window_sequence[0] == EIGHT_SHORT_SEQUENCE; + const int tns_max_order = is8 ? 7 : ac->m4ac.object_type == AOT_AAC_MAIN ? 20 : 12; + for (w = 0; w < ics->num_windows; w++) { + if ((tns->n_filt[w] = get_bits(gb, 2 - is8))) { + coef_res = get_bits1(gb); + + for (filt = 0; filt < tns->n_filt[w]; filt++) { + int tmp2_idx; + tns->length[w][filt] = get_bits(gb, 6 - 2*is8); + + if ((tns->order[w][filt] = get_bits(gb, 5 - 2*is8)) > tns_max_order) { + av_log(ac->avccontext, AV_LOG_ERROR, "TNS filter order %d is greater than maximum %d.", + tns->order[w][filt], tns_max_order); + tns->order[w][filt] = 0; + return -1; + } + if (tns->order[w][filt]) { + tns->direction[w][filt] = get_bits1(gb); + coef_compress = get_bits1(gb); + coef_len = coef_res + 3 - coef_compress; + tmp2_idx = 2*coef_compress + coef_res; + + for (i = 0; i < tns->order[w][filt]; i++) + tns->coef[w][filt][i] = tns_tmp2_map[tmp2_idx][get_bits(gb, coef_len)]; + } + } + } + } + return 0; } /** @@ -713,10 +762,19 @@ static int decode_spectrum_and_dequant(AACContext * ac, float coef[1024], GetBit } if (pulse_present) { + idx = 0; for(i = 0; i < pulse->num_pulse; i++){ float co = coef_base[ pulse->pos[i] ]; - float ico = co / sqrtf(sqrtf(fabsf(co))) + pulse->amp[i]; - coef_base[ pulse->pos[i] ] = cbrtf(fabsf(ico)) * ico; + while(offsets[idx + 1] <= pulse->pos[i]) + idx++; + if (band_type[idx] != NOISE_BT && sf[idx]) { + float ico = -pulse->amp[i]; + if (co) { + co /= sf[idx]; + ico = co / sqrtf(sqrtf(fabsf(co))) + (co > 0 ? -ico : ico); + } + coef_base[ pulse->pos[i] ] = cbrtf(fabsf(ico)) * ico * sf[idx]; + } } } return 0; @@ -761,7 +819,10 @@ static int decode_ics(AACContext * ac, SingleChannelElement * sce, GetBitContext av_log(ac->avccontext, AV_LOG_ERROR, "Pulse tool not allowed in eight short sequence.\n"); return -1; } - decode_pulses(&pulse, gb, ics->swb_offset); + if (decode_pulses(&pulse, gb, ics->swb_offset, ics->num_swb)) { + av_log(ac->avccontext, AV_LOG_ERROR, "Pulse data corrupt or invalid.\n"); + return -1; + } } if ((tns->present = get_bits1(gb)) && decode_tns(ac, tns, gb, ics)) return -1; @@ -889,7 +950,7 @@ static int decode_cpe(AACContext * ac, GetBitContext * gb, int elem_id) { */ static int decode_cce(AACContext * ac, GetBitContext * gb, ChannelElement * che) { int num_gain = 0; - int c, g, sfb, ret, idx = 0; + int c, g, sfb, ret; int sign; float scale; SingleChannelElement * sce = &che->ch[0]; @@ -906,7 +967,7 @@ static int decode_cce(AACContext * ac, GetBitContext * gb, ChannelElement * che) if (coup->ch_select[c] == 3) num_gain++; } else - coup->ch_select[c] = 1; + coup->ch_select[c] = 2; } coup->coupling_point += get_bits1(gb); @@ -918,37 +979,40 @@ static int decode_cce(AACContext * ac, GetBitContext * gb, ChannelElement * che) } sign = get_bits(gb, 1); - scale = pow(2., pow(2., get_bits(gb, 2) - 3)); + scale = pow(2., pow(2., (int)get_bits(gb, 2) - 3)); if ((ret = decode_ics(ac, sce, gb, 0, 0))) return ret; for (c = 0; c < num_gain; c++) { + int idx = 0; int cge = 1; int gain = 0; float gain_cache = 1.; if (c) { cge = coup->coupling_point == AFTER_IMDCT ? 1 : get_bits1(gb); gain = cge ? get_vlc2(gb, vlc_scalefactors.table, 7, 3) - 60: 0; - gain_cache = pow(scale, gain); + gain_cache = pow(scale, -gain); } - for (g = 0; g < sce->ics.num_window_groups; g++) - for (sfb = 0; sfb < sce->ics.max_sfb; sfb++, idx++) + for (g = 0; g < sce->ics.num_window_groups; g++) { + for (sfb = 0; sfb < sce->ics.max_sfb; sfb++, idx++) { if (sce->band_type[idx] != ZERO_BT) { if (!cge) { int t = get_vlc2(gb, vlc_scalefactors.table, 7, 3) - 60; if (t) { int s = 1; + t = gain += t; if (sign) { s -= 2 * (t & 0x1); t >>= 1; } - gain += t; - gain_cache = pow(scale, gain) * s; + gain_cache = pow(scale, -t) * s; } } coup->gain[c][idx] = gain_cache; } + } + } } return 0; } @@ -1066,6 +1130,49 @@ static int decode_extension_payload(AACContext * ac, GetBitContext * gb, int cnt return res; } +/** + * Decode Temporal Noise Shaping filter coefficients and apply all-pole filters; reference: 4.6.9.3. + * + * @param decode 1 if tool is used normally, 0 if tool is used in LTP. + * @param coef spectral coefficients + */ +static void apply_tns(float coef[1024], TemporalNoiseShaping * tns, IndividualChannelStream * ics, int decode) { + const int mmm = FFMIN(ics->tns_max_bands, ics->max_sfb); + int w, filt, m, i; + int bottom, top, order, start, end, size, inc; + float lpc[TNS_MAX_ORDER]; + + for (w = 0; w < ics->num_windows; w++) { + bottom = ics->num_swb; + for (filt = 0; filt < tns->n_filt[w]; filt++) { + top = bottom; + bottom = FFMAX(0, top - tns->length[w][filt]); + order = tns->order[w][filt]; + if (order == 0) + continue; + + // tns_decode_coef + compute_lpc_coefs(tns->coef[w][filt], order, lpc, 0, 0, 0); + + start = ics->swb_offset[FFMIN(bottom, mmm)]; + end = ics->swb_offset[FFMIN( top, mmm)]; + if ((size = end - start) <= 0) + continue; + if (tns->direction[w][filt]) { + inc = -1; start = end - 1; + } else { + inc = 1; + } + start += w * 128; + + // ar filter + for (m = 0; m < size; m++, start += inc) + for (i = 1; i <= FFMIN(m, order); i++) + coef[start] -= coef[start - i*inc] * lpc[i-1]; + } + } +} + /** * Conduct IMDCT and windowing. */ @@ -1074,60 +1181,65 @@ static void imdct_and_windowing(AACContext * ac, SingleChannelElement * sce) { float * in = sce->coeffs; float * out = sce->ret; float * saved = sce->saved; - const float * lwindow = ics->use_kb_window[0] ? ff_aac_kbd_long_1024 : ff_sine_1024; const float * swindow = ics->use_kb_window[0] ? ff_aac_kbd_short_128 : ff_sine_128; const float * lwindow_prev = ics->use_kb_window[1] ? ff_aac_kbd_long_1024 : ff_sine_1024; const float * swindow_prev = ics->use_kb_window[1] ? ff_aac_kbd_short_128 : ff_sine_128; float * buf = ac->buf_mdct; + DECLARE_ALIGNED(16, float, temp[128]); int i; + // imdct if (ics->window_sequence[0] == EIGHT_SHORT_SEQUENCE) { if (ics->window_sequence[1] == ONLY_LONG_SEQUENCE || ics->window_sequence[1] == LONG_STOP_SEQUENCE) av_log(ac->avccontext, AV_LOG_WARNING, "Transition from an ONLY_LONG or LONG_STOP to an EIGHT_SHORT sequence detected. " "If you heard an audible artifact, please submit the sample to the FFmpeg developers.\n"); - for (i = 0; i < 2048; i += 256) { - ff_imdct_calc(&ac->mdct_small, buf + i, in + i/2); - ac->dsp.vector_fmul_reverse(ac->revers + i/2, buf + i + 128, swindow, 128); - } - for (i = 0; i < 448; i++) out[i] = saved[i] + ac->add_bias; - - ac->dsp.vector_fmul_add_add(out + 448 + 0*128, buf + 0*128, swindow_prev, saved + 448 , ac->add_bias, 128, 1); - ac->dsp.vector_fmul_add_add(out + 448 + 1*128, buf + 2*128, swindow, ac->revers + 0*128, ac->add_bias, 128, 1); - ac->dsp.vector_fmul_add_add(out + 448 + 2*128, buf + 4*128, swindow, ac->revers + 1*128, ac->add_bias, 128, 1); - ac->dsp.vector_fmul_add_add(out + 448 + 3*128, buf + 6*128, swindow, ac->revers + 2*128, ac->add_bias, 128, 1); - ac->dsp.vector_fmul_add_add(out + 448 + 4*128, buf + 8*128, swindow, ac->revers + 3*128, ac->add_bias, 64, 1); - -#if 0 - vector_fmul_add_add_add(&ac->dsp, out + 448 + 1*128, buf + 2*128, swindow, saved + 448 + 1*128, ac->revers + 0*128, ac->add_bias, 128); - vector_fmul_add_add_add(&ac->dsp, out + 448 + 2*128, buf + 4*128, swindow, saved + 448 + 2*128, ac->revers + 1*128, ac->add_bias, 128); - vector_fmul_add_add_add(&ac->dsp, out + 448 + 3*128, buf + 6*128, swindow, saved + 448 + 3*128, ac->revers + 2*128, ac->add_bias, 128); - vector_fmul_add_add_add(&ac->dsp, out + 448 + 4*128, buf + 8*128, swindow, saved + 448 + 4*128, ac->revers + 3*128, ac->add_bias, 64); -#endif - - ac->dsp.vector_fmul_add_add(saved, buf + 1024 + 64, swindow + 64, ac->revers + 3*128+64, 0, 64, 1); - ac->dsp.vector_fmul_add_add(saved + 64, buf + 1024 + 2*128, swindow, ac->revers + 4*128, 0, 128, 1); - ac->dsp.vector_fmul_add_add(saved + 192, buf + 1024 + 4*128, swindow, ac->revers + 5*128, 0, 128, 1); - ac->dsp.vector_fmul_add_add(saved + 320, buf + 1024 + 6*128, swindow, ac->revers + 6*128, 0, 128, 1); - memcpy( saved + 448, ac->revers + 7*128, 128 * sizeof(float)); - memset( saved + 576, 0, 448 * sizeof(float)); + for (i = 0; i < 1024; i += 128) + ff_imdct_half(&ac->mdct_small, buf + i, in + i); + } else + ff_imdct_half(&ac->mdct, buf, in); + + /* window overlapping + * NOTE: To simplify the overlapping code, all 'meaningless' short to long + * and long to short transitions are considered to be short to short + * transitions. This leaves just two cases (long to long and short to short) + * with a little special sauce for EIGHT_SHORT_SEQUENCE. + */ + if ((ics->window_sequence[1] == ONLY_LONG_SEQUENCE || ics->window_sequence[1] == LONG_STOP_SEQUENCE) && + (ics->window_sequence[0] == ONLY_LONG_SEQUENCE || ics->window_sequence[0] == LONG_START_SEQUENCE)) { + ac->dsp.vector_fmul_window( out, saved, buf, lwindow_prev, ac->add_bias, 512); } else { - ff_imdct_calc(&ac->mdct, buf, in); - if (ics->window_sequence[0] == LONG_STOP_SEQUENCE) { - for (i = 0; i < 448; i++) out[i] = saved[i] + ac->add_bias; - ac->dsp.vector_fmul_add_add(out + 448, buf + 448, swindow_prev, saved + 448, ac->add_bias, 128, 1); - for (i = 576; i < 1024; i++) out[i] = buf[i] + saved[i] + ac->add_bias; - } else { - ac->dsp.vector_fmul_add_add(out, buf, lwindow_prev, saved, ac->add_bias, 1024, 1); - } - if (ics->window_sequence[0] == LONG_START_SEQUENCE) { - memcpy(saved, buf + 1024, 448 * sizeof(float)); - ac->dsp.vector_fmul_reverse(saved + 448, buf + 1024 + 448, swindow, 128); - memset(saved + 576, 0, 448 * sizeof(float)); + for (i = 0; i < 448; i++) + out[i] = saved[i] + ac->add_bias; + + if (ics->window_sequence[0] == EIGHT_SHORT_SEQUENCE) { + ac->dsp.vector_fmul_window(out + 448 + 0*128, saved + 448, buf + 0*128, swindow_prev, ac->add_bias, 64); + ac->dsp.vector_fmul_window(out + 448 + 1*128, buf + 0*128 + 64, buf + 1*128, swindow, ac->add_bias, 64); + ac->dsp.vector_fmul_window(out + 448 + 2*128, buf + 1*128 + 64, buf + 2*128, swindow, ac->add_bias, 64); + ac->dsp.vector_fmul_window(out + 448 + 3*128, buf + 2*128 + 64, buf + 3*128, swindow, ac->add_bias, 64); + ac->dsp.vector_fmul_window(temp, buf + 3*128 + 64, buf + 4*128, swindow, ac->add_bias, 64); + memcpy( out + 448 + 4*128, temp, 64 * sizeof(float)); } else { - ac->dsp.vector_fmul_reverse(saved, buf + 1024, lwindow, 1024); + ac->dsp.vector_fmul_window(out + 448, saved + 448, buf, swindow_prev, ac->add_bias, 64); + for (i = 576; i < 1024; i++) + out[i] = buf[i-512] + ac->add_bias; } } + + // buffer update + if (ics->window_sequence[0] == EIGHT_SHORT_SEQUENCE) { + for (i = 0; i < 64; i++) + saved[i] = temp[64 + i] - ac->add_bias; + ac->dsp.vector_fmul_window(saved + 64, buf + 4*128 + 64, buf + 5*128, swindow, 0, 64); + ac->dsp.vector_fmul_window(saved + 192, buf + 5*128 + 64, buf + 6*128, swindow, 0, 64); + ac->dsp.vector_fmul_window(saved + 320, buf + 6*128 + 64, buf + 7*128, swindow, 0, 64); + memcpy( saved + 448, buf + 7*128 + 64, 64 * sizeof(float)); + } else if (ics->window_sequence[0] == LONG_START_SEQUENCE) { + memcpy( saved, buf + 512, 448 * sizeof(float)); + memcpy( saved + 448, buf + 7*128 + 64, 64 * sizeof(float)); + } else { // LONG_STOP or ONLY_LONG + memcpy( saved, buf + 512, 512 * sizeof(float)); + } } /** @@ -1135,11 +1247,11 @@ static void imdct_and_windowing(AACContext * ac, SingleChannelElement * sce) { * * @param index index into coupling gain array */ -static void apply_dependent_coupling(AACContext * ac, SingleChannelElement * sce, ChannelElement * cc, int index) { - IndividualChannelStream * ics = &cc->ch[0].ics; +static void apply_dependent_coupling(AACContext * ac, SingleChannelElement * target, ChannelElement * cce, int index) { + IndividualChannelStream * ics = &cce->ch[0].ics; const uint16_t * offsets = ics->swb_offset; - float * dest = sce->coeffs; - const float * src = cc->ch[0].coeffs; + float * dest = target->coeffs; + const float * src = cce->ch[0].coeffs; int g, i, group, k, idx = 0; if(ac->m4ac.object_type == AOT_AAC_LTP) { av_log(ac->avccontext, AV_LOG_ERROR, @@ -1148,11 +1260,11 @@ static void apply_dependent_coupling(AACContext * ac, SingleChannelElement * sce } for (g = 0; g < ics->num_window_groups; g++) { for (i = 0; i < ics->max_sfb; i++, idx++) { - if (cc->ch[0].band_type[idx] != ZERO_BT) { + if (cce->ch[0].band_type[idx] != ZERO_BT) { for (group = 0; group < ics->group_len[g]; group++) { for (k = offsets[i]; k < offsets[i+1]; k++) { // XXX dsputil-ize - dest[group*128+k] += cc->coup.gain[index][idx] * src[group*128+k]; + dest[group*128+k] += cce->coup.gain[index][idx] * src[group*128+k]; } } } @@ -1167,10 +1279,10 @@ static void apply_dependent_coupling(AACContext * ac, SingleChannelElement * sce * * @param index index into coupling gain array */ -static void apply_independent_coupling(AACContext * ac, SingleChannelElement * sce, ChannelElement * cc, int index) { +static void apply_independent_coupling(AACContext * ac, SingleChannelElement * target, ChannelElement * cce, int index) { int i; for (i = 0; i < 1024; i++) - sce->ret[i] += cc->coup.gain[index][0] * (cc->ch[0].ret[i] - ac->add_bias); + target->ret[i] += cce->coup.gain[index][0] * (cce->ch[0].ret[i] - ac->add_bias); } /** @@ -1180,25 +1292,30 @@ static void apply_independent_coupling(AACContext * ac, SingleChannelElement * s * @param apply_coupling_method pointer to (in)dependent coupling function */ static void apply_channel_coupling(AACContext * ac, ChannelElement * cc, - void (*apply_coupling_method)(AACContext * ac, SingleChannelElement * sce, ChannelElement * cc, int index)) + enum RawDataBlockType type, int elem_id, enum CouplingPoint coupling_point, + void (*apply_coupling_method)(AACContext * ac, SingleChannelElement * target, ChannelElement * cce, int index)) { - int c; - int index = 0; - ChannelCoupling * coup = &cc->coup; - for (c = 0; c <= coup->num_coupled; c++) { - if (ac->che[coup->type[c]][coup->id_select[c]]) { - if (coup->ch_select[c] != 2) { - apply_coupling_method(ac, &ac->che[coup->type[c]][coup->id_select[c]]->ch[0], cc, index); - if (coup->ch_select[c] != 0) - index++; + int i, c; + + for (i = 0; i < MAX_ELEM_ID; i++) { + ChannelElement *cce = ac->che[TYPE_CCE][i]; + int index = 0; + + if (cce && cce->coup.coupling_point == coupling_point) { + ChannelCoupling * coup = &cce->coup; + + for (c = 0; c <= coup->num_coupled; c++) { + if (coup->type[c] == type && coup->id_select[c] == elem_id) { + if (coup->ch_select[c] != 1) { + apply_coupling_method(ac, &cc->ch[0], cce, index); + if (coup->ch_select[c] != 0) + index++; + } + if (coup->ch_select[c] != 2) + apply_coupling_method(ac, &cc->ch[1], cce, index++); + } else + index += 1 + (coup->ch_select[c] == 3); } - if (coup->ch_select[c] != 1) - apply_coupling_method(ac, &ac->che[coup->type[c]][coup->id_select[c]]->ch[1], cc, index++); - } else { - av_log(ac->avccontext, AV_LOG_ERROR, - "coupling target %sE[%d] not available\n", - coup->type[c] == TYPE_CPE ? "CP" : "SC", coup->id_select[c]); - break; } } } @@ -1208,23 +1325,24 @@ static void apply_channel_coupling(AACContext * ac, ChannelElement * cc, */ static void spectral_to_sample(AACContext * ac) { int i, type; - for (i = 0; i < MAX_ELEM_ID; i++) { - for(type = 0; type < 4; type++) { + for(type = 3; type >= 0; type--) { + for (i = 0; i < MAX_ELEM_ID; i++) { ChannelElement *che = ac->che[type][i]; if(che) { - if(che->coup.coupling_point == BEFORE_TNS) - apply_channel_coupling(ac, che, apply_dependent_coupling); + if(type <= TYPE_CPE) + apply_channel_coupling(ac, che, type, i, BEFORE_TNS, apply_dependent_coupling); if(che->ch[0].tns.present) apply_tns(che->ch[0].coeffs, &che->ch[0].tns, &che->ch[0].ics, 1); if(che->ch[1].tns.present) apply_tns(che->ch[1].coeffs, &che->ch[1].tns, &che->ch[1].ics, 1); - if(che->coup.coupling_point == BETWEEN_TNS_AND_IMDCT) - apply_channel_coupling(ac, che, apply_dependent_coupling); - imdct_and_windowing(ac, &che->ch[0]); + if(type <= TYPE_CPE) + apply_channel_coupling(ac, che, type, i, BETWEEN_TNS_AND_IMDCT, apply_dependent_coupling); + if(type != TYPE_CCE || che->coup.coupling_point == AFTER_IMDCT) + imdct_and_windowing(ac, &che->ch[0]); if(type == TYPE_CPE) imdct_and_windowing(ac, &che->ch[1]); - if(che->coup.coupling_point == AFTER_IMDCT) - apply_channel_coupling(ac, che, apply_independent_coupling); + if(type <= TYPE_CCE) + apply_channel_coupling(ac, che, type, i, AFTER_IMDCT, apply_independent_coupling); } } } @@ -1251,7 +1369,7 @@ static int aac_decode_frame(AVCodecContext * avccontext, void * data, int * data ac->che[TYPE_SCE][elem_id] = ac->che[TYPE_LFE][0]; ac->che[TYPE_LFE][0] = NULL; } - if(elem_type && elem_type < TYPE_DSE) { + if(elem_type < TYPE_DSE) { if(!ac->che[elem_type][elem_id]) return -1; if(elem_type != TYPE_CCE) @@ -1269,7 +1387,7 @@ static int aac_decode_frame(AVCodecContext * avccontext, void * data, int * data break; case TYPE_CCE: - err = decode_cce(ac, &gb, ac->che[TYPE_SCE][elem_id]); + err = decode_cce(ac, &gb, ac->che[TYPE_CCE][elem_id]); break; case TYPE_LFE: