... | ... |
@@ -57,7 +57,8 @@ OBJS-$(CONFIG_AAC_DECODER) += aacdec.o aactab.o aacsbr.o aacps.o \ |
57 | 57 |
OBJS-$(CONFIG_AAC_ENCODER) += aacenc.o aaccoder.o \ |
58 | 58 |
aacpsy.o aactab.o \ |
59 | 59 |
psymodel.o iirfilter.o \ |
60 |
- mpeg4audio.o kbdwin.o |
|
60 |
+ mpeg4audio.o kbdwin.o \ |
|
61 |
+ audio_frame_queue.o |
|
61 | 62 |
OBJS-$(CONFIG_AASC_DECODER) += aasc.o msrledec.o |
62 | 63 |
OBJS-$(CONFIG_AC3_DECODER) += ac3dec.o ac3dec_data.o ac3.o kbdwin.o |
63 | 64 |
OBJS-$(CONFIG_AC3_ENCODER) += ac3enc_float.o ac3enc.o ac3tab.o \ |
... | ... |
@@ -34,6 +34,7 @@ |
34 | 34 |
#include "avcodec.h" |
35 | 35 |
#include "put_bits.h" |
36 | 36 |
#include "dsputil.h" |
37 |
+#include "internal.h" |
|
37 | 38 |
#include "mpeg4audio.h" |
38 | 39 |
#include "kbdwin.h" |
39 | 40 |
#include "sinewin.h" |
... | ... |
@@ -476,8 +477,7 @@ static void put_bitstream_info(AVCodecContext *avctx, AACEncContext *s, |
476 | 476 |
* Deinterleave input samples. |
477 | 477 |
* Channels are reordered from Libav's default order to AAC order. |
478 | 478 |
*/ |
479 |
-static void deinterleave_input_samples(AACEncContext *s, |
|
480 |
- const float *samples, int nb_samples) |
|
479 |
+static void deinterleave_input_samples(AACEncContext *s, AVFrame *frame) |
|
481 | 480 |
{ |
482 | 481 |
int ch, i; |
483 | 482 |
const int sinc = s->channels; |
... | ... |
@@ -485,35 +485,43 @@ static void deinterleave_input_samples(AACEncContext *s, |
485 | 485 |
|
486 | 486 |
/* deinterleave and remap input samples */ |
487 | 487 |
for (ch = 0; ch < sinc; ch++) { |
488 |
- const float *sptr = samples + channel_map[ch]; |
|
489 |
- |
|
490 | 488 |
/* copy last 1024 samples of previous frame to the start of the current frame */ |
491 | 489 |
memcpy(&s->planar_samples[ch][1024], &s->planar_samples[ch][2048], 1024 * sizeof(s->planar_samples[0][0])); |
492 | 490 |
|
493 | 491 |
/* deinterleave */ |
494 |
- for (i = 2048; i < 2048 + nb_samples; i++) { |
|
495 |
- s->planar_samples[ch][i] = *sptr; |
|
496 |
- sptr += sinc; |
|
492 |
+ i = 2048; |
|
493 |
+ if (frame) { |
|
494 |
+ const float *sptr = ((const float *)frame->data[0]) + channel_map[ch]; |
|
495 |
+ for (; i < 2048 + frame->nb_samples; i++) { |
|
496 |
+ s->planar_samples[ch][i] = *sptr; |
|
497 |
+ sptr += sinc; |
|
498 |
+ } |
|
497 | 499 |
} |
498 | 500 |
memset(&s->planar_samples[ch][i], 0, |
499 | 501 |
(3072 - i) * sizeof(s->planar_samples[0][0])); |
500 | 502 |
} |
501 | 503 |
} |
502 | 504 |
|
503 |
-static int aac_encode_frame(AVCodecContext *avctx, |
|
504 |
- uint8_t *frame, int buf_size, void *data) |
|
505 |
+static int aac_encode_frame(AVCodecContext *avctx, AVPacket *avpkt, |
|
506 |
+ const AVFrame *frame, int *got_packet_ptr) |
|
505 | 507 |
{ |
506 | 508 |
AACEncContext *s = avctx->priv_data; |
507 | 509 |
float **samples = s->planar_samples, *samples2, *la, *overlap; |
508 | 510 |
ChannelElement *cpe; |
509 |
- int i, ch, w, g, chans, tag, start_ch; |
|
511 |
+ int i, ch, w, g, chans, tag, start_ch, ret; |
|
510 | 512 |
int chan_el_counter[4]; |
511 | 513 |
FFPsyWindowInfo windows[AAC_MAX_CHANNELS]; |
512 | 514 |
|
513 | 515 |
if (s->last_frame == 2) |
514 | 516 |
return 0; |
515 | 517 |
|
516 |
- deinterleave_input_samples(s, data, data ? avctx->frame_size : 0); |
|
518 |
+ /* add current frame to queue */ |
|
519 |
+ if (frame) { |
|
520 |
+ if ((ret = ff_af_queue_add(&s->afq, frame) < 0)) |
|
521 |
+ return ret; |
|
522 |
+ } |
|
523 |
+ |
|
524 |
+ deinterleave_input_samples(s, frame); |
|
517 | 525 |
if (s->psypp) |
518 | 526 |
ff_psy_preprocess(s->psypp, s->planar_samples, s->channels); |
519 | 527 |
|
... | ... |
@@ -532,7 +540,7 @@ static int aac_encode_frame(AVCodecContext *avctx, |
532 | 532 |
overlap = &samples[cur_channel][0]; |
533 | 533 |
samples2 = overlap + 1024; |
534 | 534 |
la = samples2 + (448+64); |
535 |
- if (!data) |
|
535 |
+ if (!frame) |
|
536 | 536 |
la = NULL; |
537 | 537 |
if (tag == TYPE_LFE) { |
538 | 538 |
wi[ch].window_type[0] = ONLY_LONG_SEQUENCE; |
... | ... |
@@ -565,7 +573,13 @@ static int aac_encode_frame(AVCodecContext *avctx, |
565 | 565 |
} |
566 | 566 |
do { |
567 | 567 |
int frame_bits; |
568 |
- init_put_bits(&s->pb, frame, buf_size*8); |
|
568 |
+ |
|
569 |
+ if ((ret = ff_alloc_packet(avpkt, 768 * s->channels))) { |
|
570 |
+ av_log(avctx, AV_LOG_ERROR, "Error getting output packet\n"); |
|
571 |
+ return ret; |
|
572 |
+ } |
|
573 |
+ init_put_bits(&s->pb, avpkt->data, avpkt->size); |
|
574 |
+ |
|
569 | 575 |
if ((avctx->frame_number & 0xFF)==1 && !(avctx->flags & CODEC_FLAG_BITEXACT)) |
570 | 576 |
put_bitstream_info(avctx, s, LIBAVCODEC_IDENT); |
571 | 577 |
start_ch = 0; |
... | ... |
@@ -645,10 +659,15 @@ static int aac_encode_frame(AVCodecContext *avctx, |
645 | 645 |
s->lambda = FFMIN(s->lambda, 65536.f); |
646 | 646 |
} |
647 | 647 |
|
648 |
- if (!data) |
|
648 |
+ if (!frame) |
|
649 | 649 |
s->last_frame++; |
650 | 650 |
|
651 |
- return put_bits_count(&s->pb)>>3; |
|
651 |
+ ff_af_queue_remove(&s->afq, avctx->frame_size, &avpkt->pts, |
|
652 |
+ &avpkt->duration); |
|
653 |
+ |
|
654 |
+ avpkt->size = put_bits_count(&s->pb) >> 3; |
|
655 |
+ *got_packet_ptr = 1; |
|
656 |
+ return 0; |
|
652 | 657 |
} |
653 | 658 |
|
654 | 659 |
static av_cold int aac_encode_end(AVCodecContext *avctx) |
... | ... |
@@ -662,6 +681,10 @@ static av_cold int aac_encode_end(AVCodecContext *avctx) |
662 | 662 |
ff_psy_preprocess_end(s->psypp); |
663 | 663 |
av_freep(&s->buffer.samples); |
664 | 664 |
av_freep(&s->cpe); |
665 |
+ ff_af_queue_close(&s->afq); |
|
666 |
+#if FF_API_OLD_ENCODE_AUDIO |
|
667 |
+ av_freep(&avctx->coded_frame); |
|
668 |
+#endif |
|
665 | 669 |
return 0; |
666 | 670 |
} |
667 | 671 |
|
... | ... |
@@ -695,6 +718,11 @@ static av_cold int alloc_buffers(AVCodecContext *avctx, AACEncContext *s) |
695 | 695 |
for(ch = 0; ch < s->channels; ch++) |
696 | 696 |
s->planar_samples[ch] = s->buffer.samples + 3 * 1024 * ch; |
697 | 697 |
|
698 |
+#if FF_API_OLD_ENCODE_AUDIO |
|
699 |
+ if (!(avctx->coded_frame = avcodec_alloc_frame())) |
|
700 |
+ goto alloc_fail; |
|
701 |
+#endif |
|
702 |
+ |
|
698 | 703 |
return 0; |
699 | 704 |
alloc_fail: |
700 | 705 |
return AVERROR(ENOMEM); |
... | ... |
@@ -756,6 +784,9 @@ static av_cold int aac_encode_init(AVCodecContext *avctx) |
756 | 756 |
for (i = 0; i < 428; i++) |
757 | 757 |
ff_aac_pow34sf_tab[i] = sqrt(ff_aac_pow2sf_tab[i] * sqrt(ff_aac_pow2sf_tab[i])); |
758 | 758 |
|
759 |
+ avctx->delay = 1024; |
|
760 |
+ ff_af_queue_init(avctx, &s->afq); |
|
761 |
+ |
|
759 | 762 |
return 0; |
760 | 763 |
fail: |
761 | 764 |
aac_encode_end(avctx); |
... | ... |
@@ -784,7 +815,7 @@ AVCodec ff_aac_encoder = { |
784 | 784 |
.id = CODEC_ID_AAC, |
785 | 785 |
.priv_data_size = sizeof(AACEncContext), |
786 | 786 |
.init = aac_encode_init, |
787 |
- .encode = aac_encode_frame, |
|
787 |
+ .encode2 = aac_encode_frame, |
|
788 | 788 |
.close = aac_encode_end, |
789 | 789 |
.capabilities = CODEC_CAP_SMALL_LAST_FRAME | CODEC_CAP_DELAY | CODEC_CAP_EXPERIMENTAL, |
790 | 790 |
.sample_fmts = (const enum AVSampleFormat[]){AV_SAMPLE_FMT_FLT,AV_SAMPLE_FMT_NONE}, |
... | ... |
@@ -27,7 +27,7 @@ |
27 | 27 |
#include "dsputil.h" |
28 | 28 |
|
29 | 29 |
#include "aac.h" |
30 |
- |
|
30 |
+#include "audio_frame_queue.h" |
|
31 | 31 |
#include "psymodel.h" |
32 | 32 |
|
33 | 33 |
typedef struct AACEncOptions { |
... | ... |
@@ -71,6 +71,7 @@ typedef struct AACEncContext { |
71 | 71 |
int cur_channel; |
72 | 72 |
int last_frame; |
73 | 73 |
float lambda; |
74 |
+ AudioFrameQueue afq; |
|
74 | 75 |
DECLARE_ALIGNED(16, int, qcoefs)[96]; ///< quantized coefficients |
75 | 76 |
DECLARE_ALIGNED(32, float, scoefs)[1024]; ///< scaled coefficients |
76 | 77 |
|