Merge commit '2df0c32ea12ddfa72ba88309812bfb13b674130f'

* commit '2df0c32ea12ddfa72ba88309812bfb13b674130f': lavc: use a separate field for exporting audio encoder padding Conflicts: libavcodec/audio_frame_queue.c libavcodec/avcodec.h libavcodec/libvorbisenc.c libavcodec/utils.c libavcodec/version.h libavcodec/wmaenc.c

Michael Niedermayer committed 12 years ago

Commit da218952 parents 1f20fa2 2df0c3222 files · +63 −41Browse files

Unified · Split

22 files

doc/APIchanges +4 −0
libavcodec/aacenc.c +1 −1
libavcodec/ac3enc.c +1 −1
libavcodec/audio_frame_queue.c +2 −2
libavcodec/avcodec.h +18 −10
libavcodec/g722enc.c +2 −2
libavcodec/libfaac.c +1 −1
libavcodec/libfdk-aacenc.c +1 −1
libavcodec/libmp3lame.c +1 −1
libavcodec/libopencore-amr.c +2 −2
libavcodec/libopusenc.c +2 −2
libavcodec/libspeexenc.c +1 −1
libavcodec/libtwolame.c +2 −2
libavcodec/libvo-aacenc.c +1 −1
libavcodec/libvo-amrwbenc.c +2 −2
libavcodec/libvorbisenc.c +2 −2
libavcodec/mpegaudioenc_template.c +2 −2
libavcodec/nellymoserenc.c +1 −1
libavcodec/ra144enc.c +1 −1
libavcodec/utils.c +9 −0
libavcodec/version.h +5 −2
libavcodec/wmaenc.c +2 −4

doc/APIchanges+4−0

@@ -15,6 +15,10 @@ libavutil: 2014-08-09 · 14 unmodified lines
15	15
16	16	API changes, most recent first:
17	17
	18	+2014-10-13 - xxxxxxx - lavc 55.03.0 - avcodec.h
	19	+ Add AVCodecContext.initial_padding. Deprecate the use of AVCodecContext.delay
	20	+ for audio encoding.
	21	+
18	22	2014-09-xx - xxxxxxx - lavu 54.04.0 - pixdesc.h
19	23	Add API to return the name of frame and context color properties.
20	24

libavcodec/aacenc.c+1−1

@@ -776,7 +776,7 @@ static av_cold int aac_encode_init(AVCodecContext *avctx) · 775 unmodified lines
776	776	for (i = 0; i < 428; i++)
777	777	ff_aac_pow34sf_tab[i] = sqrt(ff_aac_pow2sf_tab[i] * sqrt(ff_aac_pow2sf_tab[i]));
778	778
779		− avctx->delay = 1024;
	779	+ avctx->initial_padding = 1024;
780	780	ff_af_queue_init(avctx, &s->afq);
781	781
782	782	return 0;

libavcodec/ac3enc.c+1−1

@@ -2434,7 +2434,7 @@ av_cold int ff_ac3_encode_init(AVCodecContext *avctx) · 2433 unmodified lines
2434	2434	return ret;
2435	2435
2436	2436	avctx->frame_size = AC3_BLOCK_SIZE * s->num_blocks;
2437		− avctx->delay = AC3_BLOCK_SIZE;
	2437	+ avctx->initial_padding = AC3_BLOCK_SIZE;
2438	2438
2439	2439	s->bitstream_mode = avctx->audio_service_type;
2440	2440	if (s->bitstream_mode == AV_AUDIO_SERVICE_TYPE_KARAOKE)

libavcodec/audio_frame_queue.c+2−2

@@ -28,8 +28,8 @@ · 27 unmodified lines
28	28	av_cold void ff_af_queue_init(AVCodecContext avctx, AudioFrameQueue afq)
29	29	{
30	30	afq->avctx = avctx;
31		− afq->remaining_delay = avctx->delay;
32		− afq->remaining_samples = avctx->delay;
	31	+ afq->remaining_delay = avctx->initial_padding;
	32	+ afq->remaining_samples = avctx->initial_padding;
33	33	afq->frame_count = 0;
34	34	}
35	35

libavcodec/avcodec.h+18−10

@@ -1373,16 +1373,7 @@ typedef struct AVCodecContext { · 1372 unmodified lines
1373	1373	* encoded input.
1374	1374	*
1375	1375	* Audio:
1376		− * For encoding, this is the number of "priming" samples added by the
1377		− * encoder to the beginning of the stream. The decoded output will be
1378		− * delayed by this many samples relative to the input to the encoder (or
1379		− * more, if the decoder adds its own padding).
1380		− * The timestamps on the output packets are adjusted by the encoder so
1381		− * that they always refer to the first sample of the data actually
1382		− * contained in the packet, including any added padding.
1383		− * E.g. if the timebase is 1/samplerate and the timestamp of the first
1384		− * input sample is 0, the timestamp of the first output packet will be
1385		− * -delay.
	1376	+ * For encoding, this field is unused (see initial_padding).
1386	1377	*
1387	1378	* For decoding, this is the number of samples the decoder needs to
1388	1379	* output before the decoder's output is valid. When seeking, you should
@@ -2980,6 +2971,23 @@ typedef struct AVCodecContext { · 1591 unmodified lines
2980	2971	*/
2981	2972	int side_data_only_packets;
2982	2973
	2974	+ /**
	2975	+ * Audio only. The number of "priming" samples (padding) inserted by the
	2976	+ * encoder at the beginning of the audio. I.e. this number of leading
	2977	+ * decoded samples must be discarded by the caller to get the original audio
	2978	+ * without leading padding.
	2979	+ *
	2980	+ * - decoding: unused
	2981	+ * - encoding: Set by libavcodec. The timestamps on the output packets are
	2982	+ * adjusted by the encoder so that they always refer to the
	2983	+ * first sample of the data actually contained in the packet,
	2984	+ * including any added padding. E.g. if the timebase is
	2985	+ * 1/samplerate and the timestamp of the first input sample is
	2986	+ * 0, the timestamp of the first output packet will be
	2987	+ * -initial_padding.
	2988	+ */
	2989	+ int initial_padding;
	2990	+
2983	2991	/**
2984	2992	* Timebase in which pkt_dts/pts and AVPacket.dts/pts are.
2985	2993	* Code outside libavcodec should access this field using:

libavcodec/g722enc.c+2−2

@@ -107,7 +107,7 @@ static av_cold int g722_encode_init(AVCodecContext * avctx) · 106 unmodified lines
107	107	a common packet size for VoIP applications */
108	108	avctx->frame_size = 320;
109	109	}
110		− avctx->delay = 22;
	110	+ avctx->initial_padding = 22;
111	111
112	112	if (avctx->trellis) {
113	113	/* validate trellis */
@@ -374,7 +374,7 @@ static int g722_encode_frame(AVCodecContext avctx, AVPacket avpkt, · 260 unmodified lines
374	374	}
375	375
376	376	if (frame->pts != AV_NOPTS_VALUE)
377		− avpkt->pts = frame->pts - ff_samples_to_time_base(avctx, avctx->delay);
	377	+ avpkt->pts = frame->pts - ff_samples_to_time_base(avctx, avctx->initial_padding);
378	378	*got_packet_ptr = 1;
379	379	return 0;
380	380	}

libavcodec/libfaac.c+1−1

@@ -167,7 +167,7 @@ static av_cold int Faac_encode_init(AVCodecContext *avctx) · 166 unmodified lines
167	167	}
168	168	}
169	169
170		− avctx->delay = FAAC_DELAY_SAMPLES;
	170	+ avctx->initial_padding = FAAC_DELAY_SAMPLES;
171	171	ff_af_queue_init(avctx, &s->afq);
172	172
173	173	return 0;

libavcodec/libfdk-aacenc.c+1−1

@@ -286,7 +286,7 @@ static av_cold int aac_encode_init(AVCodecContext *avctx) · 285 unmodified lines
286	286	}
287	287
288	288	avctx->frame_size = info.frameLength;
289		− avctx->delay = info.encoderDelay;
	289	+ avctx->initial_padding = info.encoderDelay;
290	290	ff_af_queue_init(avctx, &s->afq);
291	291
292	292	if (avctx->flags & CODEC_FLAG_GLOBAL_HEADER) {

libavcodec/libmp3lame.c+1−1

@@ -138,7 +138,7 @@ static av_cold int mp3lame_encode_init(AVCodecContext *avctx) · 137 unmodified lines
138	138	}
139	139
140	140	/* get encoder delay */
141		− avctx->delay = lame_get_encoder_delay(s->gfp) + 528 + 1;
	141	+ avctx->initial_padding = lame_get_encoder_delay(s->gfp) + 528 + 1;
142	142	ff_af_queue_init(avctx, &s->afq);
143	143
144	144	avctx->frame_size = lame_get_framesize(s->gfp);

libavcodec/libopencore-amr.c+2−2

@@ -199,7 +199,7 @@ static av_cold int amr_nb_encode_init(AVCodecContext *avctx) · 198 unmodified lines
199	199	}
200	200
201	201	avctx->frame_size = 160;
202		− avctx->delay = 50;
	202	+ avctx->initial_padding = 50;
203	203	ff_af_queue_init(avctx, &s->afq);
204	204
205	205	s->enc_state = Encoder_Interface_init(s->enc_dtx);
@@ -246,7 +246,7 @@ static int amr_nb_encode_frame(AVCodecContext avctx, AVPacket avpkt, · 40 unmodified lines
246	246	return AVERROR(ENOMEM);
247	247	memcpy(flush_buf, samples, frame->nb_samples * sizeof(*flush_buf));
248	248	samples = flush_buf;
249		− if (frame->nb_samples < avctx->frame_size - avctx->delay)
	249	+ if (frame->nb_samples < avctx->frame_size - avctx->initial_padding)
250	250	s->enc_last_frame = -1;
251	251	}
252	252	if ((ret = ff_af_queue_add(&s->afq, frame)) < 0) {

libavcodec/libopusenc.c+2−2

@@ -87,7 +87,7 @@ static void libopus_write_header(AVCodecContext *avctx, int stream_count, · 86 unmodified lines
87	87	bytestream_put_buffer(&p, "OpusHead", 8);
88	88	bytestream_put_byte(&p, 1); /* Version */
89	89	bytestream_put_byte(&p, channels);
90		− bytestream_put_le16(&p, avctx->delay); /* Lookahead samples at 48kHz */
	90	+ bytestream_put_le16(&p, avctx->initial_padding); /* Lookahead samples at 48kHz */
91	91	bytestream_put_le32(&p, avctx->sample_rate); /* Original sample rate */
92	92	bytestream_put_le16(&p, 0); /* Gain of 0dB is recommended. */
93	93
@@ -284,7 +284,7 @@ static av_cold int libopus_encode_init(AVCodecContext *avctx) · 190 unmodified lines
284	284	goto fail;
285	285	}
286	286
287		− ret = opus_multistream_encoder_ctl(enc, OPUS_GET_LOOKAHEAD(&avctx->delay));
	287	+ ret = opus_multistream_encoder_ctl(enc, OPUS_GET_LOOKAHEAD(&avctx->initial_padding));
288	288	if (ret != OPUS_OK)
289	289	av_log(avctx, AV_LOG_WARNING,
290	290	"Unable to get number of lookahead samples: %s\n",

libavcodec/libspeexenc.c+1−1

@@ -236,7 +236,7 @@ static av_cold int encode_init(AVCodecContext *avctx) · 235 unmodified lines
236	236	s->header.frames_per_packet = s->frames_per_packet;
237	237
238	238	/* set encoding delay */
239		− speex_encoder_ctl(s->enc_state, SPEEX_GET_LOOKAHEAD, &avctx->delay);
	239	+ speex_encoder_ctl(s->enc_state, SPEEX_GET_LOOKAHEAD, &avctx->initial_padding);
240	240	ff_af_queue_init(avctx, &s->afq);
241	241
242	242	/* create header packet bytes from header struct */

libavcodec/libtwolame.c+2−2

@@ -60,7 +60,7 @@ static av_cold int twolame_encode_init(AVCodecContext *avctx) · 59 unmodified lines
60	60	int ret;
61	61
62	62	avctx->frame_size = TWOLAME_SAMPLES_PER_FRAME;
63		− avctx->delay = 512 - 32 + 1;
	63	+ avctx->initial_padding = 512 - 32 + 1;
64	64
65	65	s->glopts = twolame_init();
66	66	if (!s->glopts)
@@ -155,7 +155,7 @@ static int twolame_encode_frame(AVCodecContext avctx, AVPacket avpkt, · 88 unmodified lines
155	155	if (frame) {
156	156	avpkt->duration = ff_samples_to_time_base(avctx, frame->nb_samples);
157	157	if (frame->pts != AV_NOPTS_VALUE)
158		− avpkt->pts = frame->pts - ff_samples_to_time_base(avctx, avctx->delay);
	158	+ avpkt->pts = frame->pts - ff_samples_to_time_base(avctx, avctx->initial_padding);
159	159	} else {
160	160	avpkt->pts = s->next_pts;
161	161	}

libavcodec/libvo-aacenc.c+1−1

@@ -61,7 +61,7 @@ static av_cold int aac_encode_init(AVCodecContext *avctx) · 60 unmodified lines
61	61	int index, ret;
62	62
63	63	avctx->frame_size = FRAME_SIZE;
64		− avctx->delay = ENC_DELAY;
	64	+ avctx->initial_padding = ENC_DELAY;
65	65	s->last_frame = 2;
66	66	ff_af_queue_init(avctx, &s->afq);
67	67

libavcodec/libvo-amrwbenc.c+2−2

@@ -93,7 +93,7 @@ static av_cold int amr_wb_encode_init(AVCodecContext *avctx) · 92 unmodified lines
93	93	s->last_bitrate = avctx->bit_rate;
94	94
95	95	avctx->frame_size = 320;
96		− avctx->delay = 80;
	96	+ avctx->initial_padding = 80;
97	97
98	98	s->state = E_IF_init();
99	99
@@ -129,7 +129,7 @@ static int amr_wb_encode_frame(AVCodecContext avctx, AVPacket avpkt, · 29 unmodified lines
129	129	}
130	130
131	131	if (frame->pts != AV_NOPTS_VALUE)
132		− avpkt->pts = frame->pts - ff_samples_to_time_base(avctx, avctx->delay);
	132	+ avpkt->pts = frame->pts - ff_samples_to_time_base(avctx, avctx->initial_padding);
133	133
134	134	avpkt->size = size;
135	135	*got_packet_ptr = 1;

libavcodec/libvorbisenc.c+2−2

@@ -345,8 +345,8 @@ static int libvorbis_encode_frame(AVCodecContext avctx, AVPacket avpkt, · 344 unmodified lines
345	345	if (duration > 0) {
346	346	/* we do not know encoder delay until we get the first packet from
347	347	* libvorbis, so we have to update the AudioFrameQueue counts */
348		− if (!avctx->delay && s->afq.frames) {
349		− avctx->delay = duration;
	348	+ if (!avctx->initial_padding && s->afq.frames) {
	349	+ avctx->initial_padding = duration;
350	350	av_assert0(!s->afq.remaining_delay);
351	351	s->afq.frames->duration += duration;
352	352	if (s->afq.frames->pts != AV_NOPTS_VALUE)

libavcodec/mpegaudioenc_template.c+2−2

@@ -89,7 +89,7 @@ static av_cold int MPA_encode_init(AVCodecContext *avctx) · 88 unmodified lines
89	89	bitrate = bitrate / 1000;
90	90	s->nb_channels = channels;
91	91	avctx->frame_size = MPA_FRAME_SIZE;
92		− avctx->delay = 512 - 32 + 1;
	92	+ avctx->initial_padding = 512 - 32 + 1;
93	93
94	94	/* encoding freq */
95	95	s->lsf = 0;
@@ -771,7 +771,7 @@ static int MPA_encode_frame(AVCodecContext avctx, AVPacket avpkt, · 675 unmodified lines
771	771	encode_frame(s, bit_alloc, padding);
772	772
773	773	if (frame->pts != AV_NOPTS_VALUE)
774		− avpkt->pts = frame->pts - ff_samples_to_time_base(avctx, avctx->delay);
	774	+ avpkt->pts = frame->pts - ff_samples_to_time_base(avctx, avctx->initial_padding);
775	775
776	776	avpkt->size = put_bits_count(&s->pb) / 8;
777	777	*got_packet_ptr = 1;

libavcodec/nellymoserenc.c+1−1

@@ -165,7 +165,7 @@ static av_cold int encode_init(AVCodecContext *avctx) · 164 unmodified lines
165	165	}
166	166
167	167	avctx->frame_size = NELLY_SAMPLES;
168		− avctx->delay = NELLY_BUF_LEN;
	168	+ avctx->initial_padding = NELLY_BUF_LEN;
169	169	ff_af_queue_init(avctx, &s->afq);
170	170	s->avctx = avctx;
171	171	if ((ret = ff_mdct_init(&s->mdct_ctx, 8, 0, 32768.0)) < 0)

libavcodec/ra144enc.c+1−1

@@ -55,7 +55,7 @@ static av_cold int ra144_encode_init(AVCodecContext * avctx) · 54 unmodified lines
55	55	return -1;
56	56	}
57	57	avctx->frame_size = NBLOCKS * BLOCKSIZE;
58		− avctx->delay = avctx->frame_size;
	58	+ avctx->initial_padding = avctx->frame_size;
59	59	avctx->bit_rate = 8000;
60	60	ractx = avctx->priv_data;
61	61	ractx->lpc_coef[0] = ractx->lpc_tables[0];

libavcodec/utils.c+9−0

@@ -1609,6 +1609,11 @@ int attribute_align_arg avcodec_open2(AVCodecContext avctx, const AVCodec code · 1608 unmodified lines
1609	1609
1610	1610	ret=0;
1611	1611
	1612	+#if FF_API_AUDIOENC_DELAY
	1613	+ if (av_codec_is_encoder(avctx->codec))
	1614	+ avctx->delay = avctx->initial_padding;
	1615	+#endif
	1616	+
1612	1617	if (av_codec_is_decoder(avctx->codec)) {
1613	1618	if (!avctx->bit_rate)
1614	1619	avctx->bit_rate = get_bit_rate(avctx);
@@ -1921,6 +1926,10 @@ end: · 306 unmodified lines
1921	1926	av_frame_free(&padded_frame);
1922	1927	av_free(extended_frame);
1923	1928
	1929	+#if FF_API_AUDIOENC_DELAY
	1930	+ avctx->delay = avctx->initial_padding;
	1931	+#endif
	1932	+
1924	1933	return ret;
1925	1934	}
1926	1935

libavcodec/version.h+5−2

@@ -29,8 +29,8 @@ · 28 unmodified lines
29	29	#include "libavutil/version.h"
30	30
31	31	#define LIBAVCODEC_VERSION_MAJOR 56
32		−#define LIBAVCODEC_VERSION_MINOR 4
33		−#define LIBAVCODEC_VERSION_MICRO 101
	32	+#define LIBAVCODEC_VERSION_MINOR 5
	33	+#define LIBAVCODEC_VERSION_MICRO 100
34	34
35	35	#define LIBAVCODEC_VERSION_INT AV_VERSION_INT(LIBAVCODEC_VERSION_MAJOR, \
36	36	LIBAVCODEC_VERSION_MINOR, \
@@ -175,5 +175,8 @@ · 138 unmodified lines
175	175	#ifndef FF_API_DV_FRAME_PROFILE
176	176	#define FF_API_DV_FRAME_PROFILE (LIBAVCODEC_VERSION_MAJOR < 57)
177	177	#endif
	178	+#ifndef FF_API_AUDIOENC_DELAY
	179	+#define FF_API_AUDIOENC_DELAY (LIBAVCODEC_VERSION_MAJOR < 58)
	180	+#endif
178	181
179	182	#endif /* AVCODEC_VERSION_H */

libavcodec/wmaenc.c+2−4

@@ -88,9 +88,7 @@ static av_cold int encode_init(AVCodecContext *avctx) · 87 unmodified lines
88	88	(avctx->sample_rate * 8);
89	89	block_align = FFMIN(block_align, MAX_CODED_SUPERFRAME_SIZE);
90	90	avctx->block_align = block_align;
91		−
92		− avctx->frame_size =
93		− avctx->delay = s->frame_len;
	91	+ avctx->frame_size = avctx->initial_padding = s->frame_len;
94	92
95	93	return 0;
96	94	}
@@ -403,7 +401,7 @@ static int encode_superframe(AVCodecContext avctx, AVPacket avpkt, · 306 unmodified lines
403	401	av_assert0(put_bits_ptr(&s->pb) - s->pb.buf == avctx->block_align);
404	402
405	403	if (frame->pts != AV_NOPTS_VALUE)
406		− avpkt->pts = frame->pts - ff_samples_to_time_base(avctx, avctx->delay);
	404	+ avpkt->pts = frame->pts - ff_samples_to_time_base(avctx, avctx->initial_padding);
407	405
408	406	avpkt->size = avctx->block_align;
409	407	*got_packet_ptr = 1;

Merge commit '2df0c32ea12ddfa72ba88309812bfb13b674130f'

Keyboard shortcuts

General

Issues and pull requests

Issues, pull requests and Inbox

Repository code