[FFmpeg-cvslog] alacenc: add support for multi-channel encoding
Justin Ruggles
git at videolan.org
Thu Dec 6 16:18:41 CET 2012
ffmpeg | branch: master | Justin Ruggles <justin.ruggles at gmail.com> | Tue Dec 4 13:46:20 2012 -0500| [5e1bbb8c7e7008151b36ab0ba253466f3fae64ef] | committer: Justin Ruggles
alacenc: add support for multi-channel encoding
> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=5e1bbb8c7e7008151b36ab0ba253466f3fae64ef
---
Changelog | 1 +
libavcodec/Makefile | 4 +-
libavcodec/alac.c | 46 +++------------------
libavcodec/alac_data.c | 56 ++++++++++++++++++++++++++
libavcodec/alac_data.h | 46 +++++++++++++++++++++
libavcodec/alacenc.c | 104 ++++++++++++++++++++++++++++++------------------
6 files changed, 177 insertions(+), 80 deletions(-)
diff --git a/Changelog b/Changelog
index 7216622..f241e6d 100644
--- a/Changelog
+++ b/Changelog
@@ -5,6 +5,7 @@ version <next>:
- ashowinfo audio filter
- 24-bit FLAC encoding
- audio volume filter
+- multi-channel ALAC encoding up to 7.1
version 9_beta2:
diff --git a/libavcodec/Makefile b/libavcodec/Makefile
index 8162f62..502489d 100644
--- a/libavcodec/Makefile
+++ b/libavcodec/Makefile
@@ -85,8 +85,8 @@ OBJS-$(CONFIG_AC3_DECODER) += ac3dec.o ac3dec_data.o ac3.o kbdwin.o
OBJS-$(CONFIG_AC3_ENCODER) += ac3enc_float.o ac3enc.o ac3tab.o \
ac3.o kbdwin.o
OBJS-$(CONFIG_AC3_FIXED_ENCODER) += ac3enc_fixed.o ac3enc.o ac3tab.o ac3.o
-OBJS-$(CONFIG_ALAC_DECODER) += alac.o
-OBJS-$(CONFIG_ALAC_ENCODER) += alacenc.o
+OBJS-$(CONFIG_ALAC_DECODER) += alac.o alac_data.o
+OBJS-$(CONFIG_ALAC_ENCODER) += alacenc.o alac_data.o
OBJS-$(CONFIG_ALS_DECODER) += alsdec.o bgmc.o mpeg4audio.o
OBJS-$(CONFIG_AMRNB_DECODER) += amrnbdec.o celp_filters.o \
celp_math.o acelp_filters.o \
diff --git a/libavcodec/alac.c b/libavcodec/alac.c
index 64c7f54..22882a5 100644
--- a/libavcodec/alac.c
+++ b/libavcodec/alac.c
@@ -52,9 +52,9 @@
#include "internal.h"
#include "unary.h"
#include "mathops.h"
+#include "alac_data.h"
#define ALAC_EXTRADATA_SIZE 36
-#define MAX_CHANNELS 8
typedef struct {
AVCodecContext *avctx;
@@ -76,40 +76,6 @@ typedef struct {
int nb_samples; /**< number of samples in the current frame */
} ALACContext;
-enum RawDataBlockType {
- /* At the moment, only SCE, CPE, LFE, and END are recognized. */
- TYPE_SCE,
- TYPE_CPE,
- TYPE_CCE,
- TYPE_LFE,
- TYPE_DSE,
- TYPE_PCE,
- TYPE_FIL,
- TYPE_END
-};
-
-static const uint8_t alac_channel_layout_offsets[8][8] = {
- { 0 },
- { 0, 1 },
- { 2, 0, 1 },
- { 2, 0, 1, 3 },
- { 2, 0, 1, 3, 4 },
- { 2, 0, 1, 4, 5, 3 },
- { 2, 0, 1, 4, 5, 6, 3 },
- { 2, 6, 7, 0, 1, 4, 5, 3 }
-};
-
-static const uint16_t alac_channel_layouts[8] = {
- AV_CH_LAYOUT_MONO,
- AV_CH_LAYOUT_STEREO,
- AV_CH_LAYOUT_SURROUND,
- AV_CH_LAYOUT_4POINT0,
- AV_CH_LAYOUT_5POINT0_BACK,
- AV_CH_LAYOUT_5POINT1_BACK,
- AV_CH_LAYOUT_6POINT1_BACK,
- AV_CH_LAYOUT_7POINT1_WIDE_BACK
-};
-
static inline unsigned int decode_scalar(GetBitContext *gb, int k, int bps)
{
unsigned int x = get_unary_0_9(gb);
@@ -431,7 +397,7 @@ static int alac_decode_frame(AVCodecContext *avctx, void *data,
int *got_frame_ptr, AVPacket *avpkt)
{
ALACContext *alac = avctx->priv_data;
- enum RawDataBlockType element;
+ enum AlacRawDataBlockType element;
int channels;
int ch, ret, got_end;
@@ -458,7 +424,7 @@ static int alac_decode_frame(AVCodecContext *avctx, void *data,
}
ret = decode_element(avctx, data,
- alac_channel_layout_offsets[alac->channels - 1][ch],
+ ff_alac_channel_layout_offsets[alac->channels - 1][ch],
channels);
if (ret < 0 && get_bits_left(&alac->gb))
return ret;
@@ -581,17 +547,17 @@ static av_cold int alac_decode_init(AVCodecContext * avctx)
av_log(avctx, AV_LOG_WARNING, "Invalid channel count\n");
alac->channels = avctx->channels;
} else {
- if (alac->channels > MAX_CHANNELS)
+ if (alac->channels > ALAC_MAX_CHANNELS)
alac->channels = avctx->channels;
else
avctx->channels = alac->channels;
}
- if (avctx->channels > MAX_CHANNELS) {
+ if (avctx->channels > ALAC_MAX_CHANNELS) {
av_log(avctx, AV_LOG_ERROR, "Unsupported channel count: %d\n",
avctx->channels);
return AVERROR_PATCHWELCOME;
}
- avctx->channel_layout = alac_channel_layouts[alac->channels - 1];
+ avctx->channel_layout = ff_alac_channel_layouts[alac->channels - 1];
if ((ret = allocate_buffers(alac)) < 0) {
av_log(avctx, AV_LOG_ERROR, "Error allocating buffers\n");
diff --git a/libavcodec/alac_data.c b/libavcodec/alac_data.c
new file mode 100644
index 0000000..9e13119
--- /dev/null
+++ b/libavcodec/alac_data.c
@@ -0,0 +1,56 @@
+/*
+ * ALAC encoder and decoder common data
+ *
+ * This file is part of Libav.
+ *
+ * Libav is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * Libav is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include "libavutil/channel_layout.h"
+#include "alac_data.h"
+
+const uint8_t ff_alac_channel_layout_offsets[ALAC_MAX_CHANNELS][ALAC_MAX_CHANNELS] = {
+ { 0 },
+ { 0, 1 },
+ { 2, 0, 1 },
+ { 2, 0, 1, 3 },
+ { 2, 0, 1, 3, 4 },
+ { 2, 0, 1, 4, 5, 3 },
+ { 2, 0, 1, 4, 5, 6, 3 },
+ { 2, 6, 7, 0, 1, 4, 5, 3 }
+};
+
+const uint64_t ff_alac_channel_layouts[ALAC_MAX_CHANNELS + 1] = {
+ AV_CH_LAYOUT_MONO,
+ AV_CH_LAYOUT_STEREO,
+ AV_CH_LAYOUT_SURROUND,
+ AV_CH_LAYOUT_4POINT0,
+ AV_CH_LAYOUT_5POINT0_BACK,
+ AV_CH_LAYOUT_5POINT1_BACK,
+ AV_CH_LAYOUT_6POINT1_BACK,
+ AV_CH_LAYOUT_7POINT1_WIDE_BACK,
+ 0
+};
+
+const enum AlacRawDataBlockType ff_alac_channel_elements[ALAC_MAX_CHANNELS][5] = {
+ { TYPE_SCE, },
+ { TYPE_CPE, },
+ { TYPE_SCE, TYPE_CPE, },
+ { TYPE_SCE, TYPE_CPE, TYPE_SCE },
+ { TYPE_SCE, TYPE_CPE, TYPE_CPE, },
+ { TYPE_SCE, TYPE_CPE, TYPE_CPE, TYPE_SCE, },
+ { TYPE_SCE, TYPE_CPE, TYPE_CPE, TYPE_SCE, TYPE_SCE, },
+ { TYPE_SCE, TYPE_CPE, TYPE_CPE, TYPE_CPE, TYPE_SCE, },
+};
diff --git a/libavcodec/alac_data.h b/libavcodec/alac_data.h
new file mode 100644
index 0000000..ebb1f33
--- /dev/null
+++ b/libavcodec/alac_data.h
@@ -0,0 +1,46 @@
+/*
+ * ALAC encoder and decoder common data
+ *
+ * This file is part of Libav.
+ *
+ * Libav is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * Libav is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#ifndef AVCODEC_ALAC_DATA_H
+#define AVCODEC_ALAC_DATA_H
+
+#include <stdint.h>
+
+enum AlacRawDataBlockType {
+ /* At the moment, only SCE, CPE, LFE, and END are recognized. */
+ TYPE_SCE,
+ TYPE_CPE,
+ TYPE_CCE,
+ TYPE_LFE,
+ TYPE_DSE,
+ TYPE_PCE,
+ TYPE_FIL,
+ TYPE_END
+};
+
+#define ALAC_MAX_CHANNELS 8
+
+extern const uint8_t ff_alac_channel_layout_offsets[ALAC_MAX_CHANNELS][ALAC_MAX_CHANNELS];
+
+extern const uint64_t ff_alac_channel_layouts[ALAC_MAX_CHANNELS + 1];
+
+extern const enum AlacRawDataBlockType ff_alac_channel_elements[ALAC_MAX_CHANNELS][5];
+
+#endif /* AVCODEC_ALAC_DATA_H */
diff --git a/libavcodec/alacenc.c b/libavcodec/alacenc.c
index 4d6bf7b..3bc920a 100644
--- a/libavcodec/alacenc.c
+++ b/libavcodec/alacenc.c
@@ -25,9 +25,9 @@
#include "internal.h"
#include "lpc.h"
#include "mathops.h"
+#include "alac_data.h"
#define DEFAULT_FRAME_SIZE 4096
-#define MAX_CHANNELS 8
#define ALAC_EXTRADATA_SIZE 36
#define ALAC_FRAME_HEADER_SIZE 55
#define ALAC_FRAME_FOOTER_SIZE 3
@@ -66,27 +66,27 @@ typedef struct AlacEncodeContext {
int max_coded_frame_size;
int write_sample_size;
int extra_bits;
- int32_t sample_buf[MAX_CHANNELS][DEFAULT_FRAME_SIZE];
+ int32_t sample_buf[2][DEFAULT_FRAME_SIZE];
int32_t predictor_buf[DEFAULT_FRAME_SIZE];
int interlacing_shift;
int interlacing_leftweight;
PutBitContext pbctx;
RiceContext rc;
- AlacLPCContext lpc[MAX_CHANNELS];
+ AlacLPCContext lpc[2];
LPCContext lpc_ctx;
AVCodecContext *avctx;
} AlacEncodeContext;
-static void init_sample_buffers(AlacEncodeContext *s,
- uint8_t * const *samples)
+static void init_sample_buffers(AlacEncodeContext *s, int channels,
+ uint8_t const *samples[2])
{
int ch, i;
int shift = av_get_bytes_per_sample(s->avctx->sample_fmt) * 8 -
s->avctx->bits_per_raw_sample;
#define COPY_SAMPLES(type) do { \
- for (ch = 0; ch < s->avctx->channels; ch++) { \
+ for (ch = 0; ch < channels; ch++) { \
int32_t *bptr = s->sample_buf[ch]; \
const type *sptr = (const type *)samples[ch]; \
for (i = 0; i < s->frame_size; i++) \
@@ -128,15 +128,18 @@ static void encode_scalar(AlacEncodeContext *s, int x,
}
}
-static void write_frame_header(AlacEncodeContext *s)
+static void write_element_header(AlacEncodeContext *s,
+ enum AlacRawDataBlockType element,
+ int instance)
{
int encode_fs = 0;
if (s->frame_size < DEFAULT_FRAME_SIZE)
encode_fs = 1;
- put_bits(&s->pbctx, 3, s->avctx->channels-1); // No. of channels -1
- put_bits(&s->pbctx, 16, 0); // Seems to be zero
+ put_bits(&s->pbctx, 3, element); // element type
+ put_bits(&s->pbctx, 4, instance); // element instance
+ put_bits(&s->pbctx, 12, 0); // unused header bits
put_bits(&s->pbctx, 1, encode_fs); // Sample count is in the header
put_bits(&s->pbctx, 2, s->extra_bits >> 3); // Extra bytes (for 24-bit)
put_bits(&s->pbctx, 1, s->verbatim); // Audio block is verbatim
@@ -355,42 +358,51 @@ static void alac_entropy_coder(AlacEncodeContext *s)
}
}
-static int write_frame(AlacEncodeContext *s, AVPacket *avpkt,
- uint8_t * const *samples)
+static void write_element(AlacEncodeContext *s,
+ enum AlacRawDataBlockType element, int instance,
+ const uint8_t *samples0, const uint8_t *samples1)
{
- int i, j;
+ uint8_t const *samples[2] = { samples0, samples1 };
+ int i, j, channels;
int prediction_type = 0;
PutBitContext *pb = &s->pbctx;
- init_put_bits(pb, avpkt->data, avpkt->size);
+ channels = element == TYPE_CPE ? 2 : 1;
if (s->verbatim) {
- write_frame_header(s);
+ write_element_header(s, element, instance);
/* samples are channel-interleaved in verbatim mode */
if (s->avctx->sample_fmt == AV_SAMPLE_FMT_S32P) {
int shift = 32 - s->avctx->bits_per_raw_sample;
- int32_t * const *samples_s32 = (int32_t * const *)samples;
+ int32_t const *samples_s32[2] = { (const int32_t *)samples0,
+ (const int32_t *)samples1 };
for (i = 0; i < s->frame_size; i++)
- for (j = 0; j < s->avctx->channels; j++)
+ for (j = 0; j < channels; j++)
put_sbits(pb, s->avctx->bits_per_raw_sample,
samples_s32[j][i] >> shift);
} else {
- int16_t * const *samples_s16 = (int16_t * const *)samples;
+ int16_t const *samples_s16[2] = { (const int16_t *)samples0,
+ (const int16_t *)samples1 };
for (i = 0; i < s->frame_size; i++)
- for (j = 0; j < s->avctx->channels; j++)
+ for (j = 0; j < channels; j++)
put_sbits(pb, s->avctx->bits_per_raw_sample,
samples_s16[j][i]);
}
} else {
- init_sample_buffers(s, samples);
- write_frame_header(s);
+ s->write_sample_size = s->avctx->bits_per_raw_sample - s->extra_bits +
+ channels - 1;
- if (s->avctx->channels == 2)
+ init_sample_buffers(s, channels, samples);
+ write_element_header(s, element, instance);
+
+ if (channels == 2)
alac_stereo_decorrelation(s);
+ else
+ s->interlacing_shift = s->interlacing_leftweight = 0;
put_bits(pb, 8, s->interlacing_shift);
put_bits(pb, 8, s->interlacing_leftweight);
- for (i = 0; i < s->avctx->channels; i++) {
+ for (i = 0; i < channels; i++) {
calc_predictor_params(s, i);
put_bits(pb, 4, prediction_type);
@@ -407,7 +419,7 @@ static int write_frame(AlacEncodeContext *s, AVPacket *avpkt,
if (s->extra_bits) {
uint32_t mask = (1 << s->extra_bits) - 1;
for (i = 0; i < s->frame_size; i++) {
- for (j = 0; j < s->avctx->channels; j++) {
+ for (j = 0; j < channels; j++) {
put_bits(pb, s->extra_bits, s->sample_buf[j][i] & mask);
s->sample_buf[j][i] >>= s->extra_bits;
}
@@ -415,8 +427,7 @@ static int write_frame(AlacEncodeContext *s, AVPacket *avpkt,
}
// apply lpc and entropy coding to audio samples
-
- for (i = 0; i < s->avctx->channels; i++) {
+ for (i = 0; i < channels; i++) {
alac_linear_predictor(s, i);
// TODO: determine when this will actually help. for now it's not used.
@@ -425,12 +436,39 @@ static int write_frame(AlacEncodeContext *s, AVPacket *avpkt,
for (j = s->frame_size - 1; j > 0; j--)
s->predictor_buf[j] -= s->predictor_buf[j - 1];
}
-
alac_entropy_coder(s);
}
}
- put_bits(pb, 3, 7);
+}
+
+static int write_frame(AlacEncodeContext *s, AVPacket *avpkt,
+ uint8_t * const *samples)
+{
+ PutBitContext *pb = &s->pbctx;
+ const enum AlacRawDataBlockType *ch_elements = ff_alac_channel_elements[s->avctx->channels - 1];
+ const uint8_t *ch_map = ff_alac_channel_layout_offsets[s->avctx->channels - 1];
+ int ch, element, sce, cpe;
+
+ init_put_bits(pb, avpkt->data, avpkt->size);
+
+ ch = element = sce = cpe = 0;
+ while (ch < s->avctx->channels) {
+ if (ch_elements[element] == TYPE_CPE) {
+ write_element(s, TYPE_CPE, cpe, samples[ch_map[ch]],
+ samples[ch_map[ch + 1]]);
+ cpe++;
+ ch += 2;
+ } else {
+ write_element(s, TYPE_SCE, sce, samples[ch_map[ch]], NULL);
+ sce++;
+ ch++;
+ }
+ element++;
+ }
+
+ put_bits(pb, 3, TYPE_END);
flush_put_bits(pb);
+
return put_bits_count(pb) >> 3;
}
@@ -458,14 +496,6 @@ static av_cold int alac_encode_init(AVCodecContext *avctx)
avctx->frame_size = s->frame_size = DEFAULT_FRAME_SIZE;
- /* TODO: Correctly implement multi-channel ALAC.
- It is similar to multi-channel AAC, in that it has a series of
- single-channel (SCE), channel-pair (CPE), and LFE elements. */
- if (avctx->channels > 2) {
- av_log(avctx, AV_LOG_ERROR, "only mono or stereo input is currently supported\n");
- return AVERROR_PATCHWELCOME;
- }
-
if (avctx->sample_fmt == AV_SAMPLE_FMT_S32P) {
if (avctx->bits_per_raw_sample != 24)
av_log(avctx, AV_LOG_WARNING, "encoding as 24 bits-per-sample\n");
@@ -597,8 +627,6 @@ static int alac_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
s->verbatim = 1;
s->extra_bits = 0;
}
- s->write_sample_size = avctx->bits_per_raw_sample - s->extra_bits +
- avctx->channels - 1;
out_bytes = write_frame(s, avpkt, frame->extended_data);
@@ -606,7 +634,6 @@ static int alac_encode_frame(AVCodecContext *avctx, AVPacket *avpkt,
/* frame too large. use verbatim mode */
s->verbatim = 1;
s->extra_bits = 0;
- s->write_sample_size = avctx->bits_per_raw_sample + avctx->channels - 1;
out_bytes = write_frame(s, avpkt, frame->extended_data);
}
@@ -624,6 +651,7 @@ AVCodec ff_alac_encoder = {
.encode2 = alac_encode_frame,
.close = alac_encode_close,
.capabilities = CODEC_CAP_SMALL_LAST_FRAME,
+ .channel_layouts = ff_alac_channel_layouts,
.sample_fmts = (const enum AVSampleFormat[]){ AV_SAMPLE_FMT_S32P,
AV_SAMPLE_FMT_S16P,
AV_SAMPLE_FMT_NONE },
More information about the ffmpeg-cvslog
mailing list