*/
/**
- * @file libavcodec/alsdec.c
+ * @file
* MPEG-4 ALS decoder
* @author Thilo Borgmann <thilo.borgmann _at_ googlemail.com>
*/
#include "unary.h"
#include "mpeg4audio.h"
#include "bytestream.h"
+#include "bgmc.h"
+#include "dsputil.h"
+#include "libavutil/crc.h"
#include <stdint.h>
/** Rice parameters and corresponding index offsets for decoding the
- * indices of scaled PARCOR values. The table choosen is set globally
+ * indices of scaled PARCOR values. The table chosen is set globally
* by the encoder and stored in ALSSpecificConfig.
*/
static const int8_t parcor_rice_table[3][20][2] = {
};
+/** Tail codes used in arithmetic coding using block Gilbert-Moore codes.
+ */
+static const uint8_t tail_code[16][6] = {
+ { 74, 44, 25, 13, 7, 3},
+ { 68, 42, 24, 13, 7, 3},
+ { 58, 39, 23, 13, 7, 3},
+ {126, 70, 37, 19, 10, 5},
+ {132, 70, 37, 20, 10, 5},
+ {124, 70, 38, 20, 10, 5},
+ {120, 69, 37, 20, 11, 5},
+ {116, 67, 37, 20, 11, 5},
+ {108, 66, 36, 20, 10, 5},
+ {102, 62, 36, 20, 10, 5},
+ { 88, 58, 34, 19, 10, 5},
+ {162, 89, 49, 25, 13, 7},
+ {156, 87, 49, 26, 14, 7},
+ {150, 86, 47, 26, 14, 7},
+ {142, 84, 47, 26, 14, 7},
+ {131, 79, 46, 26, 14, 7}
+};
+
+
enum RA_Flag {
RA_FLAG_NONE,
RA_FLAG_FRAMES,
uint32_t samples; ///< number of samples, 0xFFFFFFFF if unknown
int resolution; ///< 000 = 8-bit; 001 = 16-bit; 010 = 24-bit; 011 = 32-bit
int floating; ///< 1 = IEEE 32-bit floating-point, 0 = integer
+ int msb_first; ///< 1 = original CRC calculated on big-endian system, 0 = little-endian
int frame_length; ///< frame length for each frame (last frame may differ)
int ra_distance; ///< distance between RA frames (in frames, 0...255)
enum RA_Flag ra_flag; ///< indicates where the size of ra units is stored
int rlslms; ///< use "Recursive Least Square-Least Mean Square" predictor: 1 = on, 0 = off
int chan_config_info; ///< mapping of channels to loudspeaker locations. Unused until setting channel configuration is implemented.
int *chan_pos; ///< original channel positions
- uint32_t header_size; ///< header size of original audio file in bytes, provided for debugging
- uint32_t trailer_size; ///< trailer size of original audio file in bytes, provided for debugging
+ int crc_enabled; ///< enable Cyclic Redundancy Checksum
} ALSSpecificConfig;
AVCodecContext *avctx;
ALSSpecificConfig sconf;
GetBitContext gb;
+ DSPContext dsp;
+ const AVCRC *crc_table;
+ uint32_t crc_org; ///< CRC value of the original input data
+ uint32_t crc; ///< CRC value calculated from decoded data
unsigned int cur_frame_length; ///< length of the current frame to decode
unsigned int frame_id; ///< the frame ID / number of the current frame
unsigned int js_switch; ///< if true, joint-stereo decoding is enforced
unsigned int num_blocks; ///< number of blocks used in the current frame
+ unsigned int s_max; ///< maximum Rice parameter allowed in entropy coding
+ uint8_t *bgmc_lut; ///< pointer at lookup tables used for BGMC
+ unsigned int *bgmc_lut_status; ///< pointer at lookup table status flags used for BGMC
int ltp_lag_length; ///< number of bits used for ltp lag value
int *use_ltp; ///< contains use_ltp flags for all channels
int *ltp_lag; ///< contains ltp lag values for all channels
int32_t *prev_raw_samples; ///< contains unshifted raw samples from the previous block
int32_t **raw_samples; ///< decoded raw samples for each channel
int32_t *raw_buffer; ///< contains all decoded raw samples including carryover samples
+ uint8_t *crc_buffer; ///< buffer of byte order corrected samples used for CRC check
} ALSDecContext;
dprintf(avctx, "chan_sort = %i\n", sconf->chan_sort);
dprintf(avctx, "RLSLMS = %i\n", sconf->rlslms);
dprintf(avctx, "chan_config_info = %i\n", sconf->chan_config_info);
- dprintf(avctx, "header_size = %i\n", sconf->header_size);
- dprintf(avctx, "trailer_size = %i\n", sconf->trailer_size);
#endif
}
-/** Reads an ALSSpecificConfig from a buffer into the output struct.
+/** Read an ALSSpecificConfig from a buffer into the output struct.
*/
static av_cold int read_specific_config(ALSDecContext *ctx)
{
GetBitContext gb;
uint64_t ht_size;
- int i, config_offset, crc_enabled;
+ int i, config_offset;
MPEG4AudioConfig m4ac;
ALSSpecificConfig *sconf = &ctx->sconf;
AVCodecContext *avctx = ctx->avctx;
- uint32_t als_id;
+ uint32_t als_id, header_size, trailer_size;
init_get_bits(&gb, avctx->extradata, avctx->extradata_size * 8);
skip_bits(&gb, 3); // skip file_type
sconf->resolution = get_bits(&gb, 3);
sconf->floating = get_bits1(&gb);
- skip_bits1(&gb); // skip msb_first
+ sconf->msb_first = get_bits1(&gb);
sconf->frame_length = get_bits(&gb, 16) + 1;
sconf->ra_distance = get_bits(&gb, 8);
sconf->ra_flag = get_bits(&gb, 2);
sconf->mc_coding = get_bits1(&gb);
sconf->chan_config = get_bits1(&gb);
sconf->chan_sort = get_bits1(&gb);
- crc_enabled = get_bits1(&gb);
+ sconf->crc_enabled = get_bits1(&gb);
sconf->rlslms = get_bits1(&gb);
skip_bits(&gb, 5); // skip 5 reserved bits
skip_bits1(&gb); // skip aux_data_enabled
if (get_bits_left(&gb) < 64)
return -1;
- sconf->header_size = get_bits_long(&gb, 32);
- sconf->trailer_size = get_bits_long(&gb, 32);
- if (sconf->header_size == 0xFFFFFFFF)
- sconf->header_size = 0;
- if (sconf->trailer_size == 0xFFFFFFFF)
- sconf->trailer_size = 0;
+ header_size = get_bits_long(&gb, 32);
+ trailer_size = get_bits_long(&gb, 32);
+ if (header_size == 0xFFFFFFFF)
+ header_size = 0;
+ if (trailer_size == 0xFFFFFFFF)
+ trailer_size = 0;
- ht_size = ((int64_t)(sconf->header_size) + (int64_t)(sconf->trailer_size)) << 3;
+ ht_size = ((int64_t)(header_size) + (int64_t)(trailer_size)) << 3;
// skip the header and trailer data
skip_bits_long(&gb, ht_size);
- // skip the crc data
- if (crc_enabled) {
+ // initialize CRC calculation
+ if (sconf->crc_enabled) {
if (get_bits_left(&gb) < 32)
return -1;
- skip_bits_long(&gb, 32);
+ if (avctx->error_recognition >= FF_ER_CAREFUL) {
+ ctx->crc_table = av_crc_get_table(AV_CRC_32_IEEE_LE);
+ ctx->crc = 0xFFFFFFFF;
+ ctx->crc_org = ~get_bits_long(&gb, 32);
+ } else
+ skip_bits_long(&gb, 32);
}
}
-/** Checks the ALSSpecificConfig for unsupported features.
+/** Check the ALSSpecificConfig for unsupported features.
*/
static int check_specific_config(ALSDecContext *ctx)
{
}
MISSING_ERR(sconf->floating, "Floating point decoding", -1);
- MISSING_ERR(sconf->bgmc, "BGMC entropy decoding", -1);
MISSING_ERR(sconf->rlslms, "Adaptive RLS-LMS prediction", -1);
MISSING_ERR(sconf->chan_sort, "Channel sorting", 0);
}
-/** Parses the bs_info field to extract the block partitioning used in
+/** Parse the bs_info field to extract the block partitioning used in
* block switching mode, refer to ISO/IEC 14496-3, section 11.6.2.
*/
static void parse_bs_info(const uint32_t bs_info, unsigned int n,
}
-/** Reads and decodes a Rice codeword.
+/** Read and decode a Rice codeword.
*/
static int32_t decode_rice(GetBitContext *gb, unsigned int k)
{
}
-/** Converts PARCOR coefficient k to direct filter coefficient.
+/** Convert PARCOR coefficient k to direct filter coefficient.
*/
static void parcor_to_lpc(unsigned int k, const int32_t *par, int32_t *cof)
{
}
-/** Reads block switching field if necessary and sets actual block sizes.
- * Also assures that the block sizes of the last frame correspond to the
+/** Read block switching field if necessary and set actual block sizes.
+ * Also assure that the block sizes of the last frame correspond to the
* actual number of samples.
*/
static void get_block_sizes(ALSDecContext *ctx, unsigned int *div_blocks,
unsigned int remaining = ctx->cur_frame_length;
for (b = 0; b < ctx->num_blocks; b++) {
- if (remaining < div_blocks[b]) {
+ if (remaining <= div_blocks[b]) {
div_blocks[b] = remaining;
ctx->num_blocks = b + 1;
break;
}
-/** Reads the block data for a constant block
+/** Read the block data for a constant block
*/
static void read_const_block_data(ALSDecContext *ctx, ALSBlockData *bd)
{
}
-/** Decodes the block data for a constant block
+/** Decode the block data for a constant block
*/
static void decode_const_block_data(ALSDecContext *ctx, ALSBlockData *bd)
{
}
-/** Reads the block data for a non-constant block
+/** Read the block data for a non-constant block
*/
static int read_var_block_data(ALSDecContext *ctx, ALSBlockData *bd)
{
GetBitContext *gb = &ctx->gb;
unsigned int k;
unsigned int s[8];
+ unsigned int sx[8];
unsigned int sub_blocks, log2_sub_blocks, sb_length;
unsigned int start = 0;
unsigned int opt_order;
int sb;
int32_t *quant_cof = bd->quant_cof;
+ int32_t *current_res;
// ensure variable block decoding by reusing this field
sb_length = bd->block_length >> log2_sub_blocks;
-
if (sconf->bgmc) {
- // TODO: BGMC mode
+ s[0] = get_bits(gb, 8 + (sconf->resolution > 1));
+ for (k = 1; k < sub_blocks; k++)
+ s[k] = s[k - 1] + decode_rice(gb, 2);
+
+ for (k = 0; k < sub_blocks; k++) {
+ sx[k] = s[k] & 0x0F;
+ s [k] >>= 4;
+ }
} else {
s[0] = get_bits(gb, 4 + (sconf->resolution > 1));
for (k = 1; k < sub_blocks; k++)
*bd->use_ltp = get_bits1(gb);
if (*bd->use_ltp) {
+ int r, c;
+
bd->ltp_gain[0] = decode_rice(gb, 1) << 3;
bd->ltp_gain[1] = decode_rice(gb, 2) << 3;
- bd->ltp_gain[2] = ltp_gain_values[get_unary(gb, 0, 4)][get_bits(gb, 2)];
+ r = get_unary(gb, 0, 4);
+ c = get_bits(gb, 2);
+ bd->ltp_gain[2] = ltp_gain_values[r][c];
bd->ltp_gain[3] = decode_rice(gb, 2) << 3;
bd->ltp_gain[4] = decode_rice(gb, 1) << 3;
if (opt_order)
bd->raw_samples[0] = decode_rice(gb, avctx->bits_per_raw_sample - 4);
if (opt_order > 1)
- bd->raw_samples[1] = decode_rice(gb, s[0] + 3);
+ bd->raw_samples[1] = decode_rice(gb, FFMIN(s[0] + 3, ctx->s_max));
if (opt_order > 2)
- bd->raw_samples[2] = decode_rice(gb, s[0] + 1);
+ bd->raw_samples[2] = decode_rice(gb, FFMIN(s[0] + 1, ctx->s_max));
start = FFMIN(opt_order, 3);
}
// read all residuals
if (sconf->bgmc) {
- // TODO: BGMC mode
+ unsigned int delta[8];
+ unsigned int k [8];
+ unsigned int b = av_clip((av_ceil_log2(bd->block_length) - 3) >> 1, 0, 5);
+ unsigned int i = start;
+
+ // read most significant bits
+ unsigned int high;
+ unsigned int low;
+ unsigned int value;
+
+ ff_bgmc_decode_init(gb, &high, &low, &value);
+
+ current_res = bd->raw_samples + start;
+
+ for (sb = 0; sb < sub_blocks; sb++, i = 0) {
+ k [sb] = s[sb] > b ? s[sb] - b : 0;
+ delta[sb] = 5 - s[sb] + k[sb];
+
+ ff_bgmc_decode(gb, sb_length, current_res,
+ delta[sb], sx[sb], &high, &low, &value, ctx->bgmc_lut, ctx->bgmc_lut_status);
+
+ current_res += sb_length;
+ }
+
+ ff_bgmc_decode_end(gb);
+
+
+ // read least significant bits and tails
+ i = start;
+ current_res = bd->raw_samples + start;
+
+ for (sb = 0; sb < sub_blocks; sb++, i = 0) {
+ unsigned int cur_tail_code = tail_code[sx[sb]][delta[sb]];
+ unsigned int cur_k = k[sb];
+ unsigned int cur_s = s[sb];
+
+ for (; i < sb_length; i++) {
+ int32_t res = *current_res;
+
+ if (res == cur_tail_code) {
+ unsigned int max_msb = (2 + (sx[sb] > 2) + (sx[sb] > 10))
+ << (5 - delta[sb]);
+
+ res = decode_rice(gb, cur_s);
+
+ if (res >= 0) {
+ res += (max_msb ) << cur_k;
+ } else {
+ res -= (max_msb - 1) << cur_k;
+ }
+ } else {
+ if (res > cur_tail_code)
+ res--;
+
+ if (res & 1)
+ res = -res;
+
+ res >>= 1;
+
+ if (cur_k) {
+ res <<= cur_k;
+ res |= get_bits_long(gb, cur_k);
+ }
+ }
+
+ *current_res++ = res;
+ }
+ }
} else {
- int32_t *current_res = bd->raw_samples + start;
+ current_res = bd->raw_samples + start;
for (sb = 0; sb < sub_blocks; sb++, start = 0)
for (; start < sb_length; start++)
}
-/** Decodes the block data for a non-constant block
+/** Decode the block data for a non-constant block
*/
static int decode_var_block_data(ALSDecContext *ctx, ALSBlockData *bd)
{
}
-/** Reads the block data.
+/** Read the block data.
*/
static int read_block(ALSDecContext *ctx, ALSBlockData *bd)
{
}
-/** Decodes the block data.
+/** Decode the block data.
*/
static int decode_block(ALSDecContext *ctx, ALSBlockData *bd)
{
}
-/** Reads and decodes block data successively.
+/** Read and decode block data successively.
*/
static int read_decode_block(ALSDecContext *ctx, ALSBlockData *bd)
{
}
-/** Computes the number of samples left to decode for the current frame and
+/** Compute the number of samples left to decode for the current frame and
* sets these samples to zero.
*/
static void zero_remaining(unsigned int b, unsigned int b_max,
}
-/** Decodes blocks independently.
+/** Decode blocks independently.
*/
static int decode_blocks_ind(ALSDecContext *ctx, unsigned int ra_frame,
unsigned int c, const unsigned int *div_blocks,
}
-/** Decodes blocks dependently.
+/** Decode blocks dependently.
*/
static int decode_blocks(ALSDecContext *ctx, unsigned int ra_frame,
unsigned int c, const unsigned int *div_blocks,
}
-/** Reads the channel data.
+/** Read the channel data.
*/
static int read_channel_data(ALSDecContext *ctx, ALSChannelData *cd, int c)
{
}
-/** Reads the frame data.
+/** Read the frame data.
*/
static int read_frame_data(ALSDecContext *ctx, unsigned int ra_frame)
{
}
-/** Decodes an ALS frame.
+/** Decode an ALS frame.
*/
static int decode_frame(AVCodecContext *avctx,
void *data, int *data_size,
INTERLEAVE_OUTPUT(32)
}
+ // update CRC
+ if (sconf->crc_enabled && avctx->error_recognition >= FF_ER_CAREFUL) {
+ int swap = HAVE_BIGENDIAN != sconf->msb_first;
+
+ if (ctx->avctx->bits_per_raw_sample == 24) {
+ int32_t *src = data;
+
+ for (sample = 0;
+ sample < ctx->cur_frame_length * avctx->channels;
+ sample++) {
+ int32_t v;
+
+ if (swap)
+ v = av_bswap32(src[sample]);
+ else
+ v = src[sample];
+ if (!HAVE_BIGENDIAN)
+ v >>= 8;
+
+ ctx->crc = av_crc(ctx->crc_table, ctx->crc, (uint8_t*)(&v), 3);
+ }
+ } else {
+ uint8_t *crc_source;
+
+ if (swap) {
+ if (ctx->avctx->bits_per_raw_sample <= 16) {
+ int16_t *src = (int16_t*) data;
+ int16_t *dest = (int16_t*) ctx->crc_buffer;
+ for (sample = 0;
+ sample < ctx->cur_frame_length * avctx->channels;
+ sample++)
+ *dest++ = av_bswap16(src[sample]);
+ } else {
+ ctx->dsp.bswap_buf((uint32_t*)ctx->crc_buffer, data,
+ ctx->cur_frame_length * avctx->channels);
+ }
+ crc_source = ctx->crc_buffer;
+ } else {
+ crc_source = data;
+ }
+
+ ctx->crc = av_crc(ctx->crc_table, ctx->crc, crc_source, size);
+ }
+
+
+ // check CRC sums if this is the last frame
+ if (ctx->cur_frame_length != sconf->frame_length &&
+ ctx->crc_org != ctx->crc) {
+ av_log(avctx, AV_LOG_ERROR, "CRC error.\n");
+ }
+ }
+
+
bytes_read = invalid_frame ? buffer_size :
(get_bits_count(&ctx->gb) + 7) >> 3;
}
-/** Uninitializes the ALS decoder.
+/** Uninitialize the ALS decoder.
*/
static av_cold int decode_end(AVCodecContext *avctx)
{
av_freep(&ctx->sconf.chan_pos);
+ ff_bgmc_end(&ctx->bgmc_lut, &ctx->bgmc_lut_status);
+
av_freep(&ctx->use_ltp);
av_freep(&ctx->ltp_lag);
av_freep(&ctx->ltp_gain);
}
-/** Initializes the ALS decoder.
+/** Initialize the ALS decoder.
*/
static av_cold int decode_init(AVCodecContext *avctx)
{
return -1;
}
+ if (sconf->bgmc)
+ ff_bgmc_init(avctx, &ctx->bgmc_lut, &ctx->bgmc_lut_status);
+
if (sconf->floating) {
avctx->sample_fmt = SAMPLE_FMT_FLT;
avctx->bits_per_raw_sample = 32;
avctx->bits_per_raw_sample = (sconf->resolution + 1) * 8;
}
+ // set maximum Rice parameter for progressive decoding based on resolution
+ // This is not specified in 14496-3 but actually done by the reference
+ // codec RM22 revision 2.
+ ctx->s_max = sconf->resolution > 1 ? 31 : 15;
+
// set lag value for long-term prediction
ctx->ltp_lag_length = 8 + (avctx->sample_rate >= 96000) +
(avctx->sample_rate >= 192000);
ctx->lpc_cof_reversed_buffer = av_malloc(sizeof(*ctx->lpc_cof_buffer) *
sconf->max_order);
- if (!ctx->quant_cof || !ctx->lpc_cof ||
- !ctx->quant_cof_buffer || !ctx->lpc_cof_buffer ||
+ if (!ctx->quant_cof || !ctx->lpc_cof ||
+ !ctx->quant_cof_buffer || !ctx->lpc_cof_buffer ||
!ctx->lpc_cof_reversed_buffer) {
av_log(avctx, AV_LOG_ERROR, "Allocating buffer memory failed.\n");
return AVERROR(ENOMEM);
// allocate and assign channel data buffer for mcc mode
if (sconf->mc_coding) {
ctx->chan_data_buffer = av_malloc(sizeof(*ctx->chan_data_buffer) *
- num_buffers);
- ctx->chan_data = av_malloc(sizeof(ALSChannelData) *
+ num_buffers * num_buffers);
+ ctx->chan_data = av_malloc(sizeof(*ctx->chan_data) *
num_buffers);
ctx->reverted_channels = av_malloc(sizeof(*ctx->reverted_channels) *
num_buffers);
}
for (c = 0; c < num_buffers; c++)
- ctx->chan_data[c] = ctx->chan_data_buffer + c;
+ ctx->chan_data[c] = ctx->chan_data_buffer + c * num_buffers;
} else {
ctx->chan_data = NULL;
ctx->chan_data_buffer = NULL;
for (c = 1; c < avctx->channels; c++)
ctx->raw_samples[c] = ctx->raw_samples[c - 1] + channel_size;
+ // allocate crc buffer
+ if (HAVE_BIGENDIAN != sconf->msb_first && sconf->crc_enabled &&
+ avctx->error_recognition >= FF_ER_CAREFUL) {
+ ctx->crc_buffer = av_malloc(sizeof(*ctx->crc_buffer) *
+ ctx->cur_frame_length *
+ avctx->channels *
+ (av_get_bits_per_sample_format(avctx->sample_fmt) >> 3));
+ if (!ctx->crc_buffer) {
+ av_log(avctx, AV_LOG_ERROR, "Allocating buffer memory failed.\n");
+ decode_end(avctx);
+ return AVERROR(ENOMEM);
+ }
+ }
+
+ dsputil_init(&ctx->dsp, avctx);
+
return 0;
}
-/** Flushes (resets) the frame ID after seeking.
+/** Flush (reset) the frame ID after seeking.
*/
static av_cold void flush(AVCodecContext *avctx)
{
AVCodec als_decoder = {
"als",
- CODEC_TYPE_AUDIO,
+ AVMEDIA_TYPE_AUDIO,
CODEC_ID_MP4ALS,
sizeof(ALSDecContext),
decode_init,