1
0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-12-23 12:43:46 +02:00

libavcodec: Implementation of AC3 fixedpoint decoder

Signed-off-by: Nedeljko Babic <nbabic@mips.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
This commit is contained in:
Nedeljko Babic 2014-04-01 16:31:08 +02:00 committed by Michael Niedermayer
parent d506deaeaa
commit 696e34a6e1
15 changed files with 483 additions and 109 deletions

View File

@ -2,6 +2,7 @@ Entries are sorted chronologically from oldest to youngest within each release,
releases are sorted from youngest to oldest.
version <next>:
- added AC3 fixed-point decoding
- shuffleplanes filter
- subfile protocol
- Phantom Cine demuxer

1
configure vendored
View File

@ -1951,6 +1951,7 @@ aac_decoder_select="mdct sinewin"
aac_encoder_select="audio_frame_queue mdct sinewin"
aac_latm_decoder_select="aac_decoder aac_latm_parser"
ac3_decoder_select="mdct ac3dsp ac3_parser dsputil"
ac3_fixed_decoder_select="mdct ac3dsp ac3_parser dsputil"
ac3_encoder_select="mdct ac3dsp dsputil"
ac3_fixed_encoder_select="mdct ac3dsp dsputil"
aic_decoder_select="dsputil golomb"

View File

@ -820,7 +820,7 @@ following image formats are supported:
@tab encoding supported through external library libaacplus
@item AAC @tab E @tab X
@tab encoding supported through external library libfaac and libvo-aacenc
@item AC-3 @tab IX @tab X
@item AC-3 @tab IX @tab IX
@item ADPCM 4X Movie @tab @tab X
@item ADPCM CDROM XA @tab @tab X
@item ADPCM Creative Technology @tab @tab X

View File

@ -94,7 +94,8 @@ OBJS-$(CONFIG_AAC_ENCODER) += aacenc.o aaccoder.o \
psymodel.o iirfilter.o \
mpeg4audio.o kbdwin.o
OBJS-$(CONFIG_AASC_DECODER) += aasc.o msrledec.o
OBJS-$(CONFIG_AC3_DECODER) += ac3dec.o ac3dec_data.o ac3.o kbdwin.o
OBJS-$(CONFIG_AC3_DECODER) += ac3dec_float.o ac3dec_data.o ac3.o kbdwin.o
OBJS-$(CONFIG_AC3_FIXED_DECODER) += ac3dec_fixed.o ac3dec_data.o ac3.o kbdwin.o
OBJS-$(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

View File

@ -51,6 +51,52 @@
#define EXP_D25 2
#define EXP_D45 3
#ifndef CONFIG_AC3_FIXED
#define CONFIG_AC3_FIXED 0
#endif
#if CONFIG_AC3_FIXED
#define FFT_FLOAT 0
#define FIXR(a) ((int)((a) * 0 + 0.5))
#define FIXR12(a) ((int)((a) * 4096 + 0.5))
#define FIXR15(a) ((int)((a) * 32768 + 0.5))
#define ROUND15(x) ((x) + 16384) >> 15
#define AC3_RENAME(x) x ## _fixed
#define AC3_NORM(norm) (1<<24)/(norm)
#define AC3_MUL(a,b) ((((int64_t) (a)) * (b))>>12)
#define AC3_RANGE(x) (x)
#define AC3_DYNAMIC_RANGE(x) (x)
#define AC3_SPX_BLEND(x) (x)
#define AC3_DYNAMIC_RANGE1 0
#define INTFLOAT int
#define SHORTFLOAT int16_t
#else /* CONFIG_AC3_FIXED */
#define FIXR(x) ((float)(x))
#define FIXR12(x) ((float)(x))
#define FIXR15(x) ((float)(x))
#define ROUND15(x) (x)
#define AC3_RENAME(x) x
#define AC3_NORM(norm) (1.0f/(norm))
#define AC3_MUL(a,b) ((a) * (b))
#define AC3_RANGE(x) (dynamic_range_tab[(x)])
#define AC3_DYNAMIC_RANGE(x) (powf(x, s->drc_scale))
#define AC3_SPX_BLEND(x) (x)* (1.0f/32)
#define AC3_DYNAMIC_RANGE1 1.0f
#define INTFLOAT float
#define SHORTFLOAT float
#endif /* CONFIG_AC3_FIXED */
#define AC3_LEVEL(x) ROUND15((x) * FIXR15(0.7071067811865476))
/* pre-defined gain values */
#define LEVEL_PLUS_3DB 1.4142135623730950
#define LEVEL_PLUS_1POINT5DB 1.1892071150027209

View File

@ -179,13 +179,22 @@ static av_cold int ac3_decode_init(AVCodecContext *avctx)
ac3_tables_init();
ff_mdct_init(&s->imdct_256, 8, 1, 1.0);
ff_mdct_init(&s->imdct_512, 9, 1, 1.0);
ff_kbd_window_init(s->window, 5.0, 256);
AC3_RENAME(ff_kbd_window_init)(s->window, 5.0, 256);
ff_dsputil_init(&s->dsp, avctx);
#if (CONFIG_AC3_FIXED)
s->fdsp = avpriv_alloc_fixed_dsp(avctx->flags & CODEC_FLAG_BITEXACT);
#else
avpriv_float_dsp_init(&s->fdsp, avctx->flags & CODEC_FLAG_BITEXACT);
#endif
ff_ac3dsp_init(&s->ac3dsp, avctx->flags & CODEC_FLAG_BITEXACT);
ff_fmt_convert_init(&s->fmt_conv, avctx);
av_lfg_init(&s->dith_state, 0);
if (CONFIG_AC3_FIXED)
avctx->sample_fmt = AV_SAMPLE_FMT_S16P;
else
avctx->sample_fmt = AV_SAMPLE_FMT_FLTP;
/* allow downmixing to stereo or mono */
@ -345,40 +354,45 @@ static void set_downmix_coeffs(AC3DecodeContext *s)
float cmix = gain_levels[s-> center_mix_level];
float smix = gain_levels[s->surround_mix_level];
float norm0, norm1;
float downmix_coeffs[AC3_MAX_CHANNELS][2];
for (i = 0; i < s->fbw_channels; i++) {
s->downmix_coeffs[i][0] = gain_levels[ac3_default_coeffs[s->channel_mode][i][0]];
s->downmix_coeffs[i][1] = gain_levels[ac3_default_coeffs[s->channel_mode][i][1]];
downmix_coeffs[i][0] = gain_levels[ac3_default_coeffs[s->channel_mode][i][0]];
downmix_coeffs[i][1] = gain_levels[ac3_default_coeffs[s->channel_mode][i][1]];
}
if (s->channel_mode > 1 && s->channel_mode & 1) {
s->downmix_coeffs[1][0] = s->downmix_coeffs[1][1] = cmix;
downmix_coeffs[1][0] = downmix_coeffs[1][1] = cmix;
}
if (s->channel_mode == AC3_CHMODE_2F1R || s->channel_mode == AC3_CHMODE_3F1R) {
int nf = s->channel_mode - 2;
s->downmix_coeffs[nf][0] = s->downmix_coeffs[nf][1] = smix * LEVEL_MINUS_3DB;
downmix_coeffs[nf][0] = downmix_coeffs[nf][1] = smix * LEVEL_MINUS_3DB;
}
if (s->channel_mode == AC3_CHMODE_2F2R || s->channel_mode == AC3_CHMODE_3F2R) {
int nf = s->channel_mode - 4;
s->downmix_coeffs[nf][0] = s->downmix_coeffs[nf+1][1] = smix;
downmix_coeffs[nf][0] = downmix_coeffs[nf+1][1] = smix;
}
/* renormalize */
norm0 = norm1 = 0.0;
for (i = 0; i < s->fbw_channels; i++) {
norm0 += s->downmix_coeffs[i][0];
norm1 += s->downmix_coeffs[i][1];
norm0 += downmix_coeffs[i][0];
norm1 += downmix_coeffs[i][1];
}
norm0 = 1.0f / norm0;
norm1 = 1.0f / norm1;
for (i = 0; i < s->fbw_channels; i++) {
s->downmix_coeffs[i][0] *= norm0;
s->downmix_coeffs[i][1] *= norm1;
downmix_coeffs[i][0] *= norm0;
downmix_coeffs[i][1] *= norm1;
}
if (s->output_mode == AC3_CHMODE_MONO) {
for (i = 0; i < s->fbw_channels; i++)
s->downmix_coeffs[i][0] = (s->downmix_coeffs[i][0] +
s->downmix_coeffs[i][1]) * LEVEL_MINUS_3DB;
downmix_coeffs[i][0] = (downmix_coeffs[i][0] +
downmix_coeffs[i][1]) * LEVEL_MINUS_3DB;
}
for (i = 0; i < s->fbw_channels; i++) {
s->downmix_coeffs[i][0] = FIXR12(downmix_coeffs[i][0]);
s->downmix_coeffs[i][1] = FIXR12(downmix_coeffs[i][1]);
}
}
@ -646,20 +660,30 @@ static inline void do_imdct(AC3DecodeContext *s, int channels)
for (ch = 1; ch <= channels; ch++) {
if (s->block_switch[ch]) {
int i;
float *x = s->tmp_output + 128;
FFTSample *x = s->tmp_output + 128;
for (i = 0; i < 128; i++)
x[i] = s->transform_coeffs[ch][2 * i];
s->imdct_256.imdct_half(&s->imdct_256, s->tmp_output, x);
#if CONFIG_AC3_FIXED
s->fdsp->vector_fmul_window_scaled(s->outptr[ch - 1], s->delay[ch - 1],
s->tmp_output, s->window, 128, 8);
#else
s->fdsp.vector_fmul_window(s->outptr[ch - 1], s->delay[ch - 1],
s->tmp_output, s->window, 128);
#endif
for (i = 0; i < 128; i++)
x[i] = s->transform_coeffs[ch][2 * i + 1];
s->imdct_256.imdct_half(&s->imdct_256, s->delay[ch - 1], x);
} else {
s->imdct_512.imdct_half(&s->imdct_512, s->tmp_output, s->transform_coeffs[ch]);
#if CONFIG_AC3_FIXED
s->fdsp->vector_fmul_window_scaled(s->outptr[ch - 1], s->delay[ch - 1],
s->tmp_output, s->window, 128, 8);
#else
s->fdsp.vector_fmul_window(s->outptr[ch - 1], s->delay[ch - 1],
s->tmp_output, s->window, 128);
memcpy(s->delay[ch - 1], s->tmp_output + 128, 128 * sizeof(float));
#endif
memcpy(s->delay[ch - 1], s->tmp_output + 128, 128 * sizeof(FFTSample));
}
}
}
@ -794,13 +818,13 @@ static int decode_audio_block(AC3DecodeContext *s, int blk)
if (get_bits1(gbc)) {
/* Allow asymmetric application of DRC when drc_scale > 1.
Amplification of quiet sounds is enhanced */
float range = dynamic_range_tab[get_bits(gbc, 8)];
INTFLOAT range = AC3_RANGE(get_bits(gbc, 8));
if (range > 1.0 || s->drc_scale <= 1.0)
s->dynamic_range[i] = powf(range, s->drc_scale);
s->dynamic_range[i] = AC3_DYNAMIC_RANGE(range);
else
s->dynamic_range[i] = range;
} else if (blk == 0) {
s->dynamic_range[i] = 1.0f;
s->dynamic_range[i] = AC3_DYNAMIC_RANGE1;
}
} while (i--);
@ -826,6 +850,9 @@ static int decode_audio_block(AC3DecodeContext *s, int blk)
if (start_subband > 7)
start_subband += start_subband - 7;
end_subband = get_bits(gbc, 3) + 5;
#if CONFIG_AC3_FIXED
s->spx_dst_end_freq = end_freq_inv_tab[end_subband];
#endif
if (end_subband > 7)
end_subband += end_subband - 7;
dst_start_freq = dst_start_freq * 12 + 25;
@ -846,7 +873,9 @@ static int decode_audio_block(AC3DecodeContext *s, int blk)
s->spx_dst_start_freq = dst_start_freq;
s->spx_src_start_freq = src_start_freq;
#if !CONFIG_AC3_FIXED
s->spx_dst_end_freq = dst_end_freq;
#endif
decode_band_structure(gbc, blk, s->eac3, 0,
start_subband, end_subband,
@ -866,18 +895,40 @@ static int decode_audio_block(AC3DecodeContext *s, int blk)
for (ch = 1; ch <= fbw_channels; ch++) {
if (s->channel_uses_spx[ch]) {
if (s->first_spx_coords[ch] || get_bits1(gbc)) {
float spx_blend;
INTFLOAT spx_blend;
int bin, master_spx_coord;
s->first_spx_coords[ch] = 0;
spx_blend = get_bits(gbc, 5) * (1.0f/32);
spx_blend = AC3_SPX_BLEND(get_bits(gbc, 5));
master_spx_coord = get_bits(gbc, 2) * 3;
bin = s->spx_src_start_freq;
for (bnd = 0; bnd < s->num_spx_bands; bnd++) {
int bandsize;
int spx_coord_exp, spx_coord_mant;
float nratio, sblend, nblend, spx_coord;
INTFLOAT nratio, sblend, nblend;
#if CONFIG_AC3_FIXED
int64_t accu;
/* calculate blending factors */
bandsize = s->spx_band_sizes[bnd];
accu = (int64_t)((bin << 23) + (bandsize << 22)) * s->spx_dst_end_freq;
nratio = (int)(accu >> 32);
nratio -= spx_blend << 18;
if (nratio < 0) {
nblend = 0;
sblend = 0x800000;
} else if (nratio > 0x7fffff) {
nblend = 0x800000;
sblend = 0;
} else {
nblend = fixed_sqrt(nratio, 23);
accu = (int64_t)nblend * 1859775393;
nblend = (int)((accu + (1<<29)) >> 30);
sblend = fixed_sqrt(0x800000 - nratio, 23);
}
#else
float spx_coord;
/* calculate blending factors */
bandsize = s->spx_band_sizes[bnd];
@ -886,6 +937,7 @@ static int decode_audio_block(AC3DecodeContext *s, int blk)
nblend = sqrtf(3.0f * nratio); // noise is scaled by sqrt(3)
// to give unity variance
sblend = sqrtf(1.0f - nratio);
#endif
bin += bandsize;
/* decode spx coordinates */
@ -894,11 +946,18 @@ static int decode_audio_block(AC3DecodeContext *s, int blk)
if (spx_coord_exp == 15) spx_coord_mant <<= 1;
else spx_coord_mant += 4;
spx_coord_mant <<= (25 - spx_coord_exp - master_spx_coord);
spx_coord = spx_coord_mant * (1.0f / (1 << 23));
/* multiply noise and signal blending factors by spx coordinate */
#if CONFIG_AC3_FIXED
accu = (int64_t)nblend * spx_coord_mant;
s->spx_noise_blend[ch][bnd] = (int)((accu + (1<<22)) >> 23);
accu = (int64_t)sblend * spx_coord_mant;
s->spx_signal_blend[ch][bnd] = (int)((accu + (1<<22)) >> 23);
#else
spx_coord = spx_coord_mant * (1.0f / (1 << 23));
s->spx_noise_blend [ch][bnd] = nblend * spx_coord;
s->spx_signal_blend[ch][bnd] = sblend * spx_coord;
#endif
}
}
} else {
@ -1255,14 +1314,19 @@ static int decode_audio_block(AC3DecodeContext *s, int blk)
/* apply scaling to coefficients (headroom, dynrng) */
for (ch = 1; ch <= s->channels; ch++) {
float gain = 1.0 / 4194304.0f;
if (s->channel_mode == AC3_CHMODE_DUALMONO) {
gain *= s->dynamic_range[2 - ch];
INTFLOAT gain;
if(s->channel_mode == AC3_CHMODE_DUALMONO) {
gain = s->dynamic_range[2-ch];
} else {
gain *= s->dynamic_range[0];
gain = s->dynamic_range[0];
}
#if CONFIG_AC3_FIXED
scale_coefs(s->transform_coeffs[ch], s->fixed_coeffs[ch], gain, 256);
#else
gain *= 1.0 / 4194304.0f;
s->fmt_conv.int32_to_float_fmul_scalar(s->transform_coeffs[ch],
s->fixed_coeffs[ch], gain, 256);
#endif
}
/* apply spectral extension to high frequency bins */
@ -1287,19 +1351,24 @@ static int decode_audio_block(AC3DecodeContext *s, int blk)
do_imdct(s, s->channels);
if (downmix_output) {
#if CONFIG_AC3_FIXED
ac3_downmix_c_fixed16(s->outptr, s->downmix_coeffs,
s->out_channels, s->fbw_channels, 256);
#else
s->ac3dsp.downmix(s->outptr, s->downmix_coeffs,
s->out_channels, s->fbw_channels, 256);
#endif
}
} else {
if (downmix_output) {
s->ac3dsp.downmix(s->xcfptr + 1, s->downmix_coeffs,
s->ac3dsp.AC3_RENAME(downmix)(s->xcfptr + 1, s->downmix_coeffs,
s->out_channels, s->fbw_channels, 256);
}
if (downmix_output && !s->downmixed) {
s->downmixed = 1;
s->ac3dsp.downmix(s->dlyptr, s->downmix_coeffs, s->out_channels,
s->fbw_channels, 128);
s->ac3dsp.AC3_RENAME(downmix)(s->dlyptr, s->downmix_coeffs,
s->out_channels, s->fbw_channels, 128);
}
do_imdct(s, s->out_channels);
@ -1320,7 +1389,7 @@ static int ac3_decode_frame(AVCodecContext * avctx, void *data,
AC3DecodeContext *s = avctx->priv_data;
int blk, ch, err, ret;
const uint8_t *channel_map;
const float *output[AC3_MAX_CHANNELS];
const SHORTFLOAT *output[AC3_MAX_CHANNELS];
enum AVMatrixEncoding matrix_encoding;
AVDownmixInfo *downmix_info;
@ -1447,7 +1516,7 @@ static int ac3_decode_frame(AVCodecContext * avctx, void *data,
}
for (ch = 0; ch < s->channels; ch++) {
if (ch < s->out_channels)
s->outptr[channel_map[ch]] = (float *)frame->data[ch];
s->outptr[channel_map[ch]] = (SHORTFLOAT *)frame->data[ch];
}
for (blk = 0; blk < s->num_blocks; blk++) {
if (!err && decode_audio_block(s, blk)) {
@ -1456,7 +1525,7 @@ static int ac3_decode_frame(AVCodecContext * avctx, void *data,
}
if (err)
for (ch = 0; ch < s->out_channels; ch++)
memcpy(((float*)frame->data[ch]) + AC3_BLOCK_SIZE*blk, output[ch], sizeof(**output) * AC3_BLOCK_SIZE);
memcpy(((SHORTFLOAT*)frame->data[ch]) + AC3_BLOCK_SIZE*blk, output[ch], AC3_BLOCK_SIZE*sizeof(SHORTFLOAT));
for (ch = 0; ch < s->out_channels; ch++)
output[ch] = s->outptr[channel_map[ch]];
for (ch = 0; ch < s->out_channels; ch++) {
@ -1469,7 +1538,7 @@ static int ac3_decode_frame(AVCodecContext * avctx, void *data,
/* keep last block for error concealment in next frame */
for (ch = 0; ch < s->out_channels; ch++)
memcpy(s->output[ch], output[ch], sizeof(**output) * AC3_BLOCK_SIZE);
memcpy(s->output[ch], output[ch], AC3_BLOCK_SIZE*sizeof(SHORTFLOAT));
/*
* AVMatrixEncoding
@ -1540,66 +1609,12 @@ static av_cold int ac3_decode_end(AVCodecContext *avctx)
AC3DecodeContext *s = avctx->priv_data;
ff_mdct_end(&s->imdct_512);
ff_mdct_end(&s->imdct_256);
#if (CONFIG_AC3_FIXED)
av_free(s->fdsp);
#endif
return 0;
}
#define OFFSET(x) offsetof(AC3DecodeContext, x)
#define PAR (AV_OPT_FLAG_DECODING_PARAM | AV_OPT_FLAG_AUDIO_PARAM)
static const AVOption options[] = {
{ "drc_scale", "percentage of dynamic range compression to apply", OFFSET(drc_scale), AV_OPT_TYPE_FLOAT, {.dbl = 1.0}, 0.0, 6.0, PAR },
{"dmix_mode", "Preferred Stereo Downmix Mode", OFFSET(preferred_stereo_downmix), AV_OPT_TYPE_INT, {.i64 = -1 }, -1, 2, 0, "dmix_mode"},
{"ltrt_cmixlev", "Lt/Rt Center Mix Level", OFFSET(ltrt_center_mix_level), AV_OPT_TYPE_FLOAT, {.dbl = -1.0 }, -1.0, 2.0, 0},
{"ltrt_surmixlev", "Lt/Rt Surround Mix Level", OFFSET(ltrt_surround_mix_level), AV_OPT_TYPE_FLOAT, {.dbl = -1.0 }, -1.0, 2.0, 0},
{"loro_cmixlev", "Lo/Ro Center Mix Level", OFFSET(loro_center_mix_level), AV_OPT_TYPE_FLOAT, {.dbl = -1.0 }, -1.0, 2.0, 0},
{"loro_surmixlev", "Lo/Ro Surround Mix Level", OFFSET(loro_surround_mix_level), AV_OPT_TYPE_FLOAT, {.dbl = -1.0 }, -1.0, 2.0, 0},
{ NULL},
};
static const AVClass ac3_decoder_class = {
.class_name = "AC3 decoder",
.item_name = av_default_item_name,
.option = options,
.version = LIBAVUTIL_VERSION_INT,
};
AVCodec ff_ac3_decoder = {
.name = "ac3",
.long_name = NULL_IF_CONFIG_SMALL("ATSC A/52A (AC-3)"),
.type = AVMEDIA_TYPE_AUDIO,
.id = AV_CODEC_ID_AC3,
.priv_data_size = sizeof (AC3DecodeContext),
.init = ac3_decode_init,
.close = ac3_decode_end,
.decode = ac3_decode_frame,
.capabilities = CODEC_CAP_DR1,
.sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_FLTP,
AV_SAMPLE_FMT_NONE },
.priv_class = &ac3_decoder_class,
};
#if CONFIG_EAC3_DECODER
static const AVClass eac3_decoder_class = {
.class_name = "E-AC3 decoder",
.item_name = av_default_item_name,
.option = options,
.version = LIBAVUTIL_VERSION_INT,
};
AVCodec ff_eac3_decoder = {
.name = "eac3",
.long_name = NULL_IF_CONFIG_SMALL("ATSC A/52B (AC-3, E-AC-3)"),
.type = AVMEDIA_TYPE_AUDIO,
.id = AV_CODEC_ID_EAC3,
.priv_data_size = sizeof (AC3DecodeContext),
.init = ac3_decode_init,
.close = ac3_decode_end,
.decode = ac3_decode_frame,
.capabilities = CODEC_CAP_DR1,
.sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_FLTP,
AV_SAMPLE_FMT_NONE },
.priv_class = &eac3_decoder_class,
};
#endif

View File

@ -51,6 +51,7 @@
#define AVCODEC_AC3DEC_H
#include "libavutil/float_dsp.h"
#include "libavutil/fixed_dsp.h"
#include "libavutil/lfg.h"
#include "ac3.h"
#include "ac3dsp.h"
@ -138,8 +139,8 @@ typedef struct AC3DecodeContext {
int num_spx_bands; ///< number of spx bands (nspxbnds)
uint8_t spx_band_sizes[SPX_MAX_BANDS]; ///< number of bins in each spx band
uint8_t first_spx_coords[AC3_MAX_CHANNELS]; ///< first spx coordinates states (firstspxcos)
float spx_noise_blend[AC3_MAX_CHANNELS][SPX_MAX_BANDS]; ///< spx noise blending factor (nblendfact)
float spx_signal_blend[AC3_MAX_CHANNELS][SPX_MAX_BANDS];///< spx signal blending factor (sblendfact)
INTFLOAT spx_noise_blend[AC3_MAX_CHANNELS][SPX_MAX_BANDS]; ///< spx noise blending factor (nblendfact)
INTFLOAT spx_signal_blend[AC3_MAX_CHANNELS][SPX_MAX_BANDS];///< spx signal blending factor (sblendfact)
///@}
///@name Adaptive hybrid transform
@ -151,15 +152,15 @@ typedef struct AC3DecodeContext {
int fbw_channels; ///< number of full-bandwidth channels
int channels; ///< number of total channels
int lfe_ch; ///< index of LFE channel
float downmix_coeffs[AC3_MAX_CHANNELS][2]; ///< stereo downmix coefficients
SHORTFLOAT downmix_coeffs[AC3_MAX_CHANNELS][2]; ///< stereo downmix coefficients
int downmixed; ///< indicates if coeffs are currently downmixed
int output_mode; ///< output channel configuration
int out_channels; ///< number of output channels
///@}
///@name Dynamic range
float dynamic_range[2]; ///< dynamic range
float drc_scale; ///< percentage of dynamic range compression to be applied
INTFLOAT dynamic_range[2]; ///< dynamic range
INTFLOAT drc_scale; ///< percentage of dynamic range compression to be applied
///@}
///@name Bandwidth
@ -207,22 +208,26 @@ typedef struct AC3DecodeContext {
///@name Optimization
DSPContext dsp; ///< for optimization
#if CONFIG_AC3_FIXED
AVFixedDSPContext *fdsp;
#else
AVFloatDSPContext fdsp;
#endif
AC3DSPContext ac3dsp;
FmtConvertContext fmt_conv; ///< optimized conversion functions
///@}
float *outptr[AC3_MAX_CHANNELS];
float *xcfptr[AC3_MAX_CHANNELS];
float *dlyptr[AC3_MAX_CHANNELS];
SHORTFLOAT *outptr[AC3_MAX_CHANNELS];
INTFLOAT *xcfptr[AC3_MAX_CHANNELS];
INTFLOAT *dlyptr[AC3_MAX_CHANNELS];
///@name Aligned arrays
DECLARE_ALIGNED(16, int32_t, fixed_coeffs)[AC3_MAX_CHANNELS][AC3_MAX_COEFS]; ///< fixed-point transform coefficients
DECLARE_ALIGNED(32, float, transform_coeffs)[AC3_MAX_CHANNELS][AC3_MAX_COEFS]; ///< transform coefficients
DECLARE_ALIGNED(32, float, delay)[AC3_MAX_CHANNELS][AC3_BLOCK_SIZE]; ///< delay - added to the next block
DECLARE_ALIGNED(32, float, window)[AC3_BLOCK_SIZE]; ///< window coefficients
DECLARE_ALIGNED(32, float, tmp_output)[AC3_BLOCK_SIZE]; ///< temporary storage for output before windowing
DECLARE_ALIGNED(32, float, output)[AC3_MAX_CHANNELS][AC3_BLOCK_SIZE]; ///< output after imdct transform and windowing
DECLARE_ALIGNED(16, int, fixed_coeffs)[AC3_MAX_CHANNELS][AC3_MAX_COEFS]; ///< fixed-point transform coefficients
DECLARE_ALIGNED(32, INTFLOAT, transform_coeffs)[AC3_MAX_CHANNELS][AC3_MAX_COEFS]; ///< transform coefficients
DECLARE_ALIGNED(32, INTFLOAT, delay)[AC3_MAX_CHANNELS][AC3_BLOCK_SIZE]; ///< delay - added to the next block
DECLARE_ALIGNED(32, INTFLOAT, window)[AC3_BLOCK_SIZE]; ///< window coefficients
DECLARE_ALIGNED(32, INTFLOAT, tmp_output)[AC3_BLOCK_SIZE]; ///< temporary storage for output before windowing
DECLARE_ALIGNED(32, SHORTFLOAT, output)[AC3_MAX_CHANNELS][AC3_BLOCK_SIZE]; ///< output after imdct transform and windowing
DECLARE_ALIGNED(32, uint8_t, input_buffer)[AC3_FRAME_BUFFER_SIZE + FF_INPUT_BUFFER_PADDING_SIZE]; ///< temp buffer to prevent overread
///@}
} AC3DecodeContext;

176
libavcodec/ac3dec_fixed.c Normal file
View File

@ -0,0 +1,176 @@
/*
* Copyright (c) 2012
* MIPS Technologies, Inc., California.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Neither the name of the MIPS Technologies, Inc., nor the names of its
* contributors may be used to endorse or promote products derived from
* this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE MIPS TECHNOLOGIES, INC. ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE MIPS TECHNOLOGIES, INC. BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* Author: Stanislav Ocovaj (socovaj@mips.com)
*
* AC3 fixed-point decoder for MIPS platforms
*
* This file is part of FFmpeg.
*
* FFmpeg 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.
*
* FFmpeg 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 FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#define FFT_FLOAT 0
#define CONFIG_AC3_FIXED 1
#define FFT_FIXED_32 1
#include "ac3dec.h"
/**
* Table for center mix levels
* reference: Section 5.4.2.4 cmixlev
*/
static const uint8_t center_levels[4] = { 4, 5, 6, 5 };
/**
* Table for surround mix levels
* reference: Section 5.4.2.5 surmixlev
*/
static const uint8_t surround_levels[4] = { 4, 6, 7, 6 };
int end_freq_inv_tab[8] =
{
50529027, 44278013, 39403370, 32292987, 27356480, 23729101, 20951060, 18755316
};
static void scale_coefs (
int32_t *dst,
const int32_t *src,
int dynrng,
int len)
{
int i, shift, round;
int16_t mul;
int temp, temp1, temp2, temp3, temp4, temp5, temp6, temp7;
mul = (dynrng & 0x1f) + 0x20;
shift = 4 - ((dynrng << 24) >> 29);
round = 1 << (shift-1);
for (i=0; i<len; i+=8) {
temp = src[i] * mul;
temp1 = src[i+1] * mul;
temp = temp + round;
temp2 = src[i+2] * mul;
temp1 = temp1 + round;
dst[i] = temp >> shift;
temp3 = src[i+3] * mul;
temp2 = temp2 + round;
dst[i+1] = temp1 >> shift;
temp4 = src[i + 4] * mul;
temp3 = temp3 + round;
dst[i+2] = temp2 >> shift;
temp5 = src[i+5] * mul;
temp4 = temp4 + round;
dst[i+3] = temp3 >> shift;
temp6 = src[i+6] * mul;
dst[i+4] = temp4 >> shift;
temp5 = temp5 + round;
temp7 = src[i+7] * mul;
temp6 = temp6 + round;
dst[i+5] = temp5 >> shift;
temp7 = temp7 + round;
dst[i+6] = temp6 >> shift;
dst[i+7] = temp7 >> shift;
}
}
/**
* Downmix samples from original signal to stereo or mono (this is for 16-bit samples
* and fixed point decoder - original (for 32-bit samples) is in ac3dsp.c).
*/
static void ac3_downmix_c_fixed16(int16_t **samples, int16_t (*matrix)[2],
int out_ch, int in_ch, int len)
{
int i, j;
int v0, v1;
if (out_ch == 2) {
for (i = 0; i < len; i++) {
v0 = v1 = 0;
for (j = 0; j < in_ch; j++) {
v0 += samples[j][i] * matrix[j][0];
v1 += samples[j][i] * matrix[j][1];
}
samples[0][i] = (v0+2048)>>12;
samples[1][i] = (v1+2048)>>12;
}
} else if (out_ch == 1) {
for (i = 0; i < len; i++) {
v0 = 0;
for (j = 0; j < in_ch; j++)
v0 += samples[j][i] * matrix[j][0];
samples[0][i] = (v0+2048)>>12;
}
}
}
#include "ac3dec.c"
static const AVOption options[] = {
{ NULL},
};
static const AVClass ac3_decoder_class = {
.class_name = "Fixed-Point AC-3 Decoder",
.item_name = av_default_item_name,
.option = options,
.version = LIBAVUTIL_VERSION_INT,
};
AVCodec ff_ac3_fixed_decoder = {
.name = "ac3_fixed",
.type = AVMEDIA_TYPE_AUDIO,
.id = CODEC_ID_AC3,
.priv_data_size = sizeof (AC3DecodeContext),
.init = ac3_decode_init,
.close = ac3_decode_end,
.decode = ac3_decode_frame,
.capabilities = CODEC_CAP_DR1,
.long_name = NULL_IF_CONFIG_SMALL("ATSC A/52A (AC-3)"),
.sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_S16P,
AV_SAMPLE_FMT_NONE },
.priv_class = &ac3_decoder_class,
};

89
libavcodec/ac3dec_float.c Normal file
View File

@ -0,0 +1,89 @@
/*
* AC-3 Audio Decoder
* This code was developed as part of Google Summer of Code 2006.
* E-AC-3 support was added as part of Google Summer of Code 2007.
*
* Copyright (c) 2006 Kartikey Mahendra BHATT (bhattkm at gmail dot com)
* Copyright (c) 2007-2008 Bartlomiej Wolowiec <bartek.wolowiec@gmail.com>
* Copyright (c) 2007 Justin Ruggles <justin.ruggles@gmail.com>
*
* This file is part of FFmpeg.
*
* FFmpeg 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.
*
* FFmpeg 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 FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
/**
* Upmix delay samples from stereo to original channel layout.
*/
#include "ac3dec.h"
#include "ac3dec.c"
static const AVOption options[] = {
{ "drc_scale", "percentage of dynamic range compression to apply", OFFSET(drc_scale), AV_OPT_TYPE_FLOAT, {.dbl = 1.0}, 0.0, 6.0, PAR },
{"dmix_mode", "Preferred Stereo Downmix Mode", OFFSET(preferred_stereo_downmix), AV_OPT_TYPE_INT, {.i64 = -1 }, -1, 2, 0, "dmix_mode"},
{"ltrt_cmixlev", "Lt/Rt Center Mix Level", OFFSET(ltrt_center_mix_level), AV_OPT_TYPE_FLOAT, {.dbl = -1.0 }, -1.0, 2.0, 0},
{"ltrt_surmixlev", "Lt/Rt Surround Mix Level", OFFSET(ltrt_surround_mix_level), AV_OPT_TYPE_FLOAT, {.dbl = -1.0 }, -1.0, 2.0, 0},
{"loro_cmixlev", "Lo/Ro Center Mix Level", OFFSET(loro_center_mix_level), AV_OPT_TYPE_FLOAT, {.dbl = -1.0 }, -1.0, 2.0, 0},
{"loro_surmixlev", "Lo/Ro Surround Mix Level", OFFSET(loro_surround_mix_level), AV_OPT_TYPE_FLOAT, {.dbl = -1.0 }, -1.0, 2.0, 0},
{ NULL},
};
static const AVClass ac3_decoder_class = {
.class_name = "AC3 decoder",
.item_name = av_default_item_name,
.option = options,
.version = LIBAVUTIL_VERSION_INT,
};
AVCodec ff_ac3_decoder = {
.name = "ac3",
.type = AVMEDIA_TYPE_AUDIO,
.id = AV_CODEC_ID_AC3,
.priv_data_size = sizeof (AC3DecodeContext),
.init = ac3_decode_init,
.close = ac3_decode_end,
.decode = ac3_decode_frame,
.capabilities = CODEC_CAP_DR1,
.long_name = NULL_IF_CONFIG_SMALL("ATSC A/52A (AC-3)"),
.sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_FLTP,
AV_SAMPLE_FMT_NONE },
.priv_class = &ac3_decoder_class,
};
#if CONFIG_EAC3_DECODER
static const AVClass eac3_decoder_class = {
.class_name = "E-AC3 decoder",
.item_name = av_default_item_name,
.option = options,
.version = LIBAVUTIL_VERSION_INT,
};
AVCodec ff_eac3_decoder = {
.name = "eac3",
.type = AVMEDIA_TYPE_AUDIO,
.id = AV_CODEC_ID_EAC3,
.priv_data_size = sizeof (AC3DecodeContext),
.init = ac3_decode_init,
.close = ac3_decode_end,
.decode = ac3_decode_frame,
.capabilities = CODEC_CAP_DR1,
.long_name = NULL_IF_CONFIG_SMALL("ATSC A/52B (AC-3, E-AC-3)"),
.sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_FLTP,
AV_SAMPLE_FMT_NONE },
.priv_class = &eac3_decoder_class,
};
#endif

View File

@ -239,6 +239,31 @@ static void ac3_downmix_c(float **samples, float (*matrix)[2],
}
}
static void ac3_downmix_c_fixed(int32_t **samples, int16_t (*matrix)[2],
int out_ch, int in_ch, int len)
{
int i, j;
int64_t v0, v1;
if (out_ch == 2) {
for (i = 0; i < len; i++) {
v0 = v1 = 0;
for (j = 0; j < in_ch; j++) {
v0 += (int64_t)samples[j][i] * matrix[j][0];
v1 += (int64_t)samples[j][i] * matrix[j][1];
}
samples[0][i] = (v0+2048)>>12;
samples[1][i] = (v1+2048)>>12;
}
} else if (out_ch == 1) {
for (i = 0; i < len; i++) {
v0 = 0;
for (j = 0; j < in_ch; j++)
v0 += (int64_t)samples[j][i] * matrix[j][0];
samples[0][i] = (v0+2048)>>12;
}
}
}
static void apply_window_int16_c(int16_t *output, const int16_t *input,
const int16_t *window, unsigned int len)
{
@ -266,6 +291,7 @@ av_cold void ff_ac3dsp_init(AC3DSPContext *c, int bit_exact)
c->sum_square_butterfly_int32 = ac3_sum_square_butterfly_int32_c;
c->sum_square_butterfly_float = ac3_sum_square_butterfly_float_c;
c->downmix = ac3_downmix_c;
c->downmix_fixed = ac3_downmix_c_fixed;
c->apply_window_int16 = apply_window_int16_c;
if (ARCH_ARM)

View File

@ -135,6 +135,9 @@ typedef struct AC3DSPContext {
void (*downmix)(float **samples, float (*matrix)[2], int out_ch,
int in_ch, int len);
void (*downmix_fixed)(int32_t **samples, int16_t (*matrix)[2], int out_ch,
int in_ch, int len);
/**
* Apply symmetric window in 16-bit fixed-point.
* @param output destination array

View File

@ -323,7 +323,7 @@ void avcodec_register_all(void)
REGISTER_ENCDEC (AAC, aac);
REGISTER_DECODER(AAC_LATM, aac_latm);
REGISTER_ENCDEC (AC3, ac3);
REGISTER_ENCODER(AC3_FIXED, ac3_fixed);
REGISTER_ENCDEC (AC3_FIXED, ac3_fixed);
REGISTER_ENCDEC (ALAC, alac);
REGISTER_DECODER(ALS, als);
REGISTER_DECODER(AMRNB, amrnb);

View File

@ -45,3 +45,13 @@ av_cold void ff_kbd_window_init(float *window, float alpha, int n)
for (i = 0; i < n; i++)
window[i] = sqrt(local_window[i] / sum);
}
av_cold void ff_kbd_window_init_fixed(int32_t *window, float alpha, int n)
{
int i;
float local_window[FF_KBD_WINDOW_MAX];
ff_kbd_window_init(local_window, alpha, n);
for (i = 0; i < n; i++)
window[i] = (int)floor(2147483647.0 * local_window[i] + 0.5);
}

View File

@ -31,5 +31,6 @@
* @param n size of half window, max FF_KBD_WINDOW_MAX
*/
void ff_kbd_window_init(float *window, float alpha, int n);
void ff_kbd_window_init_fixed(int32_t *window, float alpha, int n);
#endif /* AVCODEC_KBDWIN_H */

View File

@ -29,7 +29,7 @@
#include "libavutil/version.h"
#define LIBAVCODEC_VERSION_MAJOR 55
#define LIBAVCODEC_VERSION_MINOR 55
#define LIBAVCODEC_VERSION_MINOR 56
#define LIBAVCODEC_VERSION_MICRO 107
#define LIBAVCODEC_VERSION_INT AV_VERSION_INT(LIBAVCODEC_VERSION_MAJOR, \