10e26bc7 |
/* |
7df9e693 |
* ATRAC3 compatible decoder |
d311f8f3 |
* Copyright (c) 2006-2008 Maxim Poliakovski
* Copyright (c) 2006-2008 Benjamin Larsson |
10e26bc7 |
*
* 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
*/
/** |
ba87f080 |
* @file |
7df9e693 |
* ATRAC3 compatible decoder. |
d311f8f3 |
* This decoder handles Sony's ATRAC3 data.
* |
7df9e693 |
* Container formats used to store ATRAC3 data: |
d311f8f3 |
* RealMedia (.rm), RIFF WAV (.wav, .at3), Sony OpenMG (.oma, .aa3). |
10e26bc7 |
*
* To use this decoder, a calling application must supply the extradata |
d311f8f3 |
* bytes provided in the containers above. |
10e26bc7 |
*/
#include <math.h>
#include <stddef.h>
#include <stdio.h>
|
6fee1b90 |
#include "libavutil/attributes.h" |
d5a7229b |
#include "libavutil/float_dsp.h" |
c5f0b6bf |
#include "libavutil/libm.h" |
10e26bc7 |
#include "avcodec.h"
#include "bytestream.h" |
1429224b |
#include "fft.h" |
e55d5390 |
#include "get_bits.h" |
594d4d5d |
#include "internal.h" |
10e26bc7 |
|
0e1baede |
#include "atrac.h" |
10e26bc7 |
#include "atrac3data.h"
|
c61b28e0 |
#define MIN_CHANNELS 1
#define MAX_CHANNELS 8
#define MAX_JS_PAIRS 8 / 2
|
10e26bc7 |
#define JOINT_STEREO 0x12 |
cab0f3ab |
#define SINGLE 0x2 |
10e26bc7 |
|
c9161385 |
#define SAMPLES_PER_FRAME 1024
#define MDCT_SIZE 512 |
10e26bc7 |
|
e55d5390 |
typedef struct GainBlock { |
dc80e250 |
AtracGainInfo g_block[4]; |
e55d5390 |
} GainBlock;
typedef struct TonalComponent {
int pos;
int num_coefs;
float coef[8];
} TonalComponent;
typedef struct ChannelUnit {
int bands_coded;
int num_components;
float prev_frame[SAMPLES_PER_FRAME];
int gc_blk_switch;
TonalComponent components[64];
GainBlock gain_block[2]; |
10e26bc7 |
|
c9161385 |
DECLARE_ALIGNED(32, float, spectrum)[SAMPLES_PER_FRAME]; |
e55d5390 |
DECLARE_ALIGNED(32, float, imdct_buf)[SAMPLES_PER_FRAME]; |
10e26bc7 |
|
e55d5390 |
float delay_buf1[46]; ///<qmf delay buffers
float delay_buf2[46];
float delay_buf3[46];
} ChannelUnit; |
10e26bc7 |
|
e55d5390 |
typedef struct ATRAC3Context {
GetBitContext gb; |
10e26bc7 |
//@{
/** stream data */ |
e55d5390 |
int coding_mode;
ChannelUnit *units; |
10e26bc7 |
//@}
//@{
/** joint-stereo related variables */ |
c61b28e0 |
int matrix_coeff_index_prev[MAX_JS_PAIRS][4];
int matrix_coeff_index_now[MAX_JS_PAIRS][4];
int matrix_coeff_index_next[MAX_JS_PAIRS][4];
int weighting_delay[MAX_JS_PAIRS][6]; |
10e26bc7 |
//@}
//@{
/** data buffers */ |
e55d5390 |
uint8_t *decoded_bytes_buffer;
float temp_buf[1070]; |
10e26bc7 |
//@}
//@{
/** extradata */ |
e55d5390 |
int scrambled_stream; |
10e26bc7 |
//@} |
a28cccf6 |
|
d49f3fa5 |
AtracGCContext gainc_ctx;
FFTContext mdct_ctx; |
93f959b6 |
AVFloatDSPContext *fdsp; |
10e26bc7 |
} ATRAC3Context;
|
c9161385 |
static DECLARE_ALIGNED(32, float, mdct_window)[MDCT_SIZE]; |
5d1007f7 |
static VLC_TYPE atrac3_vlc_table[4096][2]; |
e55d5390 |
static VLC spectral_coeff_tab[7]; |
10e26bc7 |
|
9ccc349f |
/** |
e55d5390 |
* Regular 512 points IMDCT without overlapping, with the exception of the
* swapping of odd bands caused by the reverse spectra of the QMF. |
10e26bc7 |
*
* @param odd_band 1 if the band is an odd band
*/ |
e55d5390 |
static void imlt(ATRAC3Context *q, float *input, float *output, int odd_band) |
10e26bc7 |
{ |
e55d5390 |
int i; |
10e26bc7 |
if (odd_band) {
/** |
e55d5390 |
* Reverse the odd bands before IMDCT, this is an effect of the QMF
* transform or it gives better compression to do it this way.
* FIXME: It should be possible to handle this in imdct_calc
* for that to happen a modification of the prerotation step of
* all SIMD code and C code is needed.
* Or fix the functions before so they generate a pre reversed spectrum.
*/
for (i = 0; i < 128; i++)
FFSWAP(float, input[i], input[255 - i]); |
10e26bc7 |
}
|
e55d5390 |
q->mdct_ctx.imdct_calc(&q->mdct_ctx, output, input); |
10e26bc7 |
/* Perform windowing on the output. */ |
93f959b6 |
q->fdsp->vector_fmul(output, output, mdct_window, MDCT_SIZE); |
10e26bc7 |
}
|
e55d5390 |
/*
* indata descrambling, only used for data coming from the rm container |
10e26bc7 |
*/ |
e55d5390 |
static int decode_bytes(const uint8_t *input, uint8_t *out, int bytes)
{ |
10e26bc7 |
int i, off;
uint32_t c; |
e55d5390 |
const uint32_t *buf;
uint32_t *output = (uint32_t *)out; |
10e26bc7 |
|
e55d5390 |
off = (intptr_t)input & 3;
buf = (const uint32_t *)(input - off); |
eba1ff31 |
if (off)
c = av_be2ne32((0x537F6103U >> (off * 8)) | (0x537F6103U << (32 - (off * 8))));
else
c = av_be2ne32(0x537F6103U); |
10e26bc7 |
bytes += 3 + off; |
e55d5390 |
for (i = 0; i < bytes / 4; i++)
output[i] = c ^ buf[i]; |
10e26bc7 |
if (off) |
6d97484d |
avpriv_request_sample(NULL, "Offset of %d", off); |
10e26bc7 |
return off;
}
|
2d528349 |
static av_cold void init_imdct_window(void) |
e55d5390 |
{ |
327747de |
int i, j; |
10e26bc7 |
|
e55d5390 |
/* generate the mdct window, for details see |
10e26bc7 |
* http://wiki.multimedia.cx/index.php?title=RealAudio_atrc#Windows */ |
327747de |
for (i = 0, j = 255; i < 128; i++, j--) {
float wi = sin(((i + 0.5) / 256.0 - 0.5) * M_PI) + 1.0;
float wj = sin(((j + 0.5) / 256.0 - 0.5) * M_PI) + 1.0;
float w = 0.5 * (wi * wi + wj * wj);
mdct_window[i] = mdct_window[511 - i] = wi / w;
mdct_window[j] = mdct_window[511 - j] = wj / w; |
e55d5390 |
} |
10e26bc7 |
}
|
5ef251e5 |
static av_cold int atrac3_decode_close(AVCodecContext *avctx) |
10e26bc7 |
{
ATRAC3Context *q = avctx->priv_data;
|
ea77d3b8 |
av_freep(&q->units);
av_freep(&q->decoded_bytes_buffer); |
93f959b6 |
av_freep(&q->fdsp); |
5e76b8bb |
|
a28cccf6 |
ff_mdct_end(&q->mdct_ctx); |
10e26bc7 |
return 0;
}
|
9ccc349f |
/** |
e55d5390 |
* Mantissa decoding |
10e26bc7 |
* |
e55d5390 |
* @param selector which table the output values are coded with
* @param coding_flag constant length coding or variable length coding
* @param mantissas mantissa output table
* @param num_codes number of values to get |
10e26bc7 |
*/ |
e55d5390 |
static void read_quant_spectral_coeffs(GetBitContext *gb, int selector,
int coding_flag, int *mantissas,
int num_codes) |
10e26bc7 |
{ |
e55d5390 |
int i, code, huff_symb; |
10e26bc7 |
if (selector == 1) |
e55d5390 |
num_codes /= 2; |
10e26bc7 |
|
e55d5390 |
if (coding_flag != 0) { |
10e26bc7 |
/* constant length coding (CLC) */ |
e55d5390 |
int num_bits = clc_length_tab[selector]; |
10e26bc7 |
if (selector > 1) { |
e55d5390 |
for (i = 0; i < num_codes; i++) {
if (num_bits)
code = get_sbits(gb, num_bits); |
10e26bc7 |
else
code = 0; |
e55d5390 |
mantissas[i] = code; |
10e26bc7 |
}
} else { |
e55d5390 |
for (i = 0; i < num_codes; i++) {
if (num_bits)
code = get_bits(gb, num_bits); // num_bits is always 4 in this case |
10e26bc7 |
else
code = 0; |
e55d5390 |
mantissas[i * 2 ] = mantissa_clc_tab[code >> 2];
mantissas[i * 2 + 1] = mantissa_clc_tab[code & 3]; |
10e26bc7 |
}
}
} else {
/* variable length coding (VLC) */
if (selector != 1) { |
e55d5390 |
for (i = 0; i < num_codes; i++) {
huff_symb = get_vlc2(gb, spectral_coeff_tab[selector-1].table,
spectral_coeff_tab[selector-1].bits, 3);
huff_symb += 1;
code = huff_symb >> 1;
if (huff_symb & 1) |
10e26bc7 |
code = -code; |
e55d5390 |
mantissas[i] = code; |
10e26bc7 |
}
} else { |
e55d5390 |
for (i = 0; i < num_codes; i++) {
huff_symb = get_vlc2(gb, spectral_coeff_tab[selector - 1].table,
spectral_coeff_tab[selector - 1].bits, 3);
mantissas[i * 2 ] = mantissa_vlc_tab[huff_symb * 2 ];
mantissas[i * 2 + 1] = mantissa_vlc_tab[huff_symb * 2 + 1]; |
10e26bc7 |
}
}
}
}
|
9ccc349f |
/** |
10e26bc7 |
* Restore the quantized band spectrum coefficients
* |
e55d5390 |
* @return subband count, fix for broken specification/files |
10e26bc7 |
*/ |
e55d5390 |
static int decode_spectrum(GetBitContext *gb, float *output) |
10e26bc7 |
{ |
e55d5390 |
int num_subbands, coding_mode, i, j, first, last, subband_size;
int subband_vlc_index[32], sf_index[32];
int mantissas[128];
float scale_factor;
num_subbands = get_bits(gb, 5); // number of coded subbands
coding_mode = get_bits1(gb); // coding Mode: 0 - VLC/ 1-CLC
/* get the VLC selector table for the subbands, 0 means not coded */
for (i = 0; i <= num_subbands; i++)
subband_vlc_index[i] = get_bits(gb, 3);
/* read the scale factor indexes from the stream */
for (i = 0; i <= num_subbands; i++) {
if (subband_vlc_index[i] != 0)
sf_index[i] = get_bits(gb, 6); |
10e26bc7 |
}
|
e55d5390 |
for (i = 0; i <= num_subbands; i++) {
first = subband_tab[i ];
last = subband_tab[i + 1]; |
10e26bc7 |
|
e55d5390 |
subband_size = last - first; |
10e26bc7 |
|
e55d5390 |
if (subband_vlc_index[i] != 0) {
/* decode spectral coefficients for this subband */ |
10e26bc7 |
/* TODO: This can be done faster is several blocks share the
* same VLC selector (subband_vlc_index) */ |
e55d5390 |
read_quant_spectral_coeffs(gb, subband_vlc_index[i], coding_mode,
mantissas, subband_size); |
10e26bc7 |
|
e55d5390 |
/* decode the scale factor for this subband */
scale_factor = ff_atrac_sf_table[sf_index[i]] *
inv_max_quant[subband_vlc_index[i]]; |
10e26bc7 |
|
e55d5390 |
/* inverse quantize the coefficients */
for (j = 0; first < last; first++, j++)
output[first] = mantissas[j] * scale_factor; |
10e26bc7 |
} else { |
e55d5390 |
/* this subband was not coded, so zero the entire subband */ |
89a6c32b |
memset(output + first, 0, subband_size * sizeof(*output)); |
10e26bc7 |
}
}
|
e55d5390 |
/* clear the subbands that were not coded */
first = subband_tab[i]; |
89a6c32b |
memset(output + first, 0, (SAMPLES_PER_FRAME - first) * sizeof(*output)); |
e55d5390 |
return num_subbands; |
10e26bc7 |
}
|
9ccc349f |
/** |
10e26bc7 |
* Restore the quantized tonal components
* |
e55d5390 |
* @param components tonal components
* @param num_bands number of coded bands |
10e26bc7 |
*/ |
e55d5390 |
static int decode_tonal_components(GetBitContext *gb,
TonalComponent *components, int num_bands) |
10e26bc7 |
{ |
e55d5390 |
int i, b, c, m;
int nb_components, coding_mode_selector, coding_mode;
int band_flags[4], mantissa[8];
int component_count = 0; |
10e26bc7 |
|
e55d5390 |
nb_components = get_bits(gb, 5); |
10e26bc7 |
/* no tonal components */ |
e55d5390 |
if (nb_components == 0) |
10e26bc7 |
return 0;
|
e55d5390 |
coding_mode_selector = get_bits(gb, 2); |
10e26bc7 |
if (coding_mode_selector == 2) |
8f98577d |
return AVERROR_INVALIDDATA; |
10e26bc7 |
coding_mode = coding_mode_selector & 1;
|
e55d5390 |
for (i = 0; i < nb_components; i++) {
int coded_values_per_component, quant_step_index;
for (b = 0; b <= num_bands; b++)
band_flags[b] = get_bits1(gb); |
10e26bc7 |
|
e55d5390 |
coded_values_per_component = get_bits(gb, 3); |
10e26bc7 |
|
e55d5390 |
quant_step_index = get_bits(gb, 3); |
10e26bc7 |
if (quant_step_index <= 1) |
8f98577d |
return AVERROR_INVALIDDATA; |
10e26bc7 |
if (coding_mode_selector == 3)
coding_mode = get_bits1(gb);
|
e55d5390 |
for (b = 0; b < (num_bands + 1) * 4; b++) {
int coded_components;
if (band_flags[b >> 2] == 0) |
10e26bc7 |
continue;
|
e55d5390 |
coded_components = get_bits(gb, 3);
for (c = 0; c < coded_components; c++) {
TonalComponent *cmp = &components[component_count];
int sf_index, coded_values, max_coded_values;
float scale_factor; |
10e26bc7 |
|
e55d5390 |
sf_index = get_bits(gb, 6); |
c509f4f7 |
if (component_count >= 64) |
9af6abdc |
return AVERROR_INVALIDDATA; |
10e26bc7 |
|
e55d5390 |
cmp->pos = b * 64 + get_bits(gb, 6);
max_coded_values = SAMPLES_PER_FRAME - cmp->pos;
coded_values = coded_values_per_component + 1;
coded_values = FFMIN(max_coded_values, coded_values); |
10e26bc7 |
|
e55d5390 |
scale_factor = ff_atrac_sf_table[sf_index] *
inv_max_quant[quant_step_index]; |
10e26bc7 |
|
e55d5390 |
read_quant_spectral_coeffs(gb, quant_step_index, coding_mode,
mantissa, coded_values);
cmp->num_coefs = coded_values; |
10e26bc7 |
/* inverse quant */ |
e55d5390 |
for (m = 0; m < coded_values; m++)
cmp->coef[m] = mantissa[m] * scale_factor; |
10e26bc7 |
component_count++;
}
}
}
|
b8c4a515 |
return component_count; |
10e26bc7 |
}
|
9ccc349f |
/** |
10e26bc7 |
* Decode gain parameters for the coded bands
* |
e55d5390 |
* @param block the gainblock for the current band
* @param num_bands amount of coded bands |
10e26bc7 |
*/ |
e55d5390 |
static int decode_gain_control(GetBitContext *gb, GainBlock *block,
int num_bands) |
10e26bc7 |
{ |
4a63c69f |
int b, j; |
e55d5390 |
int *level, *loc;
|
dc80e250 |
AtracGainInfo *gain = block->g_block; |
e55d5390 |
|
4fa24840 |
for (b = 0; b <= num_bands; b++) {
gain[b].num_points = get_bits(gb, 3); |
4a63c69f |
level = gain[b].lev_code;
loc = gain[b].loc_code; |
e55d5390 |
|
c2df9597 |
for (j = 0; j < gain[b].num_points; j++) { |
be0b4c70 |
level[j] = get_bits(gb, 4);
loc[j] = get_bits(gb, 5);
if (j && loc[j] <= loc[j - 1]) |
8f98577d |
return AVERROR_INVALIDDATA; |
10e26bc7 |
}
}
/* Clear the unused blocks. */ |
4fa24840 |
for (; b < 4 ; b++)
gain[b].num_points = 0; |
10e26bc7 |
return 0;
}
|
9ccc349f |
/** |
10e26bc7 |
* Combine the tonal band spectrum and regular band spectrum
* |
e55d5390 |
* @param spectrum output spectrum buffer
* @param num_components number of tonal components
* @param components tonal components for this band
* @return position of the last tonal coefficient |
10e26bc7 |
*/ |
e55d5390 |
static int add_tonal_components(float *spectrum, int num_components,
TonalComponent *components) |
10e26bc7 |
{ |
e55d5390 |
int i, j, last_pos = -1;
float *input, *output; |
10e26bc7 |
|
e55d5390 |
for (i = 0; i < num_components; i++) {
last_pos = FFMAX(components[i].pos + components[i].num_coefs, last_pos);
input = components[i].coef;
output = &spectrum[components[i].pos]; |
10e26bc7 |
|
e55d5390 |
for (j = 0; j < components[i].num_coefs; j++) |
dcbb920f |
output[j] += input[j]; |
10e26bc7 |
} |
9d278d88 |
|
e55d5390 |
return last_pos; |
10e26bc7 |
}
|
e55d5390 |
#define INTERPOLATE(old, new, nsample) \
((old) + (nsample) * 0.125 * ((new) - (old))) |
10e26bc7 |
|
e55d5390 |
static void reverse_matrixing(float *su1, float *su2, int *prev_code,
int *curr_code) |
10e26bc7 |
{ |
e55d5390 |
int i, nsample, band;
float mc1_l, mc1_r, mc2_l, mc2_r; |
10e26bc7 |
|
e55d5390 |
for (i = 0, band = 0; band < 4 * 256; band += 256, i++) {
int s1 = prev_code[i];
int s2 = curr_code[i]; |
aefdb735 |
nsample = band; |
10e26bc7 |
if (s1 != s2) {
/* Selector value changed, interpolation needed. */ |
e55d5390 |
mc1_l = matrix_coeffs[s1 * 2 ];
mc1_r = matrix_coeffs[s1 * 2 + 1];
mc2_l = matrix_coeffs[s2 * 2 ];
mc2_r = matrix_coeffs[s2 * 2 + 1]; |
10e26bc7 |
/* Interpolation is done over the first eight samples. */ |
aefdb735 |
for (; nsample < band + 8; nsample++) {
float c1 = su1[nsample];
float c2 = su2[nsample];
c2 = c1 * INTERPOLATE(mc1_l, mc2_l, nsample - band) +
c2 * INTERPOLATE(mc1_r, mc2_r, nsample - band);
su1[nsample] = c2;
su2[nsample] = c1 * 2.0 - c2; |
10e26bc7 |
}
}
/* Apply the matrix without interpolation. */
switch (s2) { |
e55d5390 |
case 0: /* M/S decoding */ |
aefdb735 |
for (; nsample < band + 256; nsample++) {
float c1 = su1[nsample];
float c2 = su2[nsample];
su1[nsample] = c2 * 2.0;
su2[nsample] = (c1 - c2) * 2.0; |
e55d5390 |
}
break;
case 1: |
aefdb735 |
for (; nsample < band + 256; nsample++) {
float c1 = su1[nsample];
float c2 = su2[nsample];
su1[nsample] = (c1 + c2) * 2.0;
su2[nsample] = c2 * -2.0; |
e55d5390 |
}
break;
case 2:
case 3: |
aefdb735 |
for (; nsample < band + 256; nsample++) {
float c1 = su1[nsample];
float c2 = su2[nsample];
su1[nsample] = c1 + c2;
su2[nsample] = c1 - c2; |
e55d5390 |
}
break;
default: |
dcb0d119 |
av_assert1(0); |
10e26bc7 |
}
}
}
|
e55d5390 |
static void get_channel_weights(int index, int flag, float ch[2])
{
if (index == 7) { |
10e26bc7 |
ch[0] = 1.0;
ch[1] = 1.0;
} else { |
e55d5390 |
ch[0] = (index & 7) / 7.0;
ch[1] = sqrt(2 - ch[0] * ch[0]);
if (flag) |
10e26bc7 |
FFSWAP(float, ch[0], ch[1]);
}
}
|
e55d5390 |
static void channel_weighting(float *su1, float *su2, int *p3) |
10e26bc7 |
{ |
e55d5390 |
int band, nsample; |
10e26bc7 |
/* w[x][y] y=0 is left y=1 is right */
float w[2][2];
|
e55d5390 |
if (p3[1] != 7 || p3[3] != 7) {
get_channel_weights(p3[1], p3[0], w[0]);
get_channel_weights(p3[3], p3[2], w[1]); |
10e26bc7 |
|
aefdb735 |
for (band = 256; band < 4 * 256; band += 256) {
for (nsample = band; nsample < band + 8; nsample++) {
su1[nsample] *= INTERPOLATE(w[0][0], w[0][1], nsample - band);
su2[nsample] *= INTERPOLATE(w[1][0], w[1][1], nsample - band); |
10e26bc7 |
} |
aefdb735 |
for(; nsample < band + 256; nsample++) {
su1[nsample] *= w[1][0];
su2[nsample] *= w[1][1]; |
10e26bc7 |
}
}
}
}
|
9ccc349f |
/** |
10e26bc7 |
* Decode a Sound Unit
* |
e55d5390 |
* @param snd the channel unit to be used
* @param output the decoded samples before IQMF in float representation
* @param channel_num channel number |
cab0f3ab |
* @param coding_mode the coding mode (JOINT_STEREO or single channels) |
10e26bc7 |
*/ |
e55d5390 |
static int decode_channel_sound_unit(ATRAC3Context *q, GetBitContext *gb,
ChannelUnit *snd, float *output,
int channel_num, int coding_mode) |
10e26bc7 |
{ |
e55d5390 |
int band, ret, num_subbands, last_tonal, num_bands;
GainBlock *gain1 = &snd->gain_block[ snd->gc_blk_switch];
GainBlock *gain2 = &snd->gain_block[1 - snd->gc_blk_switch]; |
10e26bc7 |
|
c61b28e0 |
if (coding_mode == JOINT_STEREO && (channel_num % 2) == 1) { |
e55d5390 |
if (get_bits(gb, 2) != 3) { |
10e26bc7 |
av_log(NULL,AV_LOG_ERROR,"JS mono Sound Unit id != 3.\n"); |
8f98577d |
return AVERROR_INVALIDDATA; |
10e26bc7 |
}
} else { |
e55d5390 |
if (get_bits(gb, 6) != 0x28) { |
10e26bc7 |
av_log(NULL,AV_LOG_ERROR,"Sound Unit id != 0x28.\n"); |
8f98577d |
return AVERROR_INVALIDDATA; |
10e26bc7 |
}
}
/* number of coded QMF bands */ |
e55d5390 |
snd->bands_coded = get_bits(gb, 2); |
10e26bc7 |
|
e55d5390 |
ret = decode_gain_control(gb, gain2, snd->bands_coded);
if (ret)
return ret; |
10e26bc7 |
|
e55d5390 |
snd->num_components = decode_tonal_components(gb, snd->components,
snd->bands_coded); |
5eaed6d3 |
if (snd->num_components < 0)
return snd->num_components; |
10e26bc7 |
|
e55d5390 |
num_subbands = decode_spectrum(gb, snd->spectrum); |
10e26bc7 |
/* Merge the decoded spectrum and tonal components. */ |
e55d5390 |
last_tonal = add_tonal_components(snd->spectrum, snd->num_components,
snd->components); |
10e26bc7 |
|
e55d5390 |
/* calculate number of used MLT/QMF bands according to the amount of coded
spectral lines */
num_bands = (subband_tab[num_subbands] - 1) >> 8;
if (last_tonal >= 0)
num_bands = FFMAX((last_tonal + 256) >> 8, num_bands); |
10e26bc7 |
/* Reconstruct time domain samples. */ |
e55d5390 |
for (band = 0; band < 4; band++) { |
10e26bc7 |
/* Perform the IMDCT step without overlapping. */ |
e55d5390 |
if (band <= num_bands)
imlt(q, &snd->spectrum[band * 256], snd->imdct_buf, band & 1);
else |
89a6c32b |
memset(snd->imdct_buf, 0, 512 * sizeof(*snd->imdct_buf)); |
10e26bc7 |
/* gain compensation and overlapping */ |
d49f3fa5 |
ff_atrac_gain_compensation(&q->gainc_ctx, snd->imdct_buf,
&snd->prev_frame[band * 256],
&gain1->g_block[band], &gain2->g_block[band],
256, &output[band * 256]); |
10e26bc7 |
}
/* Swap the gain control buffers for the next frame. */ |
e55d5390 |
snd->gc_blk_switch ^= 1; |
10e26bc7 |
return 0;
}
|
5ac673b5 |
static int decode_frame(AVCodecContext *avctx, const uint8_t *databuf, |
e55d5390 |
float **out_samples) |
10e26bc7 |
{ |
5ac673b5 |
ATRAC3Context *q = avctx->priv_data; |
c61b28e0 |
int ret, i, ch; |
15ae1959 |
uint8_t *ptr1; |
10e26bc7 |
|
e55d5390 |
if (q->coding_mode == JOINT_STEREO) { |
10e26bc7 |
/* channel coupling mode */
|
c61b28e0 |
/* Decode sound unit pairs (channels are expected to be even).
* Multichannel joint stereo interleaves pairs (6ch: 2ch + 2ch + 2ch) */ |
b47582f4 |
const uint8_t *js_databuf; |
c61b28e0 |
int js_pair, js_block_align; |
10e26bc7 |
|
c61b28e0 |
js_block_align = (avctx->block_align / avctx->channels) * 2; /* block pair */ |
10e26bc7 |
|
c61b28e0 |
for (ch = 0; ch < avctx->channels; ch = ch + 2) {
js_pair = ch/2;
js_databuf = databuf + js_pair * js_block_align; /* align to current pair */ |
10e26bc7 |
|
c61b28e0 |
/* Set the bitstream reader at the start of first channel sound unit. */
init_get_bits(&q->gb,
js_databuf, js_block_align * 8); |
10e26bc7 |
|
c61b28e0 |
/* decode Sound Unit 1 */
ret = decode_channel_sound_unit(q, &q->gb, &q->units[ch],
out_samples[ch], ch, JOINT_STEREO);
if (ret != 0)
return ret;
/* Framedata of the su2 in the joint-stereo mode is encoded in
* reverse byte order so we need to swap it first. */
if (js_databuf == q->decoded_bytes_buffer) {
uint8_t *ptr2 = q->decoded_bytes_buffer + js_block_align - 1;
ptr1 = q->decoded_bytes_buffer;
for (i = 0; i < js_block_align / 2; i++, ptr1++, ptr2--)
FFSWAP(uint8_t, *ptr1, *ptr2);
} else {
const uint8_t *ptr2 = js_databuf + js_block_align - 1;
for (i = 0; i < js_block_align; i++)
q->decoded_bytes_buffer[i] = *ptr2--;
}
/* Skip the sync codes (0xF8). */
ptr1 = q->decoded_bytes_buffer;
for (i = 4; *ptr1 == 0xF8; i++, ptr1++) {
if (i >= js_block_align)
return AVERROR_INVALIDDATA;
} |
10e26bc7 |
|
c61b28e0 |
/* set the bitstream reader at the start of the second Sound Unit */ |
e976e68f |
ret = init_get_bits8(&q->gb, |
c61b28e0 |
ptr1, q->decoded_bytes_buffer + js_block_align - ptr1); |
e976e68f |
if (ret < 0)
return ret; |
10e26bc7 |
|
c61b28e0 |
/* Fill the Weighting coeffs delay buffer */
memmove(q->weighting_delay[js_pair], &q->weighting_delay[js_pair][2],
4 * sizeof(*q->weighting_delay[js_pair]));
q->weighting_delay[js_pair][4] = get_bits1(&q->gb);
q->weighting_delay[js_pair][5] = get_bits(&q->gb, 3);
for (i = 0; i < 4; i++) {
q->matrix_coeff_index_prev[js_pair][i] = q->matrix_coeff_index_now[js_pair][i];
q->matrix_coeff_index_now[js_pair][i] = q->matrix_coeff_index_next[js_pair][i];
q->matrix_coeff_index_next[js_pair][i] = get_bits(&q->gb, 2);
}
/* Decode Sound Unit 2. */
ret = decode_channel_sound_unit(q, &q->gb, &q->units[ch+1],
out_samples[ch+1], ch+1, JOINT_STEREO);
if (ret != 0)
return ret;
/* Reconstruct the channel coefficients. */
reverse_matrixing(out_samples[ch], out_samples[ch+1],
q->matrix_coeff_index_prev[js_pair],
q->matrix_coeff_index_now[js_pair]);
channel_weighting(out_samples[ch], out_samples[ch+1], q->weighting_delay[js_pair]);
} |
10e26bc7 |
} else { |
cab0f3ab |
/* single channels */ |
10e26bc7 |
/* Decode the channel sound units. */ |
5ac673b5 |
for (i = 0; i < avctx->channels; i++) { |
10e26bc7 |
/* Set the bitstream reader at the start of a channel sound unit. */ |
ee41963f |
init_get_bits(&q->gb, |
cdd0e0de |
databuf + i * avctx->block_align / avctx->channels,
avctx->block_align * 8 / avctx->channels); |
10e26bc7 |
|
e55d5390 |
ret = decode_channel_sound_unit(q, &q->gb, &q->units[i],
out_samples[i], i, q->coding_mode);
if (ret != 0)
return ret; |
10e26bc7 |
}
}
/* Apply the iQMF synthesis filter. */ |
5ac673b5 |
for (i = 0; i < avctx->channels; i++) { |
e55d5390 |
float *p1 = out_samples[i];
float *p2 = p1 + 256;
float *p3 = p2 + 256;
float *p4 = p3 + 256;
ff_atrac_iqmf(p1, p2, 256, p1, q->units[i].delay_buf1, q->temp_buf);
ff_atrac_iqmf(p4, p3, 256, p3, q->units[i].delay_buf2, q->temp_buf);
ff_atrac_iqmf(p1, p3, 512, p1, q->units[i].delay_buf3, q->temp_buf); |
10e26bc7 |
}
return 0;
}
|
280a40dd |
static int al_decode_frame(AVCodecContext *avctx, const uint8_t *databuf,
int size, float **out_samples)
{
ATRAC3Context *q = avctx->priv_data;
int ret, i;
/* Set the bitstream reader at the start of a channel sound unit. */
init_get_bits(&q->gb, databuf, size * 8);
/* single channels */
/* Decode the channel sound units. */
for (i = 0; i < avctx->channels; i++) {
ret = decode_channel_sound_unit(q, &q->gb, &q->units[i],
out_samples[i], i, q->coding_mode);
if (ret != 0)
return ret;
while (i < avctx->channels && get_bits_left(&q->gb) > 6 && show_bits(&q->gb, 6) != 0x28) {
skip_bits(&q->gb, 1);
}
}
/* Apply the iQMF synthesis filter. */
for (i = 0; i < avctx->channels; i++) {
float *p1 = out_samples[i];
float *p2 = p1 + 256;
float *p3 = p2 + 256;
float *p4 = p3 + 256;
ff_atrac_iqmf(p1, p2, 256, p1, q->units[i].delay_buf1, q->temp_buf);
ff_atrac_iqmf(p4, p3, 256, p3, q->units[i].delay_buf2, q->temp_buf);
ff_atrac_iqmf(p1, p3, 512, p1, q->units[i].delay_buf3, q->temp_buf);
}
return 0;
}
|
0eea2129 |
static int atrac3_decode_frame(AVCodecContext *avctx, void *data,
int *got_frame_ptr, AVPacket *avpkt)
{ |
9a75ace2 |
AVFrame *frame = data; |
7a00bbad |
const uint8_t *buf = avpkt->data;
int buf_size = avpkt->size; |
10e26bc7 |
ATRAC3Context *q = avctx->priv_data; |
e55d5390 |
int ret;
const uint8_t *databuf; |
10e26bc7 |
|
46a76dec |
if (buf_size < avctx->block_align) {
av_log(avctx, AV_LOG_ERROR,
"Frame too small (%d bytes). Truncated file?\n", buf_size); |
1fead73d |
return AVERROR_INVALIDDATA; |
46a76dec |
} |
10e26bc7 |
|
0eea2129 |
/* get output buffer */ |
9a75ace2 |
frame->nb_samples = SAMPLES_PER_FRAME; |
1ec94b0f |
if ((ret = ff_get_buffer(avctx, frame, 0)) < 0) |
e55d5390 |
return ret; |
10e26bc7 |
/* Check if we need to descramble and what buffer to pass on. */
if (q->scrambled_stream) {
decode_bytes(buf, q->decoded_bytes_buffer, avctx->block_align);
databuf = q->decoded_bytes_buffer;
} else {
databuf = buf;
}
|
9a75ace2 |
ret = decode_frame(avctx, databuf, (float **)frame->extended_data); |
e55d5390 |
if (ret) { |
c9fb81ff |
av_log(avctx, AV_LOG_ERROR, "Frame decoding error!\n"); |
e55d5390 |
return ret; |
10e26bc7 |
}
|
9a75ace2 |
*got_frame_ptr = 1; |
10e26bc7 |
return avctx->block_align;
}
|
280a40dd |
static int atrac3al_decode_frame(AVCodecContext *avctx, void *data,
int *got_frame_ptr, AVPacket *avpkt)
{
AVFrame *frame = data;
int ret;
frame->nb_samples = SAMPLES_PER_FRAME;
if ((ret = ff_get_buffer(avctx, frame, 0)) < 0)
return ret;
ret = al_decode_frame(avctx, avpkt->data, avpkt->size,
(float **)frame->extended_data);
if (ret) {
av_log(avctx, AV_LOG_ERROR, "Frame decoding error!\n");
return ret;
}
*got_frame_ptr = 1;
return avpkt->size;
}
|
0aa09548 |
static av_cold void atrac3_init_static_data(void) |
5d1007f7 |
{
int i;
|
2d528349 |
init_imdct_window(); |
5d1007f7 |
ff_atrac_generate_tables();
/* Initialize the VLC tables. */
for (i = 0; i < 7; i++) {
spectral_coeff_tab[i].table = &atrac3_vlc_table[atrac3_vlc_offs[i]];
spectral_coeff_tab[i].table_allocated = atrac3_vlc_offs[i + 1] -
atrac3_vlc_offs[i ];
init_vlc(&spectral_coeff_tab[i], 9, huff_tab_sizes[i],
huff_bits[i], 1, 1,
huff_codes[i], 1, 1, INIT_VLC_USE_NEW_STATIC);
}
}
|
5ef251e5 |
static av_cold int atrac3_decode_init(AVCodecContext *avctx) |
10e26bc7 |
{ |
8aa29f06 |
static int static_init_done; |
c61b28e0 |
int i, js_pair, ret; |
c51311b9 |
int version, delay, samples_per_frame, frame_factor; |
8687f767 |
const uint8_t *edata_ptr = avctx->extradata; |
10e26bc7 |
ATRAC3Context *q = avctx->priv_data;
|
c61b28e0 |
if (avctx->channels < MIN_CHANNELS || avctx->channels > MAX_CHANNELS) { |
5ac673b5 |
av_log(avctx, AV_LOG_ERROR, "Channel configuration error!\n");
return AVERROR(EINVAL);
}
|
8aa29f06 |
if (!static_init_done)
atrac3_init_static_data();
static_init_done = 1;
|
10e26bc7 |
/* Take care of the codec-specific extradata. */ |
280a40dd |
if (avctx->codec_id == AV_CODEC_ID_ATRAC3AL) {
version = 4;
samples_per_frame = SAMPLES_PER_FRAME * avctx->channels;
delay = 0x88E;
q->coding_mode = SINGLE;
} else if (avctx->extradata_size == 14) { |
10e26bc7 |
/* Parse the extradata, WAV format */ |
e55d5390 |
av_log(avctx, AV_LOG_DEBUG, "[0-1] %d\n",
bytestream_get_le16(&edata_ptr)); // Unknown value always 1 |
7c1f93af |
edata_ptr += 4; // samples per channel |
e55d5390 |
q->coding_mode = bytestream_get_le16(&edata_ptr);
av_log(avctx, AV_LOG_DEBUG,"[8-9] %d\n",
bytestream_get_le16(&edata_ptr)); //Dupe of coding mode |
c51311b9 |
frame_factor = bytestream_get_le16(&edata_ptr); // Unknown always 1 |
e55d5390 |
av_log(avctx, AV_LOG_DEBUG,"[12-13] %d\n",
bytestream_get_le16(&edata_ptr)); // Unknown always 0 |
10e26bc7 |
/* setup */ |
a2664c91 |
samples_per_frame = SAMPLES_PER_FRAME * avctx->channels; |
56a9d2b4 |
version = 4; |
64ebbb8f |
delay = 0x88E; |
cab0f3ab |
q->coding_mode = q->coding_mode ? JOINT_STEREO : SINGLE; |
e55d5390 |
q->scrambled_stream = 0;
|
c51311b9 |
if (avctx->block_align != 96 * avctx->channels * frame_factor &&
avctx->block_align != 152 * avctx->channels * frame_factor &&
avctx->block_align != 192 * avctx->channels * frame_factor) { |
e55d5390 |
av_log(avctx, AV_LOG_ERROR, "Unknown frame/channel/frame_factor " |
cdd0e0de |
"configuration %d/%d/%d\n", avctx->block_align, |
c51311b9 |
avctx->channels, frame_factor); |
8f98577d |
return AVERROR_INVALIDDATA; |
10e26bc7 |
} |
034a125c |
} else if (avctx->extradata_size == 12 || avctx->extradata_size == 10) { |
10e26bc7 |
/* Parse the extradata, RM format. */ |
56a9d2b4 |
version = bytestream_get_be32(&edata_ptr); |
a2664c91 |
samples_per_frame = bytestream_get_be16(&edata_ptr); |
64ebbb8f |
delay = bytestream_get_be16(&edata_ptr); |
e55d5390 |
q->coding_mode = bytestream_get_be16(&edata_ptr);
q->scrambled_stream = 1; |
10e26bc7 |
} else { |
c9fb81ff |
av_log(avctx, AV_LOG_ERROR, "Unknown extradata size %d.\n", |
e55d5390 |
avctx->extradata_size); |
44d854a5 |
return AVERROR(EINVAL); |
10e26bc7 |
}
|
e55d5390 |
/* Check the extradata */
|
56a9d2b4 |
if (version != 4) {
av_log(avctx, AV_LOG_ERROR, "Version %d != 4.\n", version); |
8f98577d |
return AVERROR_INVALIDDATA; |
10e26bc7 |
}
|
cab0f3ab |
if (samples_per_frame != SAMPLES_PER_FRAME * avctx->channels) { |
e55d5390 |
av_log(avctx, AV_LOG_ERROR, "Unknown amount of samples per frame %d.\n", |
a2664c91 |
samples_per_frame); |
8f98577d |
return AVERROR_INVALIDDATA; |
10e26bc7 |
}
|
64ebbb8f |
if (delay != 0x88E) { |
e55d5390 |
av_log(avctx, AV_LOG_ERROR, "Unknown amount of delay %x != 0x88E.\n", |
64ebbb8f |
delay); |
8f98577d |
return AVERROR_INVALIDDATA; |
10e26bc7 |
}
|
cab0f3ab |
if (q->coding_mode == SINGLE)
av_log(avctx, AV_LOG_DEBUG, "Single channels detected.\n"); |
50cf5a7f |
else if (q->coding_mode == JOINT_STEREO) { |
c61b28e0 |
if (avctx->channels % 2 == 1) { /* Joint stereo channels must be even */
av_log(avctx, AV_LOG_ERROR, "Invalid joint stereo channel configuration.\n"); |
50cf5a7f |
return AVERROR_INVALIDDATA; |
a3d9a216 |
} |
e55d5390 |
av_log(avctx, AV_LOG_DEBUG, "Joint stereo detected.\n"); |
50cf5a7f |
} else { |
e55d5390 |
av_log(avctx, AV_LOG_ERROR, "Unknown channel coding mode %x!\n",
q->coding_mode); |
8f98577d |
return AVERROR_INVALIDDATA; |
10e26bc7 |
}
|
ed04ecd2 |
if (avctx->block_align > 1024 || avctx->block_align <= 0) |
8f98577d |
return AVERROR(EINVAL); |
10e26bc7 |
|
a1f4cd37 |
q->decoded_bytes_buffer = av_mallocz(FFALIGN(avctx->block_align, 4) + |
059a9348 |
AV_INPUT_BUFFER_PADDING_SIZE); |
f929ab05 |
if (!q->decoded_bytes_buffer) |
6611c0b4 |
return AVERROR(ENOMEM); |
10e26bc7 |
|
9af4eaa8 |
avctx->sample_fmt = AV_SAMPLE_FMT_FLTP; |
20732246 |
|
78edce3f |
/* initialize the MDCT transform */
if ((ret = ff_mdct_init(&q->mdct_ctx, 9, 1, 1.0 / 32768)) < 0) { |
47b61702 |
av_log(avctx, AV_LOG_ERROR, "Error initializing MDCT\n");
av_freep(&q->decoded_bytes_buffer);
return ret;
} |
10e26bc7 |
/* init the joint-stereo decoding data */ |
c61b28e0 |
for (js_pair = 0; js_pair < MAX_JS_PAIRS; js_pair++) {
q->weighting_delay[js_pair][0] = 0;
q->weighting_delay[js_pair][1] = 7;
q->weighting_delay[js_pair][2] = 0;
q->weighting_delay[js_pair][3] = 7;
q->weighting_delay[js_pair][4] = 0;
q->weighting_delay[js_pair][5] = 7;
for (i = 0; i < 4; i++) {
q->matrix_coeff_index_prev[js_pair][i] = 3;
q->matrix_coeff_index_now[js_pair][i] = 3;
q->matrix_coeff_index_next[js_pair][i] = 3;
} |
10e26bc7 |
}
|
d49f3fa5 |
ff_atrac_init_gain_compensation(&q->gainc_ctx, 4, 3); |
94d68a41 |
q->fdsp = avpriv_float_dsp_alloc(avctx->flags & AV_CODEC_FLAG_BITEXACT); |
10e26bc7 |
|
606a49d2 |
q->units = av_mallocz_array(avctx->channels, sizeof(*q->units)); |
93f959b6 |
if (!q->units || !q->fdsp) { |
47b61702 |
atrac3_decode_close(avctx); |
6654296c |
return AVERROR(ENOMEM);
} |
10e26bc7 |
return 0;
}
|
e55d5390 |
AVCodec ff_atrac3_decoder = {
.name = "atrac3", |
7df9e693 |
.long_name = NULL_IF_CONFIG_SMALL("ATRAC3 (Adaptive TRansform Acoustic Coding 3)"), |
e55d5390 |
.type = AVMEDIA_TYPE_AUDIO,
.id = AV_CODEC_ID_ATRAC3,
.priv_data_size = sizeof(ATRAC3Context),
.init = atrac3_decode_init,
.close = atrac3_decode_close,
.decode = atrac3_decode_frame, |
def97856 |
.capabilities = AV_CODEC_CAP_SUBFRAMES | AV_CODEC_CAP_DR1, |
e55d5390 |
.sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_FLTP,
AV_SAMPLE_FMT_NONE }, |
10e26bc7 |
}; |
280a40dd |
AVCodec ff_atrac3al_decoder = {
.name = "atrac3al",
.long_name = NULL_IF_CONFIG_SMALL("ATRAC3 AL (Adaptive TRansform Acoustic Coding 3 Advanced Lossless)"),
.type = AVMEDIA_TYPE_AUDIO,
.id = AV_CODEC_ID_ATRAC3AL,
.priv_data_size = sizeof(ATRAC3Context),
.init = atrac3_decode_init,
.close = atrac3_decode_close,
.decode = atrac3al_decode_frame,
.capabilities = AV_CODEC_CAP_SUBFRAMES | AV_CODEC_CAP_DR1,
.sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_FLTP,
AV_SAMPLE_FMT_NONE },
}; |