*/
/**
- * @file libavcodec/aacsbr.c
+ * @file
* AAC Spectral Band Replication decoding functions
* @author Robert Swain ( rob opendot cl )
*/
1.18509277094158210129f, //2^(0.49/2)
1.11987160404675912501f }; //2^(0.49/3)
const float lim_bands_per_octave_warped = bands_warped[sbr->bs_limiter_bands - 1];
- int16_t patch_borders[5];
+ int16_t patch_borders[7];
uint16_t *in = sbr->f_tablelim + 1, *out = sbr->f_tablelim;
patch_borders[0] = sbr->kx[1];
odd = (sb + sbr->k[0]) & 1;
}
+ // Requirements (14496-3 sp04 p205) sets the maximum number of patches to 5.
+ // After this check the final number of patches can still be six which is
+ // illegal however the Coding Technologies decoder check stream has a final
+ // count of 6 patches
+ if (sbr->num_patches > 5) {
+ av_log(ac->avccontext, AV_LOG_ERROR, "Too many patches: %d\n", sbr->num_patches);
+ return -1;
+ }
+
sbr->patch_num_subbands[sbr->num_patches] = FFMAX(sb - usb, 0);
sbr->patch_start_subband[sbr->num_patches] = sbr->k[0] - odd - sbr->patch_num_subbands[sbr->num_patches];
if (sbr->patch_num_subbands[sbr->num_patches-1] < 3 && sbr->num_patches > 1)
sbr->num_patches--;
- // Requirements (14496-3 sp04 p205) sets the maximum number of patches to 5
- // However the Coding Technologies decoder check uses 6 patches
- if (sbr->num_patches > 6) {
- av_log(ac->avccontext, AV_LOG_ERROR, "Too many patches: %d\n", sbr->num_patches);
- return -1;
- }
-
return 0;
}
GetBitContext *gb, SBRData *ch_data)
{
int i;
- unsigned bs_pointer;
+ unsigned bs_pointer = 0;
// frameLengthFlag ? 15 : 16; 960 sample length frames unsupported; this value is numTimeSlots
int abs_bord_trail = 16;
int num_rel_lead, num_rel_trail;
switch (ch_data->bs_frame_class = get_bits(gb, 2)) {
case FIXFIX:
- ch_data->bs_num_env = 1 << get_bits(gb, 2);
- num_rel_lead = ch_data->bs_num_env - 1;
+ ch_data->bs_num_env = 1 << get_bits(gb, 2);
+ num_rel_lead = ch_data->bs_num_env - 1;
if (ch_data->bs_num_env == 1)
ch_data->bs_amp_res = 0;
for (i = 0; i < num_rel_lead; i++)
ch_data->t_env[i + 1] = ch_data->t_env[i] + abs_bord_trail;
- bs_pointer = 0;
-
ch_data->bs_freq_res[1] = get_bits1(gb);
for (i = 1; i < ch_data->bs_num_env; i++)
ch_data->bs_freq_res[i + 1] = ch_data->bs_freq_res[1];
break;
case FIXVAR:
- abs_bord_trail += get_bits(gb, 2);
- num_rel_trail = get_bits(gb, 2);
- num_rel_lead = 0;
- ch_data->bs_num_env = num_rel_trail + 1;
- ch_data->t_env[0] = 0;
+ abs_bord_trail += get_bits(gb, 2);
+ num_rel_trail = get_bits(gb, 2);
+ ch_data->bs_num_env = num_rel_trail + 1;
+ ch_data->t_env[0] = 0;
ch_data->t_env[ch_data->bs_num_env] = abs_bord_trail;
for (i = 0; i < num_rel_trail; i++)
- ch_data->t_env[ch_data->bs_num_env - 1 - i] = ch_data->t_env[ch_data->bs_num_env - i] - 2 * get_bits(gb, 2) - 2;
+ ch_data->t_env[ch_data->bs_num_env - 1 - i] =
+ ch_data->t_env[ch_data->bs_num_env - i] - 2 * get_bits(gb, 2) - 2;
bs_pointer = get_bits(gb, ceil_log2[ch_data->bs_num_env]);
ch_data->bs_freq_res[ch_data->bs_num_env - i] = get_bits1(gb);
break;
case VARFIX:
- ch_data->t_env[0] = get_bits(gb, 2);
- num_rel_lead = get_bits(gb, 2);
- ch_data->bs_num_env = num_rel_lead + 1;
+ ch_data->t_env[0] = get_bits(gb, 2);
+ num_rel_lead = get_bits(gb, 2);
+ ch_data->bs_num_env = num_rel_lead + 1;
ch_data->t_env[ch_data->bs_num_env] = abs_bord_trail;
for (i = 0; i < num_rel_lead; i++)
get_bits1_vector(gb, ch_data->bs_freq_res + 1, ch_data->bs_num_env);
break;
case VARVAR:
- ch_data->t_env[0] = get_bits(gb, 2);
- abs_bord_trail += get_bits(gb, 2);
- num_rel_lead = get_bits(gb, 2);
- num_rel_trail = get_bits(gb, 2);
- ch_data->bs_num_env = num_rel_lead + num_rel_trail + 1;
- ch_data->t_env[ch_data->bs_num_env] = abs_bord_trail;
+ ch_data->t_env[0] = get_bits(gb, 2);
+ abs_bord_trail += get_bits(gb, 2);
+ num_rel_lead = get_bits(gb, 2);
+ num_rel_trail = get_bits(gb, 2);
+ ch_data->bs_num_env = num_rel_lead + num_rel_trail + 1;
if (ch_data->bs_num_env > 5) {
av_log(ac->avccontext, AV_LOG_ERROR,
return -1;
}
+ ch_data->t_env[ch_data->bs_num_env] = abs_bord_trail;
+
for (i = 0; i < num_rel_lead; i++)
ch_data->t_env[i + 1] = ch_data->t_env[i] + 2 * get_bits(gb, 2) + 2;
for (i = 0; i < num_rel_trail; i++)
- ch_data->t_env[ch_data->bs_num_env - 1 - i] = ch_data->t_env[ch_data->bs_num_env - i] - 2 * get_bits(gb, 2) - 2;
+ ch_data->t_env[ch_data->bs_num_env - 1 - i] =
+ ch_data->t_env[ch_data->bs_num_env - i] - 2 * get_bits(gb, 2) - 2;
bs_pointer = get_bits(gb, ceil_log2[ch_data->bs_num_env]);
return -1;
}
+ for (i = 1; i <= ch_data->bs_num_env; i++) {
+ if (ch_data->t_env[i-1] > ch_data->t_env[i]) {
+ av_log(ac->avccontext, AV_LOG_ERROR, "Non monotone time borders\n");
+ return -1;
+ }
+ }
+
ch_data->bs_num_noise = (ch_data->bs_num_env > 1) + 1;
- ch_data->t_q[0] = ch_data->t_env[0];
+ ch_data->t_q[0] = ch_data->t_env[0];
ch_data->t_q[ch_data->bs_num_noise] = ch_data->t_env[ch_data->bs_num_env];
if (ch_data->bs_num_noise > 1) {
unsigned int idx;
static void copy_sbr_grid(SBRData *dst, const SBRData *src) {
//These variables are saved from the previous frame rather than copied
- dst->bs_freq_res[0] = dst->bs_freq_res[dst->bs_num_env];
+ dst->bs_freq_res[0] = dst->bs_freq_res[dst->bs_num_env];
dst->t_env_num_env_old = dst->t_env[dst->bs_num_env];
- dst->e_a[0] = -(dst->e_a[1] != dst->bs_num_env);
+ dst->e_a[0] = -(dst->e_a[1] != dst->bs_num_env);
//These variables are read from the bitstream and therefore copied
memcpy(dst->bs_freq_res+1, src->bs_freq_res+1, sizeof(dst->bs_freq_res)-sizeof(*dst->bs_freq_res));
memcpy(dst->t_env, src->t_env, sizeof(dst->t_env));
memcpy(dst->t_q, src->t_q, sizeof(dst->t_q));
- dst->bs_num_env = src->bs_num_env;
- dst->bs_amp_res = src->bs_amp_res;
- dst->bs_num_noise = src->bs_num_noise;
- dst->bs_frame_class = src->bs_frame_class;
- dst->e_a[1] = src->e_a[1];
+ dst->bs_num_env = src->bs_num_env;
+ dst->bs_amp_res = src->bs_amp_res;
+ dst->bs_num_noise = src->bs_num_noise;
+ dst->bs_frame_class = src->bs_frame_class;
+ dst->e_a[1] = src->e_a[1];
}
/// Read how the envelope and noise floor data is delta coded
static void read_sbr_extension(AACContext *ac, SpectralBandReplication *sbr,
GetBitContext *gb,
- int bs_extension_id, int *num_bits_left)
+ int bs_extension_id, int *num_bits_left)
{
//TODO - implement ps_data for parametric stereo parsing
switch (bs_extension_id) {
case EXTENSION_ID_PS:
+ if (!ac->m4ac.ps) {
+ av_log(ac->avccontext, AV_LOG_ERROR, "Parametric Stereo signaled to be not-present but was found in the bitstream.\n");
+ skip_bits_long(gb, *num_bits_left); // bs_fill_bits
+ *num_bits_left = 0;
+ } else {
#if 0
- *num_bits_left -= ff_ps_data(gb, ps);
+ *num_bits_left -= ff_ps_data(gb, ps);
#else
- av_log_missing_feature(ac->avccontext, "Parametric Stereo is", 0);
- skip_bits_long(gb, *num_bits_left); // bs_fill_bits
- *num_bits_left = 0;
+ av_log_missing_feature(ac->avccontext, "Parametric Stereo is", 0);
+ skip_bits_long(gb, *num_bits_left); // bs_fill_bits
+ *num_bits_left = 0;
#endif
+ }
break;
default:
av_log_missing_feature(ac->avccontext, "Reserved SBR extensions are", 1);
*/
static void sbr_qmf_analysis(DSPContext *dsp, RDFTContext *rdft, const float *in, float *x,
float z[320], float W[2][32][32][2],
- float bias, float scale)
+ float scale)
{
int i, k;
memcpy(W[0], W[1], sizeof(W[0]));
memcpy(x , x+1024, (320-32)*sizeof(x[0]));
- if (scale != 1.0f || bias != 0.0f)
- for (i = 0; i < 1024; i++)
- x[288 + i] = (in[i] - bias) * scale;
+ if (scale != 1.0f)
+ dsp->vector_fmul_scalar(x+288, in, scale, 1024);
else
memcpy(x+288, in, 1024*sizeof(*x));
for (i = 0; i < 32; i++) { // numTimeSlots*RATE = 16*2 as 960 sample frames
ch_data->f_indexsine = indexsine;
}
-void ff_sbr_dequant(AACContext *ac, SpectralBandReplication *sbr, int id_aac)
-{
- if (sbr->start) {
- sbr_dequant(sbr, id_aac);
- }
-}
-
-void ff_sbr_apply(AACContext *ac, SpectralBandReplication *sbr, int ch,
- const float* in, float* out)
+void ff_sbr_apply(AACContext *ac, SpectralBandReplication *sbr, int id_aac,
+ float* L, float* R)
{
int downsampled = ac->m4ac.ext_sample_rate < sbr->sample_rate;
+ int ch;
+ int nch = (id_aac == TYPE_CPE) ? 2 : 1;
- /* decode channel */
- sbr_qmf_analysis(&ac->dsp, &sbr->rdft, in, sbr->data[ch].analysis_filterbank_samples,
- (float*)sbr->qmf_filter_scratch,
- sbr->data[ch].W, ac->add_bias, 1/(-1024 * ac->sf_scale));
- sbr_lf_gen(ac, sbr, sbr->X_low, sbr->data[ch].W);
if (sbr->start) {
- sbr_hf_inverse_filter(sbr->alpha0, sbr->alpha1, sbr->X_low, sbr->k[0]);
- sbr_chirp(sbr, &sbr->data[ch]);
- sbr_hf_gen(ac, sbr, sbr->X_high, sbr->X_low, sbr->alpha0, sbr->alpha1,
- sbr->data[ch].bw_array, sbr->data[ch].t_env,
- sbr->data[ch].bs_num_env);
-
- // hf_adj
- sbr_mapping(ac, sbr, &sbr->data[ch], sbr->data[ch].e_a);
- sbr_env_estimate(sbr->e_curr, sbr->X_high, sbr, &sbr->data[ch]);
- sbr_gain_calc(ac, sbr, &sbr->data[ch], sbr->data[ch].e_a);
- sbr_hf_assemble(sbr->data[ch].Y, sbr->X_high, sbr, &sbr->data[ch],
- sbr->data[ch].e_a);
+ sbr_dequant(sbr, id_aac);
}
+ for (ch = 0; ch < nch; ch++) {
+ /* decode channel */
+ sbr_qmf_analysis(&ac->dsp, &sbr->rdft, ch ? R : L, sbr->data[ch].analysis_filterbank_samples,
+ (float*)sbr->qmf_filter_scratch,
+ sbr->data[ch].W, 1/(-1024 * ac->sf_scale));
+ sbr_lf_gen(ac, sbr, sbr->X_low, sbr->data[ch].W);
+ if (sbr->start) {
+ sbr_hf_inverse_filter(sbr->alpha0, sbr->alpha1, sbr->X_low, sbr->k[0]);
+ sbr_chirp(sbr, &sbr->data[ch]);
+ sbr_hf_gen(ac, sbr, sbr->X_high, sbr->X_low, sbr->alpha0, sbr->alpha1,
+ sbr->data[ch].bw_array, sbr->data[ch].t_env,
+ sbr->data[ch].bs_num_env);
+
+ // hf_adj
+ sbr_mapping(ac, sbr, &sbr->data[ch], sbr->data[ch].e_a);
+ sbr_env_estimate(sbr->e_curr, sbr->X_high, sbr, &sbr->data[ch]);
+ sbr_gain_calc(ac, sbr, &sbr->data[ch], sbr->data[ch].e_a);
+ sbr_hf_assemble(sbr->data[ch].Y, sbr->X_high, sbr, &sbr->data[ch],
+ sbr->data[ch].e_a);
+ }
- /* synthesis */
- sbr_x_gen(sbr, sbr->X, sbr->X_low, sbr->data[ch].Y, ch);
- sbr_qmf_synthesis(&ac->dsp, &sbr->mdct, out, sbr->X, sbr->qmf_filter_scratch,
- sbr->data[ch].synthesis_filterbank_samples,
- &sbr->data[ch].synthesis_filterbank_samples_offset,
+ /* synthesis */
+ sbr_x_gen(sbr, sbr->X[ch], sbr->X_low, sbr->data[ch].Y, ch);
+ }
+ sbr_qmf_synthesis(&ac->dsp, &sbr->mdct, L, sbr->X[0], sbr->qmf_filter_scratch,
+ sbr->data[0].synthesis_filterbank_samples,
+ &sbr->data[0].synthesis_filterbank_samples_offset,
downsampled,
ac->add_bias, -1024 * ac->sf_scale);
+ if (nch == 2)
+ sbr_qmf_synthesis(&ac->dsp, &sbr->mdct, R, sbr->X[1], sbr->qmf_filter_scratch,
+ sbr->data[1].synthesis_filterbank_samples,
+ &sbr->data[1].synthesis_filterbank_samples_offset,
+ downsampled,
+ ac->add_bias, -1024 * ac->sf_scale);
}
projects / ffmpeg / blobdiff