/***********************************
* TODOs:
* add sane pulse detection
- * add temporal noise shaping
***********************************/
#include "libavutil/float_dsp.h"
AACEncContext *s = avctx->priv_data;
init_put_bits(&pb, avctx->extradata, avctx->extradata_size);
- put_bits(&pb, 5, 2); //object type - AAC-LC
+ put_bits(&pb, 5, s->profile+1); //profile
put_bits(&pb, 4, s->samplerate_index); //sample rate index
put_bits(&pb, 4, s->channels);
//GASpecificConfig
put_bits(&s->pb, 1, info->use_kb_window[0]);
if (info->window_sequence[0] != EIGHT_SHORT_SEQUENCE) {
put_bits(&s->pb, 6, info->max_sfb);
- put_bits(&s->pb, 1, 0); // no prediction
+ put_bits(&s->pb, 1, !!info->predictor_present);
} else {
put_bits(&s->pb, 4, info->max_sfb);
for (w = 1; w < 8; w++)
start += sce->ics.swb_sizes[i];
continue;
}
- for (w2 = w; w2 < w + sce->ics.group_len[w]; w2++)
- s->coder->quantize_and_encode_band(s, &s->pb, sce->coeffs + start + w2*128,
- sce->ics.swb_sizes[i],
+ for (w2 = w; w2 < w + sce->ics.group_len[w]; w2++) {
+ s->coder->quantize_and_encode_band(s, &s->pb,
+ &sce->coeffs[start + w2*128],
+ NULL, sce->ics.swb_sizes[i],
sce->sf_idx[w*16 + i],
sce->band_type[w*16 + i],
- s->lambda, sce->ics.window_clipping[w]);
+ s->lambda,
+ sce->ics.window_clipping[w]);
+ }
start += sce->ics.swb_sizes[i];
}
}
int common_window)
{
put_bits(&s->pb, 8, sce->sf_idx[0]);
- if (!common_window)
+ if (!common_window) {
put_ics_info(s, &sce->ics);
+ if (s->coder->encode_main_pred)
+ s->coder->encode_main_pred(s, sce);
+ }
encode_band_info(s, sce);
encode_scale_factors(avctx, s, sce);
encode_pulses(s, &sce->pulse);
- put_bits(&s->pb, 1, 0); //tns
+ put_bits(&s->pb, 1, !!sce->tns.present);
+ if (s->coder->encode_tns_info)
+ s->coder->encode_tns_info(s, sce);
put_bits(&s->pb, 1, 0); //ssr
encode_spectral_coeffs(s, sce);
return 0;
AACEncContext *s = avctx->priv_data;
float **samples = s->planar_samples, *samples2, *la, *overlap;
ChannelElement *cpe;
- int i, ch, w, g, chans, tag, start_ch, ret, ms_mode = 0, is_mode = 0;
+ SingleChannelElement *sce;
+ int i, ch, w, g, chans, tag, start_ch, ret;
+ int ms_mode = 0, is_mode = 0, tns_mode = 0, pred_mode = 0;
int chan_el_counter[4];
FFPsyWindowInfo windows[AAC_MAX_CHANNELS];
ics->swb_offset = wi[ch].window_type[0] == EIGHT_SHORT_SEQUENCE ?
ff_swb_offset_128 [s->samplerate_index]:
ff_swb_offset_1024[s->samplerate_index];
+ ics->tns_max_bands = wi[ch].window_type[0] == EIGHT_SHORT_SEQUENCE ?
+ ff_tns_max_bands_128 [s->samplerate_index]:
+ ff_tns_max_bands_1024[s->samplerate_index];
clip_avoidance_factor = 0.0f;
for (w = 0; w < ics->num_windows; w++)
ics->group_len[w] = wi[ch].grouping[w];
tag = s->chan_map[i+1];
chans = tag == TYPE_CPE ? 2 : 1;
cpe = &s->cpe[i];
+ cpe->common_window = 0;
memset(cpe->is_mask, 0, sizeof(cpe->is_mask));
memset(cpe->ms_mask, 0, sizeof(cpe->ms_mask));
put_bits(&s->pb, 3, tag);
for (ch = 0; ch < chans; ch++) {
sce = &cpe->ch[ch];
coeffs[ch] = sce->coeffs;
+ sce->ics.predictor_present = 0;
+ memset(&sce->ics.prediction_used, 0, sizeof(sce->ics.prediction_used));
+ memset(&sce->tns, 0, sizeof(TemporalNoiseShaping));
for (w = 0; w < 128; w++)
if (sce->band_type[w] > RESERVED_BT)
sce->band_type[w] = 0;
s->cur_channel = start_ch + ch;
s->coder->search_for_quantizers(avctx, s, &cpe->ch[ch], s->lambda);
}
- cpe->common_window = 0;
if (chans > 1
&& wi[0].window_type[0] == wi[1].window_type[0]
&& wi[0].window_shape == wi[1].window_shape) {
}
}
}
- if (s->options.pns && s->coder->search_for_pns) {
- for (ch = 0; ch < chans; ch++) {
- s->cur_channel = start_ch + ch;
- s->coder->search_for_pns(s, avctx, &cpe->ch[ch]);
- }
+ for (ch = 0; ch < chans; ch++) {
+ sce = &cpe->ch[ch];
+ s->cur_channel = start_ch + ch;
+ if (s->options.pns && s->coder->search_for_pns)
+ s->coder->search_for_pns(s, avctx, sce);
+ if (s->options.tns && s->coder->search_for_tns)
+ s->coder->search_for_tns(s, sce);
+ if (s->options.tns && s->coder->apply_tns_filt)
+ s->coder->apply_tns_filt(s, sce);
+ if (sce->tns.present)
+ tns_mode = 1;
}
s->cur_channel = start_ch;
if (s->options.stereo_mode && cpe->common_window) {
s->coder->search_for_ms(s, cpe);
}
}
- if (chans > 1 && s->options.intensity_stereo && s->coder->search_for_is) {
+ if (s->options.intensity_stereo && s->coder->search_for_is) {
s->coder->search_for_is(s, avctx, cpe);
if (cpe->is_mode) is_mode = 1;
}
for (ch = 0; ch < chans; ch++)
s->coder->set_special_band_scalefactors(s, &cpe->ch[ch]);
adjust_frame_information(cpe, chans);
+ for (ch = 0; ch < chans; ch++) {
+ sce = &cpe->ch[ch];
+ s->cur_channel = start_ch + ch;
+ if (s->options.pred && s->coder->search_for_pred)
+ s->coder->search_for_pred(s, sce);
+ if (cpe->ch[ch].ics.predictor_present) pred_mode = 1;
+ }
+ if (s->options.pred && s->coder->adjust_common_prediction)
+ s->coder->adjust_common_prediction(s, cpe);
+ for (ch = 0; ch < chans; ch++) {
+ sce = &cpe->ch[ch];
+ s->cur_channel = start_ch + ch;
+ if (s->options.pred && s->coder->apply_main_pred)
+ s->coder->apply_main_pred(s, sce);
+ }
+ s->cur_channel = start_ch;
if (chans == 2) {
put_bits(&s->pb, 1, cpe->common_window);
if (cpe->common_window) {
put_ics_info(s, &cpe->ch[0].ics);
+ if (s->coder->encode_main_pred)
+ s->coder->encode_main_pred(s, &cpe->ch[0]);
encode_ms_info(&s->pb, cpe);
if (cpe->ms_mode) ms_mode = 1;
}
s->psy.bitres.bits = frame_bits / s->channels;
break;
}
- if (is_mode || ms_mode) {
+ if (is_mode || ms_mode || tns_mode || pred_mode) {
for (i = 0; i < s->chan_map[0]; i++) {
// Must restore coeffs
chans = tag == TYPE_CPE ? 2 : 1;
ff_mdct_end(&s->mdct1024);
ff_mdct_end(&s->mdct128);
ff_psy_end(&s->psy);
+ ff_lpc_end(&s->lpc);
if (s->psypp)
ff_psy_preprocess_end(s->psypp);
av_freep(&s->buffer.samples);
s->channels = avctx->channels;
- ERROR_IF(i == 16 || i >= swb_size_1024_len || i >= swb_size_128_len,
+ ERROR_IF(i == 16 || i >= ff_aac_swb_size_1024_len || i >= ff_aac_swb_size_128_len,
"Unsupported sample rate %d\n", avctx->sample_rate);
ERROR_IF(s->channels > AAC_MAX_CHANNELS,
"Unsupported number of channels: %d\n", s->channels);
- ERROR_IF(avctx->profile != FF_PROFILE_UNKNOWN && avctx->profile != FF_PROFILE_AAC_LOW,
- "Unsupported profile %d\n", avctx->profile);
WARN_IF(1024.0 * avctx->bit_rate / avctx->sample_rate > 6144 * s->channels,
"Too many bits per frame requested, clamping to max\n");
+ if (avctx->profile == FF_PROFILE_AAC_MAIN) {
+ s->options.pred = 1;
+ } else if ((avctx->profile == FF_PROFILE_AAC_LOW ||
+ avctx->profile == FF_PROFILE_UNKNOWN) && s->options.pred) {
+ s->profile = 0; /* Main */
+ WARN_IF(1, "Prediction requested, changing profile to AAC-Main\n");
+ } else if (avctx->profile == FF_PROFILE_AAC_LOW ||
+ avctx->profile == FF_PROFILE_UNKNOWN) {
+ s->profile = 1; /* Low */
+ } else {
+ ERROR_IF(1, "Unsupported profile %d\n", avctx->profile);
+ }
avctx->bit_rate = (int)FFMIN(
6144 * s->channels / 1024.0 * avctx->sample_rate,
avctx->extradata_size = 5;
put_audio_specific_config(avctx);
- sizes[0] = swb_size_1024[i];
- sizes[1] = swb_size_128[i];
+ sizes[0] = ff_aac_swb_size_1024[i];
+ sizes[1] = ff_aac_swb_size_128[i];
lengths[0] = ff_aac_num_swb_1024[i];
lengths[1] = ff_aac_num_swb_128[i];
for (i = 0; i < s->chan_map[0]; i++)
goto fail;
s->psypp = ff_psy_preprocess_init(avctx);
s->coder = &ff_aac_coders[s->options.aac_coder];
+ ff_lpc_init(&s->lpc, 2*avctx->frame_size, TNS_MAX_ORDER, FF_LPC_TYPE_LEVINSON);
if (HAVE_MIPSDSPR1)
ff_aac_coder_init_mips(s);
{"aac_is", "Intensity stereo coding", offsetof(AACEncContext, options.intensity_stereo), AV_OPT_TYPE_INT, {.i64 = 0}, 0, 1, AACENC_FLAGS, "intensity_stereo"},
{"disable", "Disable intensity stereo coding", 0, AV_OPT_TYPE_CONST, {.i64 = 0}, INT_MIN, INT_MAX, AACENC_FLAGS, "intensity_stereo"},
{"enable", "Enable intensity stereo coding", 0, AV_OPT_TYPE_CONST, {.i64 = 1}, INT_MIN, INT_MAX, AACENC_FLAGS, "intensity_stereo"},
+ {"aac_tns", "Temporal noise shaping", offsetof(AACEncContext, options.tns), AV_OPT_TYPE_INT, {.i64 = 0}, 0, 1, AACENC_FLAGS, "aac_tns"},
+ {"disable", "Disable temporal noise shaping", 0, AV_OPT_TYPE_CONST, {.i64 = 0}, INT_MIN, INT_MAX, AACENC_FLAGS, "aac_tns"},
+ {"enable", "Enable temporal noise shaping", 0, AV_OPT_TYPE_CONST, {.i64 = 1}, INT_MIN, INT_MAX, AACENC_FLAGS, "aac_tns"},
+ {"aac_pred", "AAC-Main prediction", offsetof(AACEncContext, options.pred), AV_OPT_TYPE_INT, {.i64 = 0}, 0, 1, AACENC_FLAGS, "aac_pred"},
+ {"disable", "Disable AAC-Main prediction", 0, AV_OPT_TYPE_CONST, {.i64 = 0}, INT_MIN, INT_MAX, AACENC_FLAGS, "aac_pred"},
+ {"enable", "Enable AAC-Main prediction", 0, AV_OPT_TYPE_CONST, {.i64 = 1}, INT_MIN, INT_MAX, AACENC_FLAGS, "aac_pred"},
{NULL}
};
projects / ffmpeg / blobdiff