*/
/**
- * @file aacenc.c
+ * @file libavcodec/aacenc.c
* AAC encoder
*/
/***********************************
* TODOs:
- * psy model selection with some option
- * change greedy codebook search into something more optimal, like Viterbi algorithm
- * determine run lengths along with codebook
+ * add sane pulse detection
+ * add temporal noise shaping
***********************************/
#include "avcodec.h"
-#include "bitstream.h"
+#include "put_bits.h"
#include "dsputil.h"
#include "mpeg4audio.h"
-#include "aacpsy.h"
#include "aac.h"
#include "aactab.h"
+#include "aacenc.h"
+
+#include "psymodel.h"
static const uint8_t swb_size_1024_96[] = {
4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 8, 8, 8, 8, 8,
/** default channel configurations */
static const uint8_t aac_chan_configs[6][5] = {
- {1, ID_SCE}, // 1 channel - single channel element
- {1, ID_CPE}, // 2 channels - channel pair
- {2, ID_SCE, ID_CPE}, // 3 channels - center + stereo
- {3, ID_SCE, ID_CPE, ID_SCE}, // 4 channels - front center + stereo + back center
- {3, ID_SCE, ID_CPE, ID_CPE}, // 5 channels - front center + stereo + back stereo
- {4, ID_SCE, ID_CPE, ID_CPE, ID_LFE}, // 6 channels - front center + stereo + back stereo + LFE
+ {1, TYPE_SCE}, // 1 channel - single channel element
+ {1, TYPE_CPE}, // 2 channels - channel pair
+ {2, TYPE_SCE, TYPE_CPE}, // 3 channels - center + stereo
+ {3, TYPE_SCE, TYPE_CPE, TYPE_SCE}, // 4 channels - front center + stereo + back center
+ {3, TYPE_SCE, TYPE_CPE, TYPE_CPE}, // 5 channels - front center + stereo + back stereo
+ {4, TYPE_SCE, TYPE_CPE, TYPE_CPE, TYPE_LFE}, // 6 channels - front center + stereo + back stereo + LFE
};
/**
{
AACEncContext *s = avctx->priv_data;
int i;
+ const uint8_t *sizes[2];
+ int lengths[2];
avctx->frame_size = 1024;
- for(i = 0; i < 16; i++)
- if(avctx->sample_rate == ff_mpeg4audio_sample_rates[i])
+ for (i = 0; i < 16; i++)
+ if (avctx->sample_rate == ff_mpeg4audio_sample_rates[i])
break;
- if(i == 16){
+ if (i == 16) {
av_log(avctx, AV_LOG_ERROR, "Unsupported sample rate %d\n", avctx->sample_rate);
return -1;
}
- if(avctx->channels > 6){
+ if (avctx->channels > 6) {
av_log(avctx, AV_LOG_ERROR, "Unsupported number of channels: %d\n", avctx->channels);
return -1;
}
s->samplerate_index = i;
- s->swb_sizes1024 = swb_size_1024[i];
- s->swb_num1024 = ff_aac_num_swb_1024[i];
- s->swb_sizes128 = swb_size_128[i];
- s->swb_num128 = ff_aac_num_swb_128[i];
dsputil_init(&s->dsp, avctx);
- ff_mdct_init(&s->mdct1024, 11, 0);
- ff_mdct_init(&s->mdct128, 8, 0);
-
- s->samples = av_malloc(2 * 1024 * avctx->channels * sizeof(s->samples[0]));
- s->cpe = av_mallocz(sizeof(ChannelElement) * aac_chan_configs[avctx->channels-1][0]);
- if(ff_aac_psy_init(&s->psy, avctx, AAC_PSY_3GPP, aac_chan_configs[avctx->channels-1][0], 0, s->swb_sizes1024, s->swb_num1024, s->swb_sizes128, s->swb_num128) < 0){
- av_log(avctx, AV_LOG_ERROR, "Cannot initialize selected model.\n");
- return -1;
- }
- avctx->extradata = av_malloc(2);
+ ff_mdct_init(&s->mdct1024, 11, 0, 1.0);
+ ff_mdct_init(&s->mdct128, 8, 0, 1.0);
+ // window init
+ ff_kbd_window_init(ff_aac_kbd_long_1024, 4.0, 1024);
+ ff_kbd_window_init(ff_aac_kbd_short_128, 6.0, 128);
+ ff_init_ff_sine_windows(10);
+ ff_init_ff_sine_windows(7);
+
+ s->samples = av_malloc(2 * 1024 * avctx->channels * sizeof(s->samples[0]));
+ s->cpe = av_mallocz(sizeof(ChannelElement) * aac_chan_configs[avctx->channels-1][0]);
+ avctx->extradata = av_malloc(2);
avctx->extradata_size = 2;
put_audio_specific_config(avctx);
+
+ sizes[0] = swb_size_1024[i];
+ sizes[1] = swb_size_128[i];
+ lengths[0] = ff_aac_num_swb_1024[i];
+ lengths[1] = ff_aac_num_swb_128[i];
+ ff_psy_init(&s->psy, avctx, 2, sizes, lengths);
+ s->psypp = ff_psy_preprocess_init(avctx);
+ s->coder = &ff_aac_coders[0];
+
+ s->lambda = avctx->global_quality ? avctx->global_quality : 120;
+#if !CONFIG_HARDCODED_TABLES
+ for (i = 0; i < 428; i++)
+ ff_aac_pow2sf_tab[i] = pow(2, (i - 200)/4.);
+#endif /* CONFIG_HARDCODED_TABLES */
+
+ if (avctx->channels > 5)
+ av_log(avctx, AV_LOG_ERROR, "This encoder does not yet enforce the restrictions on LFEs. "
+ "The output will most likely be an illegal bitstream.\n");
+
return 0;
}
+static void apply_window_and_mdct(AVCodecContext *avctx, AACEncContext *s,
+ SingleChannelElement *sce, short *audio, int channel)
+{
+ int i, j, k;
+ const float * lwindow = sce->ics.use_kb_window[0] ? ff_aac_kbd_long_1024 : ff_sine_1024;
+ const float * swindow = sce->ics.use_kb_window[0] ? ff_aac_kbd_short_128 : ff_sine_128;
+ const float * pwindow = sce->ics.use_kb_window[1] ? ff_aac_kbd_short_128 : ff_sine_128;
+
+ if (sce->ics.window_sequence[0] != EIGHT_SHORT_SEQUENCE) {
+ memcpy(s->output, sce->saved, sizeof(float)*1024);
+ if (sce->ics.window_sequence[0] == LONG_STOP_SEQUENCE) {
+ memset(s->output, 0, sizeof(s->output[0]) * 448);
+ for (i = 448; i < 576; i++)
+ s->output[i] = sce->saved[i] * pwindow[i - 448];
+ for (i = 576; i < 704; i++)
+ s->output[i] = sce->saved[i];
+ }
+ if (sce->ics.window_sequence[0] != LONG_START_SEQUENCE) {
+ j = channel;
+ for (i = 0; i < 1024; i++, j += avctx->channels) {
+ s->output[i+1024] = audio[j] * lwindow[1024 - i - 1];
+ sce->saved[i] = audio[j] * lwindow[i];
+ }
+ } else {
+ j = channel;
+ for (i = 0; i < 448; i++, j += avctx->channels)
+ s->output[i+1024] = audio[j];
+ for (i = 448; i < 576; i++, j += avctx->channels)
+ s->output[i+1024] = audio[j] * swindow[576 - i - 1];
+ memset(s->output+1024+576, 0, sizeof(s->output[0]) * 448);
+ j = channel;
+ for (i = 0; i < 1024; i++, j += avctx->channels)
+ sce->saved[i] = audio[j];
+ }
+ ff_mdct_calc(&s->mdct1024, sce->coeffs, s->output);
+ } else {
+ j = channel;
+ for (k = 0; k < 1024; k += 128) {
+ for (i = 448 + k; i < 448 + k + 256; i++)
+ s->output[i - 448 - k] = (i < 1024)
+ ? sce->saved[i]
+ : audio[channel + (i-1024)*avctx->channels];
+ s->dsp.vector_fmul (s->output, k ? swindow : pwindow, 128);
+ s->dsp.vector_fmul_reverse(s->output+128, s->output+128, swindow, 128);
+ ff_mdct_calc(&s->mdct128, sce->coeffs + k, s->output);
+ }
+ j = channel;
+ for (i = 0; i < 1024; i++, j += avctx->channels)
+ sce->saved[i] = audio[j];
+ }
+}
+
/**
* Encode ics_info element.
* @see Table 4.6 (syntax of ics_info)
*/
-static void put_ics_info(AVCodecContext *avctx, IndividualChannelStream *info)
+static void put_ics_info(AACEncContext *s, IndividualChannelStream *info)
{
- AACEncContext *s = avctx->priv_data;
- int i;
+ int w;
put_bits(&s->pb, 1, 0); // ics_reserved bit
put_bits(&s->pb, 2, info->window_sequence[0]);
put_bits(&s->pb, 1, info->use_kb_window[0]);
- if(info->window_sequence[0] != EIGHT_SHORT_SEQUENCE){
+ if (info->window_sequence[0] != EIGHT_SHORT_SEQUENCE) {
put_bits(&s->pb, 6, info->max_sfb);
put_bits(&s->pb, 1, 0); // no prediction
- }else{
+ } else {
put_bits(&s->pb, 4, info->max_sfb);
- for(i = 1; i < info->num_windows; i++)
- put_bits(&s->pb, 1, info->group_len[i]);
+ for (w = 1; w < 8; w++)
+ put_bits(&s->pb, 1, !info->group_len[w]);
+ }
+}
+
+/**
+ * Encode MS data.
+ * @see 4.6.8.1 "Joint Coding - M/S Stereo"
+ */
+static void encode_ms_info(PutBitContext *pb, ChannelElement *cpe)
+{
+ int i, w;
+
+ put_bits(pb, 2, cpe->ms_mode);
+ if (cpe->ms_mode == 1)
+ for (w = 0; w < cpe->ch[0].ics.num_windows; w += cpe->ch[0].ics.group_len[w])
+ for (i = 0; i < cpe->ch[0].ics.max_sfb; i++)
+ put_bits(pb, 1, cpe->ms_mask[w*16 + i]);
+}
+
+/**
+ * Produce integer coefficients from scalefactors provided by the model.
+ */
+static void adjust_frame_information(AACEncContext *apc, ChannelElement *cpe, int chans)
+{
+ int i, w, w2, g, ch;
+ int start, sum, maxsfb, cmaxsfb;
+
+ for (ch = 0; ch < chans; ch++) {
+ IndividualChannelStream *ics = &cpe->ch[ch].ics;
+ start = 0;
+ maxsfb = 0;
+ cpe->ch[ch].pulse.num_pulse = 0;
+ for (w = 0; w < ics->num_windows*16; w += 16) {
+ for (g = 0; g < ics->num_swb; g++) {
+ sum = 0;
+ //apply M/S
+ if (!ch && cpe->ms_mask[w + g]) {
+ for (i = 0; i < ics->swb_sizes[g]; i++) {
+ cpe->ch[0].coeffs[start+i] = (cpe->ch[0].coeffs[start+i] + cpe->ch[1].coeffs[start+i]) / 2.0;
+ cpe->ch[1].coeffs[start+i] = cpe->ch[0].coeffs[start+i] - cpe->ch[1].coeffs[start+i];
+ }
+ }
+ start += ics->swb_sizes[g];
+ }
+ for (cmaxsfb = ics->num_swb; cmaxsfb > 0 && cpe->ch[ch].zeroes[w+cmaxsfb-1]; cmaxsfb--)
+ ;
+ maxsfb = FFMAX(maxsfb, cmaxsfb);
+ }
+ ics->max_sfb = maxsfb;
+
+ //adjust zero bands for window groups
+ for (w = 0; w < ics->num_windows; w += ics->group_len[w]) {
+ for (g = 0; g < ics->max_sfb; g++) {
+ i = 1;
+ for (w2 = w; w2 < w + ics->group_len[w]; w2++) {
+ if (!cpe->ch[ch].zeroes[w2*16 + g]) {
+ i = 0;
+ break;
+ }
+ }
+ cpe->ch[ch].zeroes[w*16 + g] = i;
+ }
+ }
+ }
+
+ if (chans > 1 && cpe->common_window) {
+ IndividualChannelStream *ics0 = &cpe->ch[0].ics;
+ IndividualChannelStream *ics1 = &cpe->ch[1].ics;
+ int msc = 0;
+ ics0->max_sfb = FFMAX(ics0->max_sfb, ics1->max_sfb);
+ ics1->max_sfb = ics0->max_sfb;
+ for (w = 0; w < ics0->num_windows*16; w += 16)
+ for (i = 0; i < ics0->max_sfb; i++)
+ if (cpe->ms_mask[w+i])
+ msc++;
+ if (msc == 0 || ics0->max_sfb == 0)
+ cpe->ms_mode = 0;
+ else
+ cpe->ms_mode = msc < ics0->max_sfb ? 1 : 2;
+ }
+}
+
+/**
+ * Encode scalefactor band coding type.
+ */
+static void encode_band_info(AACEncContext *s, SingleChannelElement *sce)
+{
+ int w;
+
+ for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w])
+ s->coder->encode_window_bands_info(s, sce, w, sce->ics.group_len[w], s->lambda);
+}
+
+/**
+ * Encode scalefactors.
+ */
+static void encode_scale_factors(AVCodecContext *avctx, AACEncContext *s,
+ SingleChannelElement *sce)
+{
+ int off = sce->sf_idx[0], diff;
+ int i, w;
+
+ for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
+ for (i = 0; i < sce->ics.max_sfb; i++) {
+ if (!sce->zeroes[w*16 + i]) {
+ diff = sce->sf_idx[w*16 + i] - off + SCALE_DIFF_ZERO;
+ if (diff < 0 || diff > 120)
+ av_log(avctx, AV_LOG_ERROR, "Scalefactor difference is too big to be coded\n");
+ off = sce->sf_idx[w*16 + i];
+ put_bits(&s->pb, ff_aac_scalefactor_bits[diff], ff_aac_scalefactor_code[diff]);
+ }
+ }
}
}
+/**
+ * Encode pulse data.
+ */
+static void encode_pulses(AACEncContext *s, Pulse *pulse)
+{
+ int i;
+
+ put_bits(&s->pb, 1, !!pulse->num_pulse);
+ if (!pulse->num_pulse)
+ return;
+
+ put_bits(&s->pb, 2, pulse->num_pulse - 1);
+ put_bits(&s->pb, 6, pulse->start);
+ for (i = 0; i < pulse->num_pulse; i++) {
+ put_bits(&s->pb, 5, pulse->pos[i]);
+ put_bits(&s->pb, 4, pulse->amp[i]);
+ }
+}
+
+/**
+ * Encode spectral coefficients processed by psychoacoustic model.
+ */
+static void encode_spectral_coeffs(AACEncContext *s, SingleChannelElement *sce)
+{
+ int start, i, w, w2;
+
+ for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) {
+ start = 0;
+ for (i = 0; i < sce->ics.max_sfb; i++) {
+ if (sce->zeroes[w*16 + i]) {
+ 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],
+ sce->sf_idx[w*16 + i],
+ sce->band_type[w*16 + i],
+ s->lambda);
+ start += sce->ics.swb_sizes[i];
+ }
+ }
+}
+
+/**
+ * Encode one channel of audio data.
+ */
+static int encode_individual_channel(AVCodecContext *avctx, AACEncContext *s,
+ SingleChannelElement *sce,
+ int common_window)
+{
+ put_bits(&s->pb, 8, sce->sf_idx[0]);
+ if (!common_window)
+ put_ics_info(s, &sce->ics);
+ 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, 0); //ssr
+ encode_spectral_coeffs(s, sce);
+ return 0;
+}
+
/**
* Write some auxiliary information about the created AAC file.
*/
-static void put_bitstream_info(AVCodecContext *avctx, AACEncContext *s, const char *name)
+static void put_bitstream_info(AVCodecContext *avctx, AACEncContext *s,
+ const char *name)
{
int i, namelen, padbits;
namelen = strlen(name) + 2;
- put_bits(&s->pb, 3, ID_FIL);
+ put_bits(&s->pb, 3, TYPE_FIL);
put_bits(&s->pb, 4, FFMIN(namelen, 15));
- if(namelen >= 15)
+ if (namelen >= 15)
put_bits(&s->pb, 8, namelen - 16);
put_bits(&s->pb, 4, 0); //extension type - filler
padbits = 8 - (put_bits_count(&s->pb) & 7);
align_put_bits(&s->pb);
- for(i = 0; i < namelen - 2; i++)
+ for (i = 0; i < namelen - 2; i++)
put_bits(&s->pb, 8, name[i]);
put_bits(&s->pb, 12 - padbits, 0);
}
+static int aac_encode_frame(AVCodecContext *avctx,
+ uint8_t *frame, int buf_size, void *data)
+{
+ AACEncContext *s = avctx->priv_data;
+ int16_t *samples = s->samples, *samples2, *la;
+ ChannelElement *cpe;
+ int i, j, chans, tag, start_ch;
+ const uint8_t *chan_map = aac_chan_configs[avctx->channels-1];
+ int chan_el_counter[4];
+ FFPsyWindowInfo windows[avctx->channels];
+
+ if (s->last_frame)
+ return 0;
+ if (data) {
+ if (!s->psypp) {
+ memcpy(s->samples + 1024 * avctx->channels, data,
+ 1024 * avctx->channels * sizeof(s->samples[0]));
+ } else {
+ start_ch = 0;
+ samples2 = s->samples + 1024 * avctx->channels;
+ for (i = 0; i < chan_map[0]; i++) {
+ tag = chan_map[i+1];
+ chans = tag == TYPE_CPE ? 2 : 1;
+ ff_psy_preprocess(s->psypp, (uint16_t*)data + start_ch,
+ samples2 + start_ch, start_ch, chans);
+ start_ch += chans;
+ }
+ }
+ }
+ if (!avctx->frame_number) {
+ memcpy(s->samples, s->samples + 1024 * avctx->channels,
+ 1024 * avctx->channels * sizeof(s->samples[0]));
+ return 0;
+ }
+
+ start_ch = 0;
+ for (i = 0; i < chan_map[0]; i++) {
+ FFPsyWindowInfo* wi = windows + start_ch;
+ tag = chan_map[i+1];
+ chans = tag == TYPE_CPE ? 2 : 1;
+ cpe = &s->cpe[i];
+ samples2 = samples + start_ch;
+ la = samples2 + 1024 * avctx->channels + start_ch;
+ if (!data)
+ la = NULL;
+ for (j = 0; j < chans; j++) {
+ IndividualChannelStream *ics = &cpe->ch[j].ics;
+ int k;
+ wi[j] = ff_psy_suggest_window(&s->psy, samples2, la, start_ch + j, ics->window_sequence[0]);
+ ics->window_sequence[1] = ics->window_sequence[0];
+ ics->window_sequence[0] = wi[j].window_type[0];
+ ics->use_kb_window[1] = ics->use_kb_window[0];
+ ics->use_kb_window[0] = wi[j].window_shape;
+ ics->num_windows = wi[j].num_windows;
+ ics->swb_sizes = s->psy.bands [ics->num_windows == 8];
+ ics->num_swb = s->psy.num_bands[ics->num_windows == 8];
+ for (k = 0; k < ics->num_windows; k++)
+ ics->group_len[k] = wi[j].grouping[k];
+
+ s->cur_channel = start_ch + j;
+ apply_window_and_mdct(avctx, s, &cpe->ch[j], samples2, j);
+ }
+ start_ch += chans;
+ }
+ do {
+ int frame_bits;
+ init_put_bits(&s->pb, frame, buf_size*8);
+ if ((avctx->frame_number & 0xFF)==1 && !(avctx->flags & CODEC_FLAG_BITEXACT))
+ put_bitstream_info(avctx, s, LIBAVCODEC_IDENT);
+ start_ch = 0;
+ memset(chan_el_counter, 0, sizeof(chan_el_counter));
+ for (i = 0; i < chan_map[0]; i++) {
+ FFPsyWindowInfo* wi = windows + start_ch;
+ tag = chan_map[i+1];
+ chans = tag == TYPE_CPE ? 2 : 1;
+ cpe = &s->cpe[i];
+ for (j = 0; j < chans; j++) {
+ s->coder->search_for_quantizers(avctx, s, &cpe->ch[j], 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) {
+
+ cpe->common_window = 1;
+ for (j = 0; j < wi[0].num_windows; j++) {
+ if (wi[0].grouping[j] != wi[1].grouping[j]) {
+ cpe->common_window = 0;
+ break;
+ }
+ }
+ }
+ if (cpe->common_window && s->coder->search_for_ms)
+ s->coder->search_for_ms(s, cpe, s->lambda);
+ adjust_frame_information(s, cpe, chans);
+ put_bits(&s->pb, 3, tag);
+ put_bits(&s->pb, 4, chan_el_counter[tag]++);
+ if (chans == 2) {
+ put_bits(&s->pb, 1, cpe->common_window);
+ if (cpe->common_window) {
+ put_ics_info(s, &cpe->ch[0].ics);
+ encode_ms_info(&s->pb, cpe);
+ }
+ }
+ for (j = 0; j < chans; j++) {
+ s->cur_channel = start_ch + j;
+ ff_psy_set_band_info(&s->psy, s->cur_channel, cpe->ch[j].coeffs, &wi[j]);
+ encode_individual_channel(avctx, s, &cpe->ch[j], cpe->common_window);
+ }
+ start_ch += chans;
+ }
+
+ frame_bits = put_bits_count(&s->pb);
+ if (frame_bits <= 6144 * avctx->channels - 3)
+ break;
+
+ s->lambda *= avctx->bit_rate * 1024.0f / avctx->sample_rate / frame_bits;
+
+ } while (1);
+
+ put_bits(&s->pb, 3, TYPE_END);
+ flush_put_bits(&s->pb);
+ avctx->frame_bits = put_bits_count(&s->pb);
+
+ // rate control stuff
+ if (!(avctx->flags & CODEC_FLAG_QSCALE)) {
+ float ratio = avctx->bit_rate * 1024.0f / avctx->sample_rate / avctx->frame_bits;
+ s->lambda *= ratio;
+ s->lambda = FFMIN(s->lambda, 65536.f);
+ }
+
+ if (!data)
+ s->last_frame = 1;
+ memcpy(s->samples, s->samples + 1024 * avctx->channels,
+ 1024 * avctx->channels * sizeof(s->samples[0]));
+ return put_bits_count(&s->pb)>>3;
+}
+
static av_cold int aac_encode_end(AVCodecContext *avctx)
{
AACEncContext *s = avctx->priv_data;
ff_mdct_end(&s->mdct1024);
ff_mdct_end(&s->mdct128);
- ff_aac_psy_end(&s->psy);
+ ff_psy_end(&s->psy);
+ ff_psy_preprocess_end(s->psypp);
av_freep(&s->samples);
av_freep(&s->cpe);
return 0;
aac_encode_frame,
aac_encode_end,
.capabilities = CODEC_CAP_SMALL_LAST_FRAME | CODEC_CAP_DELAY,
- .sample_fmts = (enum SampleFormat[]){SAMPLE_FMT_S16,SAMPLE_FMT_NONE},
+ .sample_fmts = (const enum SampleFormat[]){SAMPLE_FMT_S16,SAMPLE_FMT_NONE},
.long_name = NULL_IF_CONFIG_SMALL("Advanced Audio Coding"),
};