Originally committed as revision 19375 to svn://svn.ffmpeg.org/ffmpeg/trunktags/v0.6
@@ -27,6 +27,7 @@ version <next>: | |||
- Electronic Arts Madcow decoder | |||
- DivX (XSUB) subtitle encoder | |||
- nonfree libamr support for AMR-NB/WB decoding/encoding removed | |||
- Experimental AAC encoder | |||
@@ -36,6 +36,7 @@ OBJS-$(CONFIG_VDPAU) += vdpau.o | |||
# decoders/encoders/hardware accelerators | |||
OBJS-$(CONFIG_AAC_DECODER) += aac.o aactab.o mpeg4audio.o aac_parser.o aac_ac3_parser.o | |||
OBJS-$(CONFIG_AAC_ENCODER) += aacenc.o aaccoder.o aacpsy.o aactab.o psymodel.o iirfilter.o mdct.o fft.o mpeg4audio.o | |||
OBJS-$(CONFIG_AASC_DECODER) += aasc.o msrledec.o | |||
OBJS-$(CONFIG_AC3_DECODER) += eac3dec.o ac3dec.o ac3tab.o ac3dec_data.o ac3.o | |||
OBJS-$(CONFIG_AC3_ENCODER) += ac3enc.o ac3tab.o ac3.o | |||
@@ -116,6 +116,12 @@ typedef struct { | |||
#define MAX_PREDICTORS 672 | |||
#define SCALE_DIV_512 36 ///< scalefactor difference that corresponds to scale difference in 512 times | |||
#define SCALE_ONE_POS 140 ///< scalefactor index that corresponds to scale=1.0 | |||
#define SCALE_MAX_POS 255 ///< scalefactor index maximum value | |||
#define SCALE_MAX_DIFF 60 ///< maximum scalefactor difference allowed by standard | |||
#define SCALE_DIFF_ZERO 60 ///< codebook index corresponding to zero scalefactor indices difference | |||
/** | |||
* Individual Channel Stream | |||
*/ | |||
@@ -126,6 +132,7 @@ typedef struct { | |||
int num_window_groups; | |||
uint8_t group_len[8]; | |||
const uint16_t *swb_offset; ///< table of offsets to the lowest spectral coefficient of a scalefactor band, sfb, for a particular window | |||
const uint8_t *swb_sizes; ///< table of scalefactor band sizes for a particular window | |||
int num_swb; ///< number of scalefactor window bands | |||
int num_windows; | |||
int tns_max_bands; | |||
@@ -165,6 +172,7 @@ typedef struct { | |||
typedef struct { | |||
int num_pulse; | |||
int start; | |||
int pos[4]; | |||
int amp[4]; | |||
} Pulse; | |||
@@ -189,11 +197,14 @@ typedef struct { | |||
typedef struct { | |||
IndividualChannelStream ics; | |||
TemporalNoiseShaping tns; | |||
enum BandType band_type[120]; ///< band types | |||
Pulse pulse; | |||
enum BandType band_type[128]; ///< band types | |||
int band_type_run_end[120]; ///< band type run end points | |||
float sf[120]; ///< scalefactors | |||
int sf_idx[128]; ///< scalefactor indices (used by encoder) | |||
uint8_t zeroes[128]; ///< band is not coded (used by encoder) | |||
DECLARE_ALIGNED_16(float, coeffs[1024]); ///< coefficients for IMDCT | |||
DECLARE_ALIGNED_16(float, saved[512]); ///< overlap | |||
DECLARE_ALIGNED_16(float, saved[1024]); ///< overlap | |||
DECLARE_ALIGNED_16(float, ret[1024]); ///< PCM output | |||
PredictorState predictor_state[MAX_PREDICTORS]; | |||
} SingleChannelElement; | |||
@@ -203,7 +214,9 @@ typedef struct { | |||
*/ | |||
typedef struct { | |||
// CPE specific | |||
uint8_t ms_mask[120]; ///< Set if mid/side stereo is used for each scalefactor window band | |||
int common_window; ///< Set if channels share a common 'IndividualChannelStream' in bitstream. | |||
int ms_mode; ///< Signals mid/side stereo flags coding mode (used by encoder) | |||
uint8_t ms_mask[128]; ///< Set if mid/side stereo is used for each scalefactor window band | |||
// shared | |||
SingleChannelElement ch[2]; | |||
// CCE specific | |||
@@ -26,19 +26,20 @@ | |||
/*********************************** | |||
* TODOs: | |||
* psy model selection with some option | |||
* add sane pulse detection | |||
* add temporal noise shaping | |||
***********************************/ | |||
#include "avcodec.h" | |||
#include "get_bits.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, | |||
@@ -83,7 +84,7 @@ static const uint8_t swb_size_1024_8[] = { | |||
32, 36, 36, 40, 44, 48, 52, 56, 60, 64, 80 | |||
}; | |||
static const uint8_t * const swb_size_1024[] = { | |||
static const uint8_t *swb_size_1024[] = { | |||
swb_size_1024_96, swb_size_1024_96, swb_size_1024_64, | |||
swb_size_1024_48, swb_size_1024_48, swb_size_1024_32, | |||
swb_size_1024_24, swb_size_1024_24, swb_size_1024_16, | |||
@@ -110,7 +111,7 @@ static const uint8_t swb_size_128_8[] = { | |||
4, 4, 4, 4, 4, 4, 4, 8, 8, 8, 8, 12, 16, 20, 20 | |||
}; | |||
static const uint8_t * const swb_size_128[] = { | |||
static const uint8_t *swb_size_128[] = { | |||
/* the last entry on the following row is swb_size_128_64 but is a | |||
duplicate of swb_size_128_96 */ | |||
swb_size_128_96, swb_size_128_96, swb_size_128_96, | |||
@@ -119,23 +120,6 @@ static const uint8_t * const swb_size_128[] = { | |||
swb_size_128_16, swb_size_128_16, swb_size_128_8 | |||
}; | |||
/** bits needed to code codebook run value for long windows */ | |||
static const uint8_t run_value_bits_long[64] = { | |||
5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, | |||
5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 10, | |||
10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, | |||
10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 15 | |||
}; | |||
/** bits needed to code codebook run value for short windows */ | |||
static const uint8_t run_value_bits_short[16] = { | |||
3, 3, 3, 3, 3, 3, 3, 6, 6, 6, 6, 6, 6, 6, 6, 9 | |||
}; | |||
static const uint8_t* const run_value_bits[2] = { | |||
run_value_bits_long, run_value_bits_short | |||
}; | |||
/** default channel configurations */ | |||
static const uint8_t aac_chan_configs[6][5] = { | |||
{1, TYPE_SCE}, // 1 channel - single channel element | |||
@@ -146,33 +130,6 @@ static const uint8_t aac_chan_configs[6][5] = { | |||
{4, TYPE_SCE, TYPE_CPE, TYPE_CPE, TYPE_LFE}, // 6 channels - front center + stereo + back stereo + LFE | |||
}; | |||
/** | |||
* structure used in optimal codebook search | |||
*/ | |||
typedef struct BandCodingPath { | |||
int prev_idx; ///< pointer to the previous path point | |||
int codebook; ///< codebook for coding band run | |||
int bits; ///< number of bit needed to code given number of bands | |||
} BandCodingPath; | |||
/** | |||
* AAC encoder context | |||
*/ | |||
typedef struct { | |||
PutBitContext pb; | |||
MDCTContext mdct1024; ///< long (1024 samples) frame transform context | |||
MDCTContext mdct128; ///< short (128 samples) frame transform context | |||
DSPContext dsp; | |||
DECLARE_ALIGNED_16(FFTSample, output[2048]); ///< temporary buffer for MDCT input coefficients | |||
int16_t* samples; ///< saved preprocessed input | |||
int samplerate_index; ///< MPEG-4 samplerate index | |||
ChannelElement *cpe; ///< channel elements | |||
AACPsyContext psy; ///< psychoacoustic model context | |||
int last_frame; | |||
} AACEncContext; | |||
/** | |||
* Make AAC audio config object. | |||
* @see 1.6.2.1 "Syntax - AudioSpecificConfig" | |||
@@ -197,6 +154,8 @@ static av_cold int aac_encode_init(AVCodecContext *avctx) | |||
{ | |||
AACEncContext *s = avctx->priv_data; | |||
int i; | |||
const uint8_t *sizes[2]; | |||
int lengths[2]; | |||
avctx->frame_size = 1024; | |||
@@ -224,25 +183,90 @@ static av_cold int aac_encode_init(AVCodecContext *avctx) | |||
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, | |||
swb_size_1024[i], ff_aac_num_swb_1024[i], swb_size_128[i], ff_aac_num_swb_128[i]) < 0){ | |||
av_log(avctx, AV_LOG_ERROR, "Cannot initialize selected model.\n"); | |||
return -1; | |||
} | |||
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(AACEncContext *s, IndividualChannelStream *info) | |||
{ | |||
int i; | |||
int w; | |||
put_bits(&s->pb, 1, 0); // ics_reserved bit | |||
put_bits(&s->pb, 2, info->window_sequence[0]); | |||
@@ -252,27 +276,118 @@ static void put_ics_info(AACEncContext *s, IndividualChannelStream *info) | |||
put_bits(&s->pb, 1, 0); // no prediction | |||
}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]); | |||
} | |||
} | |||
} | |||
/** | |||
* Calculate the number of bits needed to code all coefficient signs in current band. | |||
* Encode MS data. | |||
* @see 4.6.8.1 "Joint Coding - M/S Stereo" | |||
*/ | |||
static int calculate_band_sign_bits(AACEncContext *s, SingleChannelElement *sce, | |||
int group_len, int start, int size) | |||
static void encode_ms_info(PutBitContext *pb, ChannelElement *cpe) | |||
{ | |||
int bits = 0; | |||
int i, w; | |||
for(w = 0; w < group_len; w++){ | |||
for(i = 0; i < size; i++){ | |||
if(sce->icoefs[start + i]) | |||
bits++; | |||
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]); | |||
} | |||
} | |||
start += 128; | |||
} | |||
return bits; | |||
} | |||
/** | |||
@@ -298,27 +413,43 @@ static void encode_pulses(AACEncContext *s, Pulse *pulse) | |||
*/ | |||
static void encode_spectral_coeffs(AACEncContext *s, SingleChannelElement *sce) | |||
{ | |||
int start, i, w, w2, wg; | |||
int start, i, w, w2; | |||
w = 0; | |||
for(wg = 0; wg < sce->ics.num_window_groups; wg++){ | |||
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[wg]; w2++){ | |||
encode_band_coeffs(s, sce, start + w2*128, | |||
sce->ics.swb_sizes[i], | |||
sce->band_type[w*16 + 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, | |||
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]; | |||
} | |||
w += sce->ics.group_len[wg]; | |||
} | |||
} | |||
/** | |||
* 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. | |||
*/ | |||
@@ -339,13 +470,130 @@ static void put_bitstream_info(AVCodecContext *avctx, AACEncContext *s, const ch | |||
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]; | |||
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; | |||
} | |||
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[2]; | |||
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); | |||
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; | |||
} | |||
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; | |||
} | |||
if (avctx->frame_bits > 6144*avctx->channels) { | |||
av_log(avctx, AV_LOG_ERROR, "input buffer violation %d > %d.\n", avctx->frame_bits, 6144*avctx->channels); | |||
} | |||
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; | |||
@@ -0,0 +1,71 @@ | |||
/* | |||
* AAC encoder | |||
* Copyright (C) 2008 Konstantin Shishkov | |||
* | |||
* 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 | |||
*/ | |||
#ifndef AVCODEC_AACENC_H | |||
#define AVCODEC_AACENC_H | |||
#include "avcodec.h" | |||
#include "put_bits.h" | |||
#include "dsputil.h" | |||
#include "aac.h" | |||
#include "psymodel.h" | |||
struct AACEncContext; | |||
typedef struct AACCoefficientsEncoder{ | |||
void (*search_for_quantizers)(AVCodecContext *avctx, struct AACEncContext *s, | |||
SingleChannelElement *sce, const float lambda); | |||
void (*encode_window_bands_info)(struct AACEncContext *s, SingleChannelElement *sce, | |||
int win, int group_len, const float lambda); | |||
void (*quantize_and_encode_band)(struct AACEncContext *s, PutBitContext *pb, const float *in, int size, | |||
int scale_idx, int cb, const float lambda); | |||
void (*search_for_ms)(struct AACEncContext *s, ChannelElement *cpe, const float lambda); | |||
}AACCoefficientsEncoder; | |||
extern AACCoefficientsEncoder ff_aac_coders[]; | |||
/** | |||
* AAC encoder context | |||
*/ | |||
typedef struct AACEncContext { | |||
PutBitContext pb; | |||
MDCTContext mdct1024; ///< long (1024 samples) frame transform context | |||
MDCTContext mdct128; ///< short (128 samples) frame transform context | |||
DSPContext dsp; | |||
DECLARE_ALIGNED_16(FFTSample, output[2048]); ///< temporary buffer for MDCT input coefficients | |||
int16_t* samples; ///< saved preprocessed input | |||
int samplerate_index; ///< MPEG-4 samplerate index | |||
ChannelElement *cpe; ///< channel elements | |||
FFPsyContext psy; | |||
struct FFPsyPreprocessContext* psypp; | |||
AACCoefficientsEncoder *coder; | |||
int cur_channel; | |||
int last_frame; | |||
float lambda; | |||
DECLARE_ALIGNED_16(int, qcoefs[96][2]); ///< quantized coefficients | |||
DECLARE_ALIGNED_16(float, scoefs[1024]); ///< scaled coefficients | |||
} AACEncContext; | |||
#endif /* AVCODEC_AACENC_H */ |
@@ -25,54 +25,25 @@ | |||
*/ | |||
#include "avcodec.h" | |||
#include "aacpsy.h" | |||
#include "aactab.h" | |||
#include "psymodel.h" | |||
/*********************************** | |||
* TODOs: | |||
* General: | |||
* better audio preprocessing (add DC highpass filter?) | |||
* more psy models | |||
* maybe improve coefficient quantization function in some way | |||
* | |||
* 3GPP-based psy model: | |||
* thresholds linearization after their modifications for attaining given bitrate | |||
* try other bitrate controlling mechanism (maybe use ratecontrol.c?) | |||
* control quality for quality-based output | |||
**********************************/ | |||
/** | |||
* Quantize one coefficient. | |||
* @return absolute value of the quantized coefficient | |||
* @see 3GPP TS26.403 5.6.2 "Scalefactor determination" | |||
*/ | |||
static av_always_inline int quant(float coef, const float Q) | |||
{ | |||
return av_clip((int)(pow(fabsf(coef) * Q, 0.75) + 0.4054), 0, 8191); | |||
} | |||
static inline float get_approximate_quant_error(float *c, int size, int scale_idx) | |||
{ | |||
int i; | |||
int q; | |||
float coef, unquant, sum = 0.0f; | |||
const float Q = ff_aac_pow2sf_tab[200 - scale_idx + SCALE_ONE_POS - SCALE_DIV_512]; | |||
const float IQ = ff_aac_pow2sf_tab[200 + scale_idx - SCALE_ONE_POS + SCALE_DIV_512]; | |||
for(i = 0; i < size; i++){ | |||
coef = fabs(c[i]); | |||
q = quant(c[i], Q); | |||
unquant = (q * cbrt(q)) * IQ; | |||
sum += (coef - unquant) * (coef - unquant); | |||
} | |||
return sum; | |||
} | |||
/** | |||
* constants for 3GPP AAC psychoacoustic model | |||
* @{ | |||
*/ | |||
#define PSY_3GPP_SPREAD_LOW 1.5f // spreading factor for ascending threshold spreading (15 dB/Bark) | |||
#define PSY_3GPP_SPREAD_HI 3.0f // spreading factor for descending threshold spreading (30 dB/Bark) | |||
#define PSY_3GPP_RPEMIN 0.01f | |||
#define PSY_3GPP_RPELEV 2.0f | |||
/** | |||
* @} | |||
*/ | |||
@@ -83,8 +54,24 @@ static inline float get_approximate_quant_error(float *c, int size, int scale_id | |||
typedef struct Psy3gppBand{ | |||
float energy; ///< band energy | |||
float ffac; ///< form factor | |||
float thr; ///< energy threshold | |||
float min_snr; ///< minimal SNR | |||
float thr_quiet; ///< threshold in quiet | |||
}Psy3gppBand; | |||
/** | |||
* single/pair channel context for psychoacoustic model | |||
*/ | |||
typedef struct Psy3gppChannel{ | |||
Psy3gppBand band[128]; ///< bands information | |||
Psy3gppBand prev_band[128]; ///< bands information from the previous frame | |||
float win_energy; ///< sliding average of channel energy | |||
float iir_state[2]; ///< hi-pass IIR filter state | |||
uint8_t next_grouping; ///< stored grouping scheme for the next frame (in case of 8 short window sequence) | |||
enum WindowSequence next_window_seq; ///< window sequence to be used in the next frame | |||
}Psy3gppChannel; | |||
/** | |||
* psychoacoustic model frame type-dependent coefficients | |||
*/ | |||
@@ -95,10 +82,241 @@ typedef struct Psy3gppCoeffs{ | |||
float spread_hi [64]; ///< spreading factor for high-to-low threshold spreading in long frame | |||
}Psy3gppCoeffs; | |||
/** | |||
* 3GPP TS26.403-inspired psychoacoustic model specific data | |||
*/ | |||
typedef struct Psy3gppContext{ | |||
Psy3gppCoeffs psy_coef[2]; | |||
Psy3gppChannel *ch; | |||
}Psy3gppContext; | |||
/** | |||
* Calculate Bark value for given line. | |||
*/ | |||
static inline float calc_bark(float f) | |||
static av_cold float calc_bark(float f) | |||
{ | |||
return 13.3f * atanf(0.00076f * f) + 3.5f * atanf((f / 7500.0f) * (f / 7500.0f)); | |||
} | |||
#define ATH_ADD 4 | |||
/** | |||
* Calculate ATH value for given frequency. | |||
* Borrowed from Lame. | |||
*/ | |||
static av_cold float ath(float f, float add) | |||
{ | |||
f /= 1000.0f; | |||
return 3.64 * pow(f, -0.8) | |||
- 6.8 * exp(-0.6 * (f - 3.4) * (f - 3.4)) | |||
+ 6.0 * exp(-0.15 * (f - 8.7) * (f - 8.7)) | |||
+ (0.6 + 0.04 * add) * 0.001 * f * f * f * f; | |||
} | |||
static av_cold int psy_3gpp_init(FFPsyContext *ctx){ | |||
Psy3gppContext *pctx; | |||
float barks[1024]; | |||
int i, j, g, start; | |||
float prev, minscale, minath; | |||
ctx->model_priv_data = av_mallocz(sizeof(Psy3gppContext)); | |||
pctx = (Psy3gppContext*) ctx->model_priv_data; | |||
for(i = 0; i < 1024; i++) | |||
barks[i] = calc_bark(i * ctx->avctx->sample_rate / 2048.0); | |||
minath = ath(3410, ATH_ADD); | |||
for(j = 0; j < 2; j++){ | |||
Psy3gppCoeffs *coeffs = &pctx->psy_coef[j]; | |||
i = 0; | |||
prev = 0.0; | |||
for(g = 0; g < ctx->num_bands[j]; g++){ | |||
i += ctx->bands[j][g]; | |||
coeffs->barks[g] = (barks[i - 1] + prev) / 2.0; | |||
prev = barks[i - 1]; | |||
} | |||
for(g = 0; g < ctx->num_bands[j] - 1; g++){ | |||
coeffs->spread_low[g] = pow(10.0, -(coeffs->barks[g+1] - coeffs->barks[g]) * PSY_3GPP_SPREAD_LOW); | |||
coeffs->spread_hi [g] = pow(10.0, -(coeffs->barks[g+1] - coeffs->barks[g]) * PSY_3GPP_SPREAD_HI); | |||
} | |||
start = 0; | |||
for(g = 0; g < ctx->num_bands[j]; g++){ | |||
minscale = ath(ctx->avctx->sample_rate * start / 1024.0, ATH_ADD); | |||
for(i = 1; i < ctx->bands[j][g]; i++){ | |||
minscale = fminf(minscale, ath(ctx->avctx->sample_rate * (start + i) / 1024.0 / 2.0, ATH_ADD)); | |||
} | |||
coeffs->ath[g] = minscale - minath; | |||
start += ctx->bands[j][g]; | |||
} | |||
} | |||
pctx->ch = av_mallocz(sizeof(Psy3gppChannel) * ctx->avctx->channels); | |||
return 0; | |||
} | |||
/** | |||
* IIR filter used in block switching decision | |||
*/ | |||
static float iir_filter(int in, float state[2]) | |||
{ | |||
float ret; | |||
ret = 0.7548f * (in - state[0]) + 0.5095f * state[1]; | |||
state[0] = in; | |||
state[1] = ret; | |||
return ret; | |||
} | |||
/** | |||
* window grouping information stored as bits (0 - new group, 1 - group continues) | |||
*/ | |||
static const uint8_t window_grouping[9] = { | |||
0xB6, 0x6C, 0xD8, 0xB2, 0x66, 0xC6, 0x96, 0x36, 0x36 | |||
}; | |||
/** | |||
* Tell encoder which window types to use. | |||
* @see 3GPP TS26.403 5.4.1 "Blockswitching" | |||
*/ | |||
static FFPsyWindowInfo psy_3gpp_window(FFPsyContext *ctx, | |||
const int16_t *audio, const int16_t *la, | |||
int channel, int prev_type) | |||
{ | |||
int i, j; | |||
int br = ctx->avctx->bit_rate / ctx->avctx->channels; | |||
int attack_ratio = br <= 16000 ? 18 : 10; | |||
Psy3gppContext *pctx = (Psy3gppContext*) ctx->model_priv_data; | |||
Psy3gppChannel *pch = &pctx->ch[channel]; | |||
uint8_t grouping = 0; | |||
FFPsyWindowInfo wi; | |||
memset(&wi, 0, sizeof(wi)); | |||
if(la){ | |||
float s[8], v; | |||
int switch_to_eight = 0; | |||
float sum = 0.0, sum2 = 0.0; | |||
int attack_n = 0; | |||
for(i = 0; i < 8; i++){ | |||
for(j = 0; j < 128; j++){ | |||
v = iir_filter(audio[(i*128+j)*ctx->avctx->channels], pch->iir_state); | |||
sum += v*v; | |||
} | |||
s[i] = sum; | |||
sum2 += sum; | |||
} | |||
for(i = 0; i < 8; i++){ | |||
if(s[i] > pch->win_energy * attack_ratio){ | |||
attack_n = i + 1; | |||
switch_to_eight = 1; | |||
break; | |||
} | |||
} | |||
pch->win_energy = pch->win_energy*7/8 + sum2/64; | |||
wi.window_type[1] = prev_type; | |||
switch(prev_type){ | |||
case ONLY_LONG_SEQUENCE: | |||
wi.window_type[0] = switch_to_eight ? LONG_START_SEQUENCE : ONLY_LONG_SEQUENCE; | |||
break; | |||
case LONG_START_SEQUENCE: | |||
wi.window_type[0] = EIGHT_SHORT_SEQUENCE; | |||
grouping = pch->next_grouping; | |||
break; | |||
case LONG_STOP_SEQUENCE: | |||
wi.window_type[0] = ONLY_LONG_SEQUENCE; | |||
break; | |||
case EIGHT_SHORT_SEQUENCE: | |||
wi.window_type[0] = switch_to_eight ? EIGHT_SHORT_SEQUENCE : LONG_STOP_SEQUENCE; | |||
grouping = switch_to_eight ? pch->next_grouping : 0; | |||
break; | |||
} | |||
pch->next_grouping = window_grouping[attack_n]; | |||
}else{ | |||
for(i = 0; i < 3; i++) | |||
wi.window_type[i] = prev_type; | |||
grouping = (prev_type == EIGHT_SHORT_SEQUENCE) ? window_grouping[0] : 0; | |||
} | |||
wi.window_shape = 1; | |||
if(wi.window_type[0] != EIGHT_SHORT_SEQUENCE){ | |||
wi.num_windows = 1; | |||
wi.grouping[0] = 1; | |||
}else{ | |||
int lastgrp = 0; | |||
wi.num_windows = 8; | |||
for(i = 0; i < 8; i++){ | |||
if(!((grouping >> i) & 1)) | |||
lastgrp = i; | |||
wi.grouping[lastgrp]++; | |||
} | |||
} | |||
return wi; | |||
} | |||
/** | |||
* Calculate band thresholds as suggested in 3GPP TS26.403 | |||
*/ | |||
static void psy_3gpp_analyze(FFPsyContext *ctx, int channel, const float *coefs, | |||
FFPsyWindowInfo *wi) | |||
{ | |||
Psy3gppContext *pctx = (Psy3gppContext*) ctx->model_priv_data; | |||
Psy3gppChannel *pch = &pctx->ch[channel]; | |||
int start = 0; | |||
int i, w, g; | |||
const int num_bands = ctx->num_bands[wi->num_windows == 8]; | |||
const uint8_t* band_sizes = ctx->bands[wi->num_windows == 8]; | |||
Psy3gppCoeffs *coeffs = &pctx->psy_coef[wi->num_windows == 8]; | |||
//calculate energies, initial thresholds and related values - 5.4.2 "Threshold Calculation" | |||
for(w = 0; w < wi->num_windows*16; w += 16){ | |||
for(g = 0; g < num_bands; g++){ | |||
Psy3gppBand *band = &pch->band[w+g]; | |||
band->energy = 0.0f; | |||
for(i = 0; i < band_sizes[g]; i++) | |||
band->energy += coefs[start+i] * coefs[start+i]; | |||
band->energy *= 1.0f / (512*512); | |||
band->thr = band->energy * 0.001258925f; | |||
start += band_sizes[g]; | |||
ctx->psy_bands[channel*PSY_MAX_BANDS+w+g].energy = band->energy; | |||
} | |||
} | |||
//modify thresholds - spread, threshold in quiet - 5.4.3 "Spreaded Energy Calculation" | |||
for(w = 0; w < wi->num_windows*16; w += 16){ | |||
Psy3gppBand *band = &pch->band[w]; | |||
for(g = 1; g < num_bands; g++){ | |||
band[g].thr = FFMAX(band[g].thr, band[g-1].thr * coeffs->spread_low[g-1]); | |||
} | |||
for(g = num_bands - 2; g >= 0; g--){ | |||
band[g].thr = FFMAX(band[g].thr, band[g+1].thr * coeffs->spread_hi [g]); | |||
} | |||
for(g = 0; g < num_bands; g++){ | |||
band[g].thr_quiet = FFMAX(band[g].thr, coeffs->ath[g]); | |||
if(wi->num_windows != 8 && wi->window_type[1] != EIGHT_SHORT_SEQUENCE){ | |||
band[g].thr_quiet = fmaxf(PSY_3GPP_RPEMIN*band[g].thr_quiet, | |||
fminf(band[g].thr_quiet, | |||
PSY_3GPP_RPELEV*pch->prev_band[w+g].thr_quiet)); | |||
} | |||
band[g].thr = FFMAX(band[g].thr, band[g].thr_quiet * 0.25); | |||
ctx->psy_bands[channel*PSY_MAX_BANDS+w+g].threshold = band[g].thr; | |||
} | |||
} | |||
memcpy(pch->prev_band, pch->band, sizeof(pch->band)); | |||
} | |||
static av_cold void psy_3gpp_end(FFPsyContext *apc) | |||
{ | |||
Psy3gppContext *pctx = (Psy3gppContext*) apc->model_priv_data; | |||
av_freep(&pctx->ch); | |||
av_freep(&apc->model_priv_data); | |||
} | |||
const FFPsyModel ff_aac_psy_model = | |||
{ | |||
.name = "3GPP TS 26.403-inspired model", | |||
.init = psy_3gpp_init, | |||
.window = psy_3gpp_window, | |||
.analyze = psy_3gpp_analyze, | |||
.end = psy_3gpp_end, | |||
}; |
@@ -195,7 +195,7 @@ void avcodec_register_all(void) | |||
REGISTER_ENCDEC (ZMBV, zmbv); | |||
/* audio codecs */ | |||
REGISTER_DECODER (AAC, aac); | |||
REGISTER_ENCDEC (AAC, aac); | |||
REGISTER_ENCDEC (AC3, ac3); | |||
REGISTER_ENCDEC (ALAC, alac); | |||
REGISTER_DECODER (APE, ape); | |||
@@ -0,0 +1,130 @@ | |||
/* | |||
* audio encoder psychoacoustic model | |||
* Copyright (C) 2008 Konstantin Shishkov | |||
* | |||
* 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 | |||
*/ | |||
#include "avcodec.h" | |||
#include "psymodel.h" | |||
#include "iirfilter.h" | |||
extern const FFPsyModel ff_aac_psy_model; | |||
av_cold int ff_psy_init(FFPsyContext *ctx, AVCodecContext *avctx, | |||
int num_lens, | |||
const uint8_t **bands, const int* num_bands) | |||
{ | |||
ctx->avctx = avctx; | |||
ctx->psy_bands = av_mallocz(sizeof(FFPsyBand) * PSY_MAX_BANDS * avctx->channels); | |||
ctx->bands = av_malloc (sizeof(ctx->bands[0]) * num_lens); | |||
ctx->num_bands = av_malloc (sizeof(ctx->num_bands[0]) * num_lens); | |||
memcpy(ctx->bands, bands, sizeof(ctx->bands[0]) * num_lens); | |||
memcpy(ctx->num_bands, num_bands, sizeof(ctx->num_bands[0]) * num_lens); | |||
switch(ctx->avctx->codec_id){ | |||
case CODEC_ID_AAC: | |||
ctx->model = &ff_aac_psy_model; | |||
break; | |||
} | |||
if(ctx->model->init) | |||
return ctx->model->init(ctx); | |||
return 0; | |||
} | |||
FFPsyWindowInfo ff_psy_suggest_window(FFPsyContext *ctx, | |||
const int16_t *audio, const int16_t *la, | |||
int channel, int prev_type) | |||
{ | |||
return ctx->model->window(ctx, audio, la, channel, prev_type); | |||
} | |||
void ff_psy_set_band_info(FFPsyContext *ctx, int channel, | |||
const float *coeffs, FFPsyWindowInfo *wi) | |||
{ | |||
ctx->model->analyze(ctx, channel, coeffs, wi); | |||
} | |||
av_cold void ff_psy_end(FFPsyContext *ctx) | |||
{ | |||
if(ctx->model->end) | |||
ctx->model->end(ctx); | |||
av_freep(&ctx->bands); | |||
av_freep(&ctx->num_bands); | |||
av_freep(&ctx->psy_bands); | |||
} | |||
typedef struct FFPsyPreprocessContext{ | |||
AVCodecContext *avctx; | |||
float stereo_att; | |||
struct FFIIRFilterCoeffs *fcoeffs; | |||
struct FFIIRFilterState **fstate; | |||
}FFPsyPreprocessContext; | |||
#define FILT_ORDER 4 | |||
av_cold struct FFPsyPreprocessContext* ff_psy_preprocess_init(AVCodecContext *avctx) | |||
{ | |||
FFPsyPreprocessContext *ctx; | |||
int i; | |||
float cutoff_coeff; | |||
ctx = av_mallocz(sizeof(FFPsyPreprocessContext)); | |||
ctx->avctx = avctx; | |||
if(avctx->flags & CODEC_FLAG_QSCALE) | |||
cutoff_coeff = 1.0f / av_clip(1 + avctx->global_quality / FF_QUALITY_SCALE, 1, 8); | |||
else | |||
cutoff_coeff = avctx->bit_rate / (4.0f * avctx->sample_rate * avctx->channels); | |||
ctx->fcoeffs = ff_iir_filter_init_coeffs(FF_FILTER_TYPE_BUTTERWORTH, FF_FILTER_MODE_LOWPASS, | |||
FILT_ORDER, cutoff_coeff, 0.0, 0.0); | |||
if(ctx->fcoeffs){ | |||
ctx->fstate = av_mallocz(sizeof(ctx->fstate[0]) * avctx->channels); | |||
for(i = 0; i < avctx->channels; i++) | |||
ctx->fstate[i] = ff_iir_filter_init_state(FILT_ORDER); | |||
} | |||
return ctx; | |||
} | |||
void ff_psy_preprocess(struct FFPsyPreprocessContext *ctx, | |||
const int16_t *audio, int16_t *dest, | |||
int tag, int channels) | |||
{ | |||
int ch, i; | |||
if(ctx->fstate){ | |||
for(ch = 0; ch < channels; ch++){ | |||
ff_iir_filter(ctx->fcoeffs, ctx->fstate[tag+ch], ctx->avctx->frame_size, | |||
audio + ch, ctx->avctx->channels, | |||
dest + ch, ctx->avctx->channels); | |||
} | |||
}else{ | |||
for(ch = 0; ch < channels; ch++){ | |||
for(i = 0; i < ctx->avctx->frame_size; i++) | |||
dest[i*ctx->avctx->channels + ch] = audio[i*ctx->avctx->channels + ch]; | |||
} | |||
} | |||
} | |||
av_cold void ff_psy_preprocess_end(struct FFPsyPreprocessContext *ctx) | |||
{ | |||
int i; | |||
ff_iir_filter_free_coeffs(ctx->fcoeffs); | |||
if (ctx->fstate) | |||
for (i = 0; i < ctx->avctx->channels; i++) | |||
ff_iir_filter_free_state(ctx->fstate[i]); | |||
av_freep(&ctx->fstate); | |||
} | |||
@@ -0,0 +1,158 @@ | |||
/* | |||
* audio encoder psychoacoustic model | |||
* Copyright (C) 2008 Konstantin Shishkov | |||
* | |||
* 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 | |||
*/ | |||
#ifndef AVCODEC_PSYMODEL_H | |||
#define AVCODEC_PSYMODEL_H | |||
#include "avcodec.h" | |||
/** maximum possible number of bands */ | |||
#define PSY_MAX_BANDS 128 | |||
/** | |||
* single band psychoacoustic information | |||
*/ | |||
typedef struct FFPsyBand{ | |||
int bits; | |||
float energy; | |||
float threshold; | |||
float distortion; | |||
float perceptual_weight; | |||
}FFPsyBand; | |||
/** | |||
* windowing related information | |||
*/ | |||
typedef struct FFPsyWindowInfo{ | |||
int window_type[3]; ///< window type (short/long/transitional, etc.) - current, previous and next | |||
int window_shape; ///< window shape (sine/KBD/whatever) | |||
int num_windows; ///< number of windows in a frame | |||
int grouping[8]; ///< window grouping (for e.g. AAC) | |||
int *window_sizes; ///< sequence of window sizes inside one frame (for eg. WMA) | |||
}FFPsyWindowInfo; | |||
/** | |||
* context used by psychoacoustic model | |||
*/ | |||
typedef struct FFPsyContext{ | |||
AVCodecContext *avctx; ///< encoder context | |||
const struct FFPsyModel *model; ///< encoder-specific model functions | |||
FFPsyBand *psy_bands; ///< frame bands information | |||
uint8_t **bands; ///< scalefactor band sizes for possible frame sizes | |||
int *num_bands; ///< number of scalefactor bands for possible frame sizes | |||
int num_lens; ///< number of scalefactor band sets | |||
void* model_priv_data; ///< psychoacoustic model implementation private data | |||
}FFPsyContext; | |||
/** | |||
* codec-specific psychoacoustic model implementation | |||
*/ | |||
typedef struct FFPsyModel { | |||
const char *name; | |||
int (*init) (FFPsyContext *apc); | |||
FFPsyWindowInfo (*window)(FFPsyContext *ctx, const int16_t *audio, const int16_t *la, int channel, int prev_type); | |||
void (*analyze)(FFPsyContext *ctx, int channel, const float *coeffs, FFPsyWindowInfo *wi); | |||
void (*end) (FFPsyContext *apc); | |||
}FFPsyModel; | |||
/** | |||
* Initialize psychoacoustic model. | |||
* | |||
* @param ctx model context | |||
* @param avctx codec context | |||
* @param num_lens number of possible frame lengths | |||
* @param bands scalefactor band lengths for all frame lengths | |||
* @param num_bands number of scalefactor bands for all frame lengths | |||
* | |||
* @return zero if successful, a negative value if not | |||
*/ | |||
av_cold int ff_psy_init(FFPsyContext *ctx, AVCodecContext *avctx, | |||
int num_lens, | |||
const uint8_t **bands, const int* num_bands); | |||
/** | |||
* Suggest window sequence for channel. | |||
* | |||
* @param ctx model context | |||
* @param audio samples for the current frame | |||
* @param la lookahead samples (NULL when unavailable) | |||
* @param channel number of channel element to analyze | |||
* @param prev_type previous window type | |||
* | |||
* @return suggested window information in a structure | |||
*/ | |||
FFPsyWindowInfo ff_psy_suggest_window(FFPsyContext *ctx, | |||
const int16_t *audio, const int16_t *la, | |||
int channel, int prev_type); | |||
/** | |||
* Perform psychoacoustic analysis and set band info (threshold, energy). | |||
* | |||
* @param ctx model context | |||
* @param channel audio channel number | |||
* @param coeffs pointer to the transformed coefficients | |||
* @param wi window information | |||
*/ | |||
void ff_psy_set_band_info(FFPsyContext *ctx, int channel, const float *coeffs, | |||
FFPsyWindowInfo *wi); | |||
/** | |||
* Cleanup model context at the end. | |||
* | |||
* @param ctx model context | |||
*/ | |||
av_cold void ff_psy_end(FFPsyContext *ctx); | |||
/************************************************************************** | |||
* Audio preprocessing stuff. * | |||
* This should be moved into some audio filter eventually. * | |||
**************************************************************************/ | |||
struct FFPsyPreprocessContext; | |||
/** | |||
* psychoacoustic model audio preprocessing initialization | |||
*/ | |||
av_cold struct FFPsyPreprocessContext* ff_psy_preprocess_init(AVCodecContext *avctx); | |||
/** | |||
* Preprocess several channel in audio frame in order to compress it better. | |||
* | |||
* @param ctx preprocessing context | |||
* @param audio samples to preprocess | |||
* @param dest place to put filtered samples | |||
* @param tag channel number | |||
* @param channels number of channel to preprocess (some additional work may be done on stereo pair) | |||
*/ | |||
void ff_psy_preprocess(struct FFPsyPreprocessContext *ctx, | |||
const int16_t *audio, int16_t *dest, | |||
int tag, int channels); | |||
/** | |||
* Cleanup audio preprocessing module. | |||
*/ | |||
av_cold void ff_psy_preprocess_end(struct FFPsyPreprocessContext *ctx); | |||
#endif /* AVCODEC_PSYMODEL_H */ |