ref: b5b68e541c7164b5be892b3f56d2d2b2a2f511a2
dir: /libfaad/decoder.c/
/*
** FAAD - Freeware Advanced Audio Decoder
** Copyright (C) 2002 M. Bakker
**
** This program is free software; you can redistribute it and/or modify
** it under the terms of the GNU General Public License as published by
** the Free Software Foundation; either version 2 of the License, or
** (at your option) any later version.
**
** This program 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 General Public License for more details.
**
** You should have received a copy of the GNU General Public License
** along with this program; if not, write to the Free Software
** Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
**
** $Id: decoder.c,v 1.46 2002/12/22 19:58:31 menno Exp $
**/
#include "common.h"
#include "structs.h"
#include <stdlib.h>
#include <string.h>
#include "decoder.h"
#include "mp4.h"
#include "syntax.h"
#include "specrec.h"
#include "data.h"
#include "tns.h"
#include "pns.h"
#include "is.h"
#include "ms.h"
#include "ic_predict.h"
#include "lt_predict.h"
#include "drc.h"
#include "error.h"
#include "output.h"
#include "dither.h"
#ifdef SSR_DEC
#include "ssr.h"
#include "ssr_fb.h"
#endif
#ifdef ANALYSIS
uint16_t dbg_count;
#endif
int8_t* FAADAPI faacDecGetErrorMessage(uint8_t errcode)
{
return err_msg[errcode];
}
faacDecHandle FAADAPI faacDecOpen()
{
uint8_t i;
faacDecHandle hDecoder = NULL;
if ((hDecoder = (faacDecHandle)malloc(sizeof(faacDecStruct))) == NULL)
return NULL;
memset(hDecoder, 0, sizeof(faacDecStruct));
hDecoder->config.outputFormat = FAAD_FMT_16BIT;
hDecoder->config.defObjectType = MAIN;
hDecoder->config.defSampleRate = 44100; /* Default: 44.1kHz */
hDecoder->adts_header_present = 0;
hDecoder->adif_header_present = 0;
#ifdef ERROR_RESILIENCE
hDecoder->aacSectionDataResilienceFlag = 0;
hDecoder->aacScalefactorDataResilienceFlag = 0;
hDecoder->aacSpectralDataResilienceFlag = 0;
#endif
hDecoder->frameLength = 1024;
hDecoder->frame = 0;
hDecoder->sample_buffer = NULL;
for (i = 0; i < MAX_CHANNELS; i++)
{
hDecoder->window_shape_prev[i] = 0;
hDecoder->time_out[i] = NULL;
#ifdef SSR_DEC
hDecoder->ssr_overlap[i] = NULL;
hDecoder->prev_fmd[i] = NULL;
#endif
#ifdef MAIN_DEC
hDecoder->pred_stat[i] = NULL;
#endif
#ifdef LTP_DEC
hDecoder->ltp_lag[i] = 0;
hDecoder->lt_pred_stat[i] = NULL;
#endif
}
hDecoder->drc = drc_init(REAL_CONST(1.0), REAL_CONST(1.0));
#if POW_TABLE_SIZE
hDecoder->pow2_table = (real_t*)malloc(POW_TABLE_SIZE*sizeof(real_t));
build_tables(hDecoder->pow2_table);
#endif
return hDecoder;
}
faacDecConfigurationPtr FAADAPI faacDecGetCurrentConfiguration(faacDecHandle hDecoder)
{
faacDecConfigurationPtr config = &(hDecoder->config);
return config;
}
uint8_t FAADAPI faacDecSetConfiguration(faacDecHandle hDecoder,
faacDecConfigurationPtr config)
{
hDecoder->config.defObjectType = config->defObjectType;
hDecoder->config.defSampleRate = config->defSampleRate;
hDecoder->config.outputFormat = config->outputFormat;
/* OK */
return 1;
}
int32_t FAADAPI faacDecInit(faacDecHandle hDecoder, uint8_t *buffer,
uint32_t buffer_size,
uint32_t *samplerate, uint8_t *channels)
{
uint32_t bits = 0;
bitfile ld;
adif_header adif;
adts_header adts;
hDecoder->sf_index = get_sr_index(hDecoder->config.defSampleRate);
hDecoder->object_type = hDecoder->config.defObjectType;
*samplerate = sample_rates[hDecoder->sf_index];
*channels = 1;
if (buffer != NULL)
{
faad_initbits(&ld, buffer, buffer_size);
/* Check if an ADIF header is present */
if ((buffer[0] == 'A') && (buffer[1] == 'D') &&
(buffer[2] == 'I') && (buffer[3] == 'F'))
{
hDecoder->adif_header_present = 1;
get_adif_header(&adif, &ld);
faad_byte_align(&ld);
hDecoder->sf_index = adif.pce.sf_index;
hDecoder->object_type = adif.pce.object_type;
*samplerate = sample_rates[hDecoder->sf_index];
*channels = adif.pce.channels;
bits = bit2byte(faad_get_processed_bits(&ld));
/* Check if an ADTS header is present */
} else if (faad_showbits(&ld, 12) == 0xfff) {
hDecoder->adts_header_present = 1;
adts_frame(&adts, &ld);
hDecoder->sf_index = adts.sf_index;
hDecoder->object_type = adts.profile;
*samplerate = sample_rates[hDecoder->sf_index];
*channels = (adts.channel_configuration > 6) ?
2 : adts.channel_configuration;
}
faad_endbits(&ld);
}
hDecoder->channelConfiguration = *channels;
/* must be done before frameLength is divided by 2 for LD */
#ifdef SSR_DEC
if (hDecoder->object_type == SSR)
hDecoder->fb = ssr_filter_bank_init(hDecoder->frameLength/SSR_BANDS);
else
#endif
hDecoder->fb = filter_bank_init(hDecoder->frameLength);
#ifdef LD_DEC
if (hDecoder->object_type == LD)
hDecoder->frameLength >>= 1;
#endif
if (can_decode_ot(hDecoder->object_type) < 0)
return -1;
#ifndef FIXED_POINT
if (hDecoder->config.outputFormat >= 5)
Init_Dither(16, hDecoder->config.outputFormat - 5);
#endif
return bits;
}
/* Init the library using a DecoderSpecificInfo */
int8_t FAADAPI faacDecInit2(faacDecHandle hDecoder, uint8_t *pBuffer,
uint32_t SizeOfDecoderSpecificInfo,
uint32_t *samplerate, uint8_t *channels)
{
int8_t rc;
uint8_t frameLengthFlag;
hDecoder->adif_header_present = 0;
hDecoder->adts_header_present = 0;
if((hDecoder == NULL)
|| (pBuffer == NULL)
|| (SizeOfDecoderSpecificInfo < 2)
|| (samplerate == NULL)
|| (channels == NULL))
{
return -1;
}
rc = AudioSpecificConfig(pBuffer, SizeOfDecoderSpecificInfo,
samplerate, channels,
&hDecoder->sf_index, &hDecoder->object_type,
#ifdef ERROR_RESILIENCE
&hDecoder->aacSectionDataResilienceFlag,
&hDecoder->aacScalefactorDataResilienceFlag,
&hDecoder->aacSpectralDataResilienceFlag,
#else
NULL, NULL, NULL,
#endif
&frameLengthFlag);
if (hDecoder->object_type < 5)
hDecoder->object_type--; /* For AAC differs from MPEG-4 */
if (rc != 0)
{
return rc;
}
hDecoder->channelConfiguration = *channels;
if (frameLengthFlag)
hDecoder->frameLength = 960;
/* must be done before frameLength is divided by 2 for LD */
#ifdef SSR_DEC
if (hDecoder->object_type == SSR)
hDecoder->fb = ssr_filter_bank_init(hDecoder->frameLength/SSR_BANDS);
else
#endif
hDecoder->fb = filter_bank_init(hDecoder->frameLength);
#ifdef LD_DEC
if (hDecoder->object_type == LD)
hDecoder->frameLength >>= 1;
#endif
#ifndef FIXED_POINT
if (hDecoder->config.outputFormat >= 5)
Init_Dither(16, hDecoder->config.outputFormat - 5);
#endif
return 0;
}
int8_t FAADAPI faacDecInitDRM(faacDecHandle hDecoder, uint32_t samplerate,
uint8_t channels)
{
/* Special object type defined for DRM */
hDecoder->config.defObjectType = DRM_ER_LC;
hDecoder->config.defSampleRate = samplerate;
hDecoder->aacSectionDataResilienceFlag = 1; /* VCB11 */
hDecoder->aacScalefactorDataResilienceFlag = 0; /* no RVLC */
hDecoder->aacSpectralDataResilienceFlag = 1; /* HCR */
hDecoder->frameLength = 960;
hDecoder->sf_index = get_sr_index(hDecoder->config.defSampleRate);
hDecoder->object_type = hDecoder->config.defObjectType;
hDecoder->channelConfiguration = channels;
/* must be done before frameLength is divided by 2 for LD */
hDecoder->fb = filter_bank_init(hDecoder->frameLength);
#ifndef FIXED_POINT
if (hDecoder->config.outputFormat >= 5)
Init_Dither(16, hDecoder->config.outputFormat - 5);
#endif
return 0;
}
void FAADAPI faacDecClose(faacDecHandle hDecoder)
{
uint8_t i;
for (i = 0; i < MAX_CHANNELS; i++)
{
if (hDecoder->time_out[i]) free(hDecoder->time_out[i]);
#ifdef SSR_DEC
if (hDecoder->ssr_overlap[i]) free(hDecoder->ssr_overlap[i]);
if (hDecoder->prev_fmd[i]) free(hDecoder->prev_fmd[i]);
#endif
#ifdef MAIN_DEC
if (hDecoder->pred_stat[i]) free(hDecoder->pred_stat[i]);
#endif
#ifdef LTP_DEC
if (hDecoder->lt_pred_stat[i]) free(hDecoder->lt_pred_stat[i]);
#endif
}
#ifdef SSR_DEC
if (hDecoder->object_type == SSR)
ssr_filter_bank_end(hDecoder->fb);
else
#endif
filter_bank_end(hDecoder->fb);
drc_end(hDecoder->drc);
#ifndef FIXED_POINT
#if POW_TABLE_SIZE
if (hDecoder->pow2_table) free(hDecoder->pow2_table);
#endif
#endif
if (hDecoder->sample_buffer) free(hDecoder->sample_buffer);
if (hDecoder) free(hDecoder);
}
void* FAADAPI faacDecDecode(faacDecHandle hDecoder,
faacDecFrameInfo *hInfo,
uint8_t *buffer, uint32_t buffer_size)
{
int32_t i;
uint8_t ch;
adts_header adts;
uint8_t channels = 0, ch_ele = 0;
bitfile *ld = (bitfile*)malloc(sizeof(bitfile));
/* local copys of globals */
uint8_t sf_index = hDecoder->sf_index;
uint8_t object_type = hDecoder->object_type;
uint8_t channelConfiguration = hDecoder->channelConfiguration;
#ifdef MAIN_DEC
pred_state **pred_stat = hDecoder->pred_stat;
#endif
#ifdef LTP_DEC
real_t **lt_pred_stat = hDecoder->lt_pred_stat;
#endif
#ifndef FIXED_POINT
#if POW_TABLE_SIZE
real_t *pow2_table = hDecoder->pow2_table;
#else
real_t *pow2_table = NULL;
#endif
#endif
uint8_t *window_shape_prev = hDecoder->window_shape_prev;
real_t **time_out = hDecoder->time_out;
#ifdef SSR_DEC
real_t **ssr_overlap = hDecoder->ssr_overlap;
real_t **prev_fmd = hDecoder->prev_fmd;
#endif
fb_info *fb = hDecoder->fb;
drc_info *drc = hDecoder->drc;
uint8_t outputFormat = hDecoder->config.outputFormat;
#ifdef LTP_DEC
uint16_t *ltp_lag = hDecoder->ltp_lag;
#endif
program_config pce;
element *syntax_elements[MAX_SYNTAX_ELEMENTS];
element **elements;
int16_t *spec_data[MAX_CHANNELS];
real_t *spec_coef[MAX_CHANNELS];
uint16_t frame_len = hDecoder->frameLength;
void *sample_buffer;
memset(hInfo, 0, sizeof(faacDecFrameInfo));
/* initialize the bitstream */
faad_initbits(ld, buffer, buffer_size);
#ifdef DRM
if (object_type == DRM_ER_LC)
{
faad_getbits(ld, 8
DEBUGVAR(1,1,"faacDecDecode(): skip CRC"));
}
#endif
if (hDecoder->adts_header_present)
{
if ((hInfo->error = adts_frame(&adts, ld)) > 0)
goto error;
/* MPEG2 does byte_alignment() here,
* but ADTS header is always multiple of 8 bits in MPEG2
* so not needed to actually do it.
*/
}
#ifdef ANALYSIS
dbg_count = 0;
#endif
elements = syntax_elements;
/* decode the complete bitstream */
elements = raw_data_block(hDecoder, hInfo, ld, syntax_elements,
spec_data, spec_coef, &pce, drc);
if (hInfo->error > 0)
goto error;
ch_ele = hDecoder->fr_ch_ele;
channels = hDecoder->fr_channels;
/* no more bit reading after this */
hInfo->bytesconsumed = bit2byte(faad_get_processed_bits(ld));
faad_endbits(ld);
if (ld) free(ld);
ld = NULL;
/* number of samples in this frame */
hInfo->samples = frame_len*channels;
/* number of samples in this frame */
hInfo->channels = channels;
/* check if frame has channel elements */
if (channels == 0)
{
hDecoder->frame++;
return NULL;
}
if (hDecoder->sample_buffer == NULL)
hDecoder->sample_buffer = malloc(frame_len*channels*sizeof(real_t));
sample_buffer = hDecoder->sample_buffer;
/* noiseless coding is done, the rest of the tools come now */
for (ch = 0; ch < channels; ch++)
{
ic_stream *ics;
/* find the syntax element to which this channel belongs */
for (i = 0; i < ch_ele; i++)
{
if (syntax_elements[i]->channel == ch)
ics = &(syntax_elements[i]->ics1);
else if (syntax_elements[i]->paired_channel == ch)
ics = &(syntax_elements[i]->ics2);
}
/* inverse quantization */
inverse_quantization(spec_coef[ch], spec_data[ch], frame_len);
/* apply scalefactors */
#ifdef FIXED_POINT
apply_scalefactors(ics, spec_coef[ch], frame_len);
#else
apply_scalefactors(ics, spec_coef[ch], pow2_table, frame_len);
#endif
/* deinterleave short block grouping */
if (ics->window_sequence == EIGHT_SHORT_SEQUENCE)
quant_to_spec(ics, spec_coef[ch], frame_len);
}
/* Because for ms, is and pns both channels spectral coefficients are needed
we have to restart running through all channels here.
*/
for (ch = 0; ch < channels; ch++)
{
int16_t pch = -1;
uint8_t right_channel;
ic_stream *ics, *icsr;
ltp_info *ltp;
/* find the syntax element to which this channel belongs */
for (i = 0; i < ch_ele; i++)
{
if (syntax_elements[i]->channel == ch)
{
ics = &(syntax_elements[i]->ics1);
icsr = &(syntax_elements[i]->ics2);
ltp = &(ics->ltp);
pch = syntax_elements[i]->paired_channel;
right_channel = 0;
} else if (syntax_elements[i]->paired_channel == ch) {
ics = &(syntax_elements[i]->ics2);
if (syntax_elements[i]->common_window)
ltp = &(ics->ltp2);
else
ltp = &(ics->ltp);
right_channel = 1;
}
}
/* pns decoding */
if ((!right_channel) && (pch != -1) && (ics->ms_mask_present))
pns_decode(ics, icsr, spec_coef[ch], spec_coef[pch], frame_len, 1);
else if ((pch == -1) || ((pch != -1) && (!ics->ms_mask_present)))
pns_decode(ics, NULL, spec_coef[ch], NULL, frame_len, 0);
if (!right_channel && (pch != -1))
{
/* mid/side decoding */
ms_decode(ics, icsr, spec_coef[ch], spec_coef[pch], frame_len);
/* intensity stereo decoding */
is_decode(ics, icsr, spec_coef[ch], spec_coef[pch], frame_len);
}
#ifdef MAIN_DEC
/* MAIN object type prediction */
if (object_type == MAIN)
{
/* allocate the state only when needed */
if (pred_stat[ch] == NULL)
{
pred_stat[ch] = (pred_state*)malloc(frame_len * sizeof(pred_state));
reset_all_predictors(pred_stat[ch], frame_len);
}
/* intra channel prediction */
ic_prediction(ics, spec_coef[ch], pred_stat[ch], frame_len);
/* In addition, for scalefactor bands coded by perceptual
noise substitution the predictors belonging to the
corresponding spectral coefficients are reset.
*/
pns_reset_pred_state(ics, pred_stat[ch]);
}
#endif
#ifdef LTP_DEC
if ((object_type == LTP)
#ifdef ERROR_RESILIENCE
|| (object_type == ER_LTP)
#endif
#ifdef LD_DEC
|| (object_type == LD)
#endif
)
{
#ifdef LD_DEC
if (object_type == LD)
{
if (ltp->data_present)
{
if (ltp->lag_update)
ltp_lag[ch] = ltp->lag;
}
ltp->lag = ltp_lag[ch];
}
#endif
/* allocate the state only when needed */
if (lt_pred_stat[ch] == NULL)
{
lt_pred_stat[ch] = (real_t*)malloc(frame_len*4 * sizeof(real_t));
memset(lt_pred_stat[ch], 0, frame_len*4 * sizeof(real_t));
}
/* long term prediction */
lt_prediction(ics, ltp, spec_coef[ch], lt_pred_stat[ch], fb,
ics->window_shape, window_shape_prev[ch],
sf_index, object_type, frame_len);
}
#endif
/* tns decoding */
tns_decode_frame(ics, &(ics->tns), sf_index, object_type,
spec_coef[ch], frame_len);
/* drc decoding */
if (drc->present)
{
if (!drc->exclude_mask[ch] || !drc->excluded_chns_present)
drc_decode(drc, spec_coef[ch]);
}
if (time_out[ch] == NULL)
{
time_out[ch] = (real_t*)malloc(frame_len*2*sizeof(real_t));
memset(time_out[ch], 0, frame_len*2*sizeof(real_t));
}
/* filter bank */
#ifdef SSR_DEC
if (object_type != SSR)
{
#endif
ifilter_bank(fb, ics->window_sequence, ics->window_shape,
window_shape_prev[ch], spec_coef[ch],
time_out[ch], object_type, frame_len);
#ifdef SSR_DEC
} else {
if (ssr_overlap[ch] == NULL)
{
ssr_overlap[ch] = (real_t*)malloc(2*frame_len*sizeof(real_t));
memset(ssr_overlap[ch], 0, 2*frame_len*sizeof(real_t));
}
if (prev_fmd[ch] == NULL)
{
uint16_t k;
prev_fmd[ch] = (real_t*)malloc(2*frame_len*sizeof(real_t));
for (k = 0; k < 2*frame_len; k++)
prev_fmd[ch][k] = REAL_CONST(-1);
}
ssr_decode(&(ics->ssr), fb, ics->window_sequence, ics->window_shape,
window_shape_prev[ch], spec_coef[ch], time_out[ch],
ssr_overlap[ch], hDecoder->ipqf_buffer[ch], prev_fmd[ch], frame_len);
}
#endif
/* save window shape for next frame */
window_shape_prev[ch] = ics->window_shape;
#ifdef LTP_DEC
if ((object_type == LTP)
#ifdef ERROR_RESILIENCE
|| (object_type == ER_LTP)
#endif
#ifdef LD_DEC
|| (object_type == LD)
#endif
)
{
lt_update_state(lt_pred_stat[ch], time_out[ch], time_out[ch]+frame_len,
frame_len, object_type);
}
#endif
}
sample_buffer = output_to_PCM(time_out, sample_buffer, channels,
frame_len, outputFormat);
hDecoder->frame++;
#ifdef LD_DEC
if (object_type != LD)
{
#endif
if (hDecoder->frame <= 1)
hInfo->samples = 0;
#ifdef LD_DEC
} else {
/* LD encoders will give lower delay */
if (hDecoder->frame <= 0)
hInfo->samples = 0;
}
#endif
/* cleanup */
for (ch = 0; ch < channels; ch++)
{
if (spec_coef[ch]) free(spec_coef[ch]);
if (spec_data[ch]) free(spec_data[ch]);
}
for (i = 0; i < ch_ele; i++)
{
if (syntax_elements[i]) free(syntax_elements[i]);
}
#ifdef ANALYSIS
fflush(stdout);
#endif
return sample_buffer;
error:
/* free all memory that could have been allocated */
faad_endbits(ld);
if (ld) free(ld);
/* cleanup */
for (ch = 0; ch < channels; ch++)
{
if (spec_coef[ch]) free(spec_coef[ch]);
if (spec_data[ch]) free(spec_data[ch]);
}
for (i = 0; i < ch_ele; i++)
{
if (syntax_elements[i]) free(syntax_elements[i]);
}
#ifdef ANALYSIS
fflush(stdout);
#endif
return NULL;
}