* Copyright (c) 2006-2008 Maxim Poliakovski
* Copyright (c) 2006-2008 Benjamin Larsson
*
- * This file is part of FFmpeg.
+ * This file is part of Libav.
*
- * FFmpeg is free software; you can redistribute it and/or
+ * Libav 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,
+ * Libav 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
+ * License along with Libav; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include "dsputil.h"
#include "bytestream.h"
#include "fft.h"
+#include "fmtconvert.h"
#include "atrac.h"
#include "atrac3data.h"
#define JOINT_STEREO 0x12
#define STEREO 0x2
+#define SAMPLES_PER_FRAME 1024
+#define MDCT_SIZE 512
/* These structures are needed to store the parsed gain control data. */
typedef struct {
int bandsCoded;
int numComponents;
tonal_component components[64];
- float prevFrame[1024];
+ float prevFrame[SAMPLES_PER_FRAME];
int gcBlkSwitch;
gain_block gainBlock[2];
- DECLARE_ALIGNED(16, float, spectrum)[1024];
- DECLARE_ALIGNED(16, float, IMDCT_buf)[1024];
+ DECLARE_ALIGNED(32, float, spectrum)[SAMPLES_PER_FRAME];
+ DECLARE_ALIGNED(32, float, IMDCT_buf)[SAMPLES_PER_FRAME];
float delayBuf1[46]; ///<qmf delay buffers
float delayBuf2[46];
} channel_unit;
typedef struct {
+ AVFrame frame;
GetBitContext gb;
//@{
/** stream data */
//@}
//@{
/** data buffers */
- float outSamples[2048];
+ float *outSamples[2];
uint8_t* decoded_bytes_buffer;
float tempBuf[1070];
//@}
//@}
FFTContext mdct_ctx;
+ FmtConvertContext fmt_conv;
} ATRAC3Context;
-static DECLARE_ALIGNED(16, float,mdct_window)[512];
+static DECLARE_ALIGNED(32, float, mdct_window)[MDCT_SIZE];
static VLC spectral_coeff_tab[7];
static float gain_tab1[16];
static float gain_tab2[31];
/**
* Reverse the odd bands before IMDCT, this is an effect of the QMF transform
* or it gives better compression to do it this way.
- * FIXME: It should be possible to handle this in ff_imdct_calc
+ * FIXME: It should be possible to handle this in imdct_calc
* for that to happen a modification of the prerotation step of
* all SIMD code and C code is needed.
* Or fix the functions before so they generate a pre reversed spectrum.
FFSWAP(float, pInput[i], pInput[255-i]);
}
- ff_imdct_calc(&q->mdct_ctx,pOutput,pInput);
+ q->mdct_ctx.imdct_calc(&q->mdct_ctx,pOutput,pInput);
/* Perform windowing on the output. */
- dsp.vector_fmul(pOutput,mdct_window,512);
+ dsp.vector_fmul(pOutput, pOutput, mdct_window, MDCT_SIZE);
}
obuf[i] = c ^ buf[i];
if (off)
- av_log(NULL,AV_LOG_DEBUG,"Offset of %d not handled, post sample on ffmpeg-dev.\n",off);
+ av_log_ask_for_sample(NULL, "Offset of %d not handled.\n", off);
return off;
}
-static av_cold void init_atrac3_transforms(ATRAC3Context *q) {
+static av_cold int init_atrac3_transforms(ATRAC3Context *q, int is_float) {
float enc_window[256];
int i;
}
/* Initialize the MDCT transform. */
- ff_mdct_init(&q->mdct_ctx, 9, 1, 1.0);
+ return ff_mdct_init(&q->mdct_ctx, 9, 1, is_float ? 1.0 / 32768 : 1.0);
}
/**
av_free(q->pUnits);
av_free(q->decoded_bytes_buffer);
+ av_freep(&q->outSamples[0]);
+
ff_mdct_end(&q->mdct_ctx);
return 0;
readQuantSpectralCoeffs (gb, subband_vlc_index[cnt], codingMode, mantissas, subbWidth);
/* Decode the scale factor for this subband. */
- SF = sf_table[SF_idxs[cnt]] * iMaxQuant[subband_vlc_index[cnt]];
+ SF = ff_atrac_sf_table[SF_idxs[cnt]] * iMaxQuant[subband_vlc_index[cnt]];
/* Inverse quantize the coefficients. */
for (pIn=mantissas ; first<last; first++, pIn++)
/* Clear the subbands that were not coded. */
first = subbandTab[cnt];
- memset(pOut+first, 0, (1024 - first) * sizeof(float));
+ memset(pOut+first, 0, (SAMPLES_PER_FRAME - first) * sizeof(float));
return numSubbands;
}
coding_mode_selector = get_bits(gb,2);
if (coding_mode_selector == 2)
- return -1;
+ return AVERROR_INVALIDDATA;
coding_mode = coding_mode_selector & 1;
quant_step_index = get_bits(gb,3);
if (quant_step_index <= 1)
- return -1;
+ return AVERROR_INVALIDDATA;
if (coding_mode_selector == 3)
coding_mode = get_bits1(gb);
for (k=0; k<coded_components; k++) {
sfIndx = get_bits(gb,6);
+ if (component_count >= 64)
+ return AVERROR_INVALIDDATA;
pComponent[component_count].pos = j * 64 + (get_bits(gb,6));
- max_coded_values = 1024 - pComponent[component_count].pos;
+ max_coded_values = SAMPLES_PER_FRAME - pComponent[component_count].pos;
coded_values = coded_values_per_component + 1;
coded_values = FFMIN(max_coded_values,coded_values);
- scalefactor = sf_table[sfIndx] * iMaxQuant[quant_step_index];
+ scalefactor = ff_atrac_sf_table[sfIndx] * iMaxQuant[quant_step_index];
readQuantSpectralCoeffs(gb, quant_step_index, coding_mode, mantissa, coded_values);
pLevel[cf]= get_bits(gb,4);
pLoc [cf]= get_bits(gb,5);
if(cf && pLoc[cf] <= pLoc[cf-1])
- return -1;
+ return AVERROR_INVALIDDATA;
}
}
if (codingMode == JOINT_STEREO && channelNum == 1) {
if (get_bits(gb,2) != 3) {
av_log(NULL,AV_LOG_ERROR,"JS mono Sound Unit id != 3.\n");
- return -1;
+ return AVERROR_INVALIDDATA;
}
} else {
if (get_bits(gb,6) != 0x28) {
av_log(NULL,AV_LOG_ERROR,"Sound Unit id != 0x28.\n");
- return -1;
+ return AVERROR_INVALIDDATA;
}
}
memset(pSnd->IMDCT_buf, 0, 512 * sizeof(float));
/* gain compensation and overlapping */
- gainCompensateAndOverlap (pSnd->IMDCT_buf, &(pSnd->prevFrame[band*256]), &(pOut[band*256]),
- &((pSnd->gainBlock[1 - (pSnd->gcBlkSwitch)]).gBlock[band]),
- &((pSnd->gainBlock[pSnd->gcBlkSwitch]).gBlock[band]));
+ gainCompensateAndOverlap(pSnd->IMDCT_buf, &pSnd->prevFrame[band * 256],
+ &pOut[band * 256],
+ &pSnd->gainBlock[1 - pSnd->gcBlkSwitch].gBlock[band],
+ &pSnd->gainBlock[ pSnd->gcBlkSwitch].gBlock[band]);
}
/* Swap the gain control buffers for the next frame. */
* @param databuf the input data
*/
-static int decodeFrame(ATRAC3Context *q, const uint8_t* databuf)
+static int decodeFrame(ATRAC3Context *q, const uint8_t* databuf,
+ float **out_samples)
{
int result, i;
float *p1, *p2, *p3, *p4;
/* decode Sound Unit 1 */
init_get_bits(&q->gb,databuf,q->bits_per_frame);
- result = decodeChannelSoundUnit(q,&q->gb, q->pUnits, q->outSamples, 0, JOINT_STEREO);
+ result = decodeChannelSoundUnit(q,&q->gb, q->pUnits, out_samples[0], 0, JOINT_STEREO);
if (result != 0)
- return (result);
+ return result;
/* Framedata of the su2 in the joint-stereo mode is encoded in
* reverse byte order so we need to swap it first. */
ptr1 = q->decoded_bytes_buffer;
for (i = 4; *ptr1 == 0xF8; i++, ptr1++) {
if (i >= q->bytes_per_frame)
- return -1;
+ return AVERROR_INVALIDDATA;
}
}
/* Decode Sound Unit 2. */
- result = decodeChannelSoundUnit(q,&q->gb, &q->pUnits[1], &q->outSamples[1024], 1, JOINT_STEREO);
+ result = decodeChannelSoundUnit(q,&q->gb, &q->pUnits[1], out_samples[1], 1, JOINT_STEREO);
if (result != 0)
- return (result);
+ return result;
/* Reconstruct the channel coefficients. */
- reverseMatrixing(q->outSamples, &q->outSamples[1024], q->matrix_coeff_index_prev, q->matrix_coeff_index_now);
+ reverseMatrixing(out_samples[0], out_samples[1], q->matrix_coeff_index_prev, q->matrix_coeff_index_now);
- channelWeighting(q->outSamples, &q->outSamples[1024], q->weighting_delay);
+ channelWeighting(out_samples[0], out_samples[1], q->weighting_delay);
} else {
/* normal stereo mode or mono */
for (i=0 ; i<q->channels ; i++) {
/* Set the bitstream reader at the start of a channel sound unit. */
- init_get_bits(&q->gb, databuf+((i*q->bytes_per_frame)/q->channels), (q->bits_per_frame)/q->channels);
+ init_get_bits(&q->gb,
+ databuf + i * q->bytes_per_frame / q->channels,
+ q->bits_per_frame / q->channels);
- result = decodeChannelSoundUnit(q,&q->gb, &q->pUnits[i], &q->outSamples[i*1024], i, q->codingMode);
+ result = decodeChannelSoundUnit(q,&q->gb, &q->pUnits[i], out_samples[i], i, q->codingMode);
if (result != 0)
- return (result);
+ return result;
}
}
/* Apply the iQMF synthesis filter. */
- p1= q->outSamples;
for (i=0 ; i<q->channels ; i++) {
+ p1 = out_samples[i];
p2= p1+256;
p3= p2+256;
p4= p3+256;
- atrac_iqmf (p1, p2, 256, p1, q->pUnits[i].delayBuf1, q->tempBuf);
- atrac_iqmf (p4, p3, 256, p3, q->pUnits[i].delayBuf2, q->tempBuf);
- atrac_iqmf (p1, p3, 512, p1, q->pUnits[i].delayBuf3, q->tempBuf);
- p1 +=1024;
+ ff_atrac_iqmf (p1, p2, 256, p1, q->pUnits[i].delayBuf1, q->tempBuf);
+ ff_atrac_iqmf (p4, p3, 256, p3, q->pUnits[i].delayBuf2, q->tempBuf);
+ ff_atrac_iqmf (p1, p3, 512, p1, q->pUnits[i].delayBuf3, q->tempBuf);
}
return 0;
* @param avctx pointer to the AVCodecContext
*/
-static int atrac3_decode_frame(AVCodecContext *avctx,
- void *data, int *data_size,
- AVPacket *avpkt) {
+static int atrac3_decode_frame(AVCodecContext *avctx, void *data,
+ int *got_frame_ptr, AVPacket *avpkt)
+{
const uint8_t *buf = avpkt->data;
int buf_size = avpkt->size;
ATRAC3Context *q = avctx->priv_data;
- int result = 0, i;
+ int result;
const uint8_t* databuf;
- int16_t* samples = data;
+ float *samples_flt;
+ int16_t *samples_s16;
if (buf_size < avctx->block_align) {
av_log(avctx, AV_LOG_ERROR,
"Frame too small (%d bytes). Truncated file?\n", buf_size);
- *data_size = 0;
- return buf_size;
+ return AVERROR_INVALIDDATA;
+ }
+
+ /* get output buffer */
+ q->frame.nb_samples = SAMPLES_PER_FRAME;
+ if ((result = avctx->get_buffer(avctx, &q->frame)) < 0) {
+ av_log(avctx, AV_LOG_ERROR, "get_buffer() failed\n");
+ return result;
}
+ samples_flt = (float *)q->frame.data[0];
+ samples_s16 = (int16_t *)q->frame.data[0];
/* Check if we need to descramble and what buffer to pass on. */
if (q->scrambled_stream) {
databuf = buf;
}
- result = decodeFrame(q, databuf);
+ if (q->channels == 1 && avctx->sample_fmt == AV_SAMPLE_FMT_FLT)
+ result = decodeFrame(q, databuf, &samples_flt);
+ else
+ result = decodeFrame(q, databuf, q->outSamples);
if (result != 0) {
av_log(NULL,AV_LOG_ERROR,"Frame decoding error!\n");
- return -1;
+ return result;
}
- if (q->channels == 1) {
- /* mono */
- for (i = 0; i<1024; i++)
- samples[i] = av_clip_int16(round(q->outSamples[i]));
- *data_size = 1024 * sizeof(int16_t);
- } else {
- /* stereo */
- for (i = 0; i < 1024; i++) {
- samples[i*2] = av_clip_int16(round(q->outSamples[i]));
- samples[i*2+1] = av_clip_int16(round(q->outSamples[1024+i]));
- }
- *data_size = 2048 * sizeof(int16_t);
+ /* interleave */
+ if (q->channels == 2 && avctx->sample_fmt == AV_SAMPLE_FMT_FLT) {
+ q->fmt_conv.float_interleave(samples_flt,
+ (const float **)q->outSamples,
+ SAMPLES_PER_FRAME, 2);
+ } else if (avctx->sample_fmt == AV_SAMPLE_FMT_S16) {
+ q->fmt_conv.float_to_int16_interleave(samples_s16,
+ (const float **)q->outSamples,
+ SAMPLES_PER_FRAME, q->channels);
}
+ *got_frame_ptr = 1;
+ *(AVFrame *)data = q->frame;
+
return avctx->block_align;
}
static av_cold int atrac3_decode_init(AVCodecContext *avctx)
{
- int i;
+ int i, ret;
const uint8_t *edata_ptr = avctx->extradata;
ATRAC3Context *q = avctx->priv_data;
static VLC_TYPE atrac3_vlc_table[4096][2];
av_log(avctx,AV_LOG_DEBUG,"[12-13] %d\n",bytestream_get_le16(&edata_ptr)); //Unknown always 0
/* setup */
- q->samples_per_frame = 1024 * q->channels;
+ q->samples_per_frame = SAMPLES_PER_FRAME * q->channels;
q->atrac3version = 4;
q->delay = 0x88E;
if (q->codingMode)
if ((q->bytes_per_frame == 96*q->channels*q->frame_factor) || (q->bytes_per_frame == 152*q->channels*q->frame_factor) || (q->bytes_per_frame == 192*q->channels*q->frame_factor)) {
} else {
av_log(avctx,AV_LOG_ERROR,"Unknown frame/channel/frame_factor configuration %d/%d/%d\n", q->bytes_per_frame, q->channels, q->frame_factor);
- return -1;
+ return AVERROR_INVALIDDATA;
}
} else if (avctx->extradata_size == 10) {
if (q->atrac3version != 4) {
av_log(avctx,AV_LOG_ERROR,"Version %d != 4.\n",q->atrac3version);
- return -1;
+ return AVERROR_INVALIDDATA;
}
- if (q->samples_per_frame != 1024 && q->samples_per_frame != 2048) {
+ if (q->samples_per_frame != SAMPLES_PER_FRAME && q->samples_per_frame != SAMPLES_PER_FRAME*2) {
av_log(avctx,AV_LOG_ERROR,"Unknown amount of samples per frame %d.\n",q->samples_per_frame);
- return -1;
+ return AVERROR_INVALIDDATA;
}
if (q->delay != 0x88E) {
av_log(avctx,AV_LOG_ERROR,"Unknown amount of delay %x != 0x88E.\n",q->delay);
- return -1;
+ return AVERROR_INVALIDDATA;
}
if (q->codingMode == STEREO) {
av_log(avctx,AV_LOG_DEBUG,"Joint stereo detected.\n");
} else {
av_log(avctx,AV_LOG_ERROR,"Unknown channel coding mode %x!\n",q->codingMode);
- return -1;
+ return AVERROR_INVALIDDATA;
}
if (avctx->channels <= 0 || avctx->channels > 2 /*|| ((avctx->channels * 1024) != q->samples_per_frame)*/) {
av_log(avctx,AV_LOG_ERROR,"Channel configuration error!\n");
- return -1;
+ return AVERROR(EINVAL);
}
if(avctx->block_align >= UINT_MAX/2)
- return -1;
+ return AVERROR(EINVAL);
/* Pad the data buffer with FF_INPUT_BUFFER_PADDING_SIZE,
* this is for the bitstream reader. */
vlcs_initialized = 1;
}
- init_atrac3_transforms(q);
+ if (avctx->request_sample_fmt == AV_SAMPLE_FMT_FLT)
+ avctx->sample_fmt = AV_SAMPLE_FMT_FLT;
+ else
+ avctx->sample_fmt = AV_SAMPLE_FMT_S16;
+
+ if ((ret = init_atrac3_transforms(q, avctx->sample_fmt == AV_SAMPLE_FMT_FLT))) {
+ av_log(avctx, AV_LOG_ERROR, "Error initializing MDCT\n");
+ av_freep(&q->decoded_bytes_buffer);
+ return ret;
+ }
- atrac_generate_tables();
+ ff_atrac_generate_tables();
/* Generate gain tables. */
for (i=0 ; i<16 ; i++)
q->matrix_coeff_index_next[i] = 3;
}
- dsputil_init(&dsp, avctx);
+ ff_dsputil_init(&dsp, avctx);
+ ff_fmt_convert_init(&q->fmt_conv, avctx);
q->pUnits = av_mallocz(sizeof(channel_unit)*q->channels);
if (!q->pUnits) {
- av_free(q->decoded_bytes_buffer);
+ atrac3_decode_close(avctx);
return AVERROR(ENOMEM);
}
- avctx->sample_fmt = SAMPLE_FMT_S16;
+ if (avctx->channels > 1 || avctx->sample_fmt == AV_SAMPLE_FMT_S16) {
+ q->outSamples[0] = av_mallocz(SAMPLES_PER_FRAME * avctx->channels * sizeof(*q->outSamples[0]));
+ q->outSamples[1] = q->outSamples[0] + SAMPLES_PER_FRAME;
+ if (!q->outSamples[0]) {
+ atrac3_decode_close(avctx);
+ return AVERROR(ENOMEM);
+ }
+ }
+
+ avcodec_get_frame_defaults(&q->frame);
+ avctx->coded_frame = &q->frame;
+
return 0;
}
-AVCodec atrac3_decoder =
+AVCodec ff_atrac3_decoder =
{
- .name = "atrac3",
- .type = AVMEDIA_TYPE_AUDIO,
- .id = CODEC_ID_ATRAC3,
+ .name = "atrac3",
+ .type = AVMEDIA_TYPE_AUDIO,
+ .id = CODEC_ID_ATRAC3,
.priv_data_size = sizeof(ATRAC3Context),
- .init = atrac3_decode_init,
- .close = atrac3_decode_close,
- .decode = atrac3_decode_frame,
- .long_name = NULL_IF_CONFIG_SMALL("Atrac 3 (Adaptive TRansform Acoustic Coding 3)"),
+ .init = atrac3_decode_init,
+ .close = atrac3_decode_close,
+ .decode = atrac3_decode_frame,
+ .capabilities = CODEC_CAP_SUBFRAMES | CODEC_CAP_DR1,
+ .long_name = NULL_IF_CONFIG_SMALL("Atrac 3 (Adaptive TRansform Acoustic Coding 3)"),
};
projects / ffmpeg / blobdiff