#include "libavutil/internal.h"
#include "libavutil/intreadwrite.h"
#include "libavutil/mathematics.h"
+#include "libavutil/opt.h"
#include "libavutil/samplefmt.h"
#include "avcodec.h"
#include "fft.h"
#define DCA_BUFFER_PADDING_SIZE 1024
+#define DCA_NSYNCAUX 0x9A1105A0
+
/** Bit allocation */
typedef struct {
int offset; ///< code values offset
}
typedef struct {
+ AVClass *class; ///< class for AVOptions
AVCodecContext *avctx;
/* Frame header */
int frame_type; ///< type of the current frame
int scale_factor[DCA_PRIM_CHANNELS_MAX][DCA_SUBBANDS][2]; ///< scale factors (2 if transient)
int joint_huff[DCA_PRIM_CHANNELS_MAX]; ///< joint subband scale factors codebook
int joint_scale_factor[DCA_PRIM_CHANNELS_MAX][DCA_SUBBANDS]; ///< joint subband scale factors
- float downmix_coef[DCA_PRIM_CHANNELS_MAX][2]; ///< stereo downmix coefficients
+ float downmix_coef[DCA_PRIM_CHANNELS_MAX + 1][2]; ///< stereo downmix coefficients
int dynrange_coef; ///< dynamic range coefficient
+ /* Core substream's embedded downmix coefficients (cf. ETSI TS 102 114 V1.4.1)
+ * Input: primary audio channels (incl. LFE if present)
+ * Output: downmix audio channels (up to 4, no LFE) */
+ uint8_t core_downmix; ///< embedded downmix coefficients available
+ uint8_t core_downmix_amode; ///< audio channel arrangement of embedded downmix
+ uint16_t core_downmix_codes[DCA_PRIM_CHANNELS_MAX + 1][4]; ///< embedded downmix coefficients (9-bit codes)
+
int high_freq_vq[DCA_PRIM_CHANNELS_MAX][DCA_SUBBANDS]; ///< VQ encoded high frequency subbands
float lfe_data[2 * DCA_LFE_MAX * (DCA_BLOCKS_MAX + 4)]; ///< Low frequency effect data
/* XCh extension information */
int xch_present; ///< XCh extension present and valid
int xch_base_channel; ///< index of first (only) channel containing XCH data
+ int xch_disable; ///< whether the XCh extension should be decoded or not
/* ExSS header parser */
int static_fields; ///< static fields present
}
}
- /* Stereo downmix coefficients */
- if (!base_channel && s->prim_channels > 2) {
- int am = s->amode & DCA_CHANNEL_MASK;
- if (am >= FF_ARRAY_ELEMS(dca_default_coeffs)) {
- av_log(s->avctx, AV_LOG_ERROR,
- "Invalid channel mode %d\n", am);
- return AVERROR_INVALIDDATA;
- }
- if (s->prim_channels > FF_ARRAY_ELEMS(dca_default_coeffs[0])) {
- avpriv_request_sample(s->avctx, "Downmixing %d channels",
- s->prim_channels);
- return AVERROR_PATCHWELCOME;
- }
- for (j = base_channel; j < s->prim_channels; j++) {
- s->downmix_coef[j][0] = dca_default_coeffs[am][j][0];
- s->downmix_coef[j][1] = dca_default_coeffs[am][j][1];
- }
- }
-
/* Dynamic range coefficient */
if (!base_channel && s->dynrange)
s->dynrange_coef = get_bits(&s->gb, 8);
av_log(s->avctx, AV_LOG_DEBUG, "\n");
}
}
- if (!base_channel && s->prim_channels > 2) {
- av_log(s->avctx, AV_LOG_DEBUG, "Downmix coeffs:\n");
- for (j = 0; j < s->prim_channels; j++) {
- av_log(s->avctx, AV_LOG_DEBUG, "Channel 0, %d = %f\n", j,
- s->downmix_coef[j][0]);
- av_log(s->avctx, AV_LOG_DEBUG, "Channel 1, %d = %f\n", j,
- s->downmix_coef[j][1]);
- }
- av_log(s->avctx, AV_LOG_DEBUG, "\n");
- }
for (j = base_channel; j < s->prim_channels; j++)
for (k = s->vq_start_subband[j]; k < s->subband_activity[j]; k++)
av_log(s->avctx, AV_LOG_DEBUG, "VQ index: %i\n", s->high_freq_vq[j][k]);
op2 \
}
-static void dca_downmix(float **samples, int srcfmt,
- float coef[DCA_PRIM_CHANNELS_MAX][2],
+static void dca_downmix(float **samples, int srcfmt, int lfe_present,
+ float coef[DCA_PRIM_CHANNELS_MAX + 1][2],
const int8_t *channel_mapping)
{
int c, l, r, sl, sr, s;
switch (srcfmt) {
case DCA_MONO:
- case DCA_CHANNEL:
- case DCA_STEREO_TOTAL:
- case DCA_STEREO_SUMDIFF:
case DCA_4F2R:
av_log(NULL, 0, "Not implemented!\n");
break;
+ case DCA_CHANNEL:
case DCA_STEREO:
+ case DCA_STEREO_TOTAL:
+ case DCA_STEREO_SUMDIFF:
break;
case DCA_3F:
c = channel_mapping[0];
MIX_REAR2(samples, sl, sr, 3, coef));
break;
}
+ if (lfe_present) {
+ int lf_buf = dca_lfe_index[srcfmt];
+ int lf_idx = dca_channels [srcfmt];
+ for (i = 0; i < 256; i++) {
+ samples[0][i] += samples[lf_buf][i] * coef[lf_idx][0];
+ samples[1][i] += samples[lf_buf][i] * coef[lf_idx][1];
+ }
+ }
}
M_SQRT1_2 / 32768.0 /* pcm_to_double[s->source_pcm_res] */);
}
- /* Down mixing */
- if (s->prim_channels > 2 &&
- s->avctx->request_channel_layout == AV_CH_LAYOUT_STEREO) {
- dca_downmix(s->samples_chanptr, s->amode, s->downmix_coef, s->channel_order_tab);
- }
-
/* Generate LFE samples for this subsubframe FIXME!!! */
if (s->output & DCA_LFE) {
lfe_interpolation_fir(s, s->lfe, 2 * s->lfe,
/* Outputs 20bits pcm samples */
}
+ /* Downmixing to Stereo */
+ if (s->prim_channels + !!s->lfe > 2 &&
+ s->avctx->request_channel_layout == AV_CH_LAYOUT_STEREO) {
+ dca_downmix(s->samples_chanptr, s->amode, !!s->lfe, s->downmix_coef,
+ s->channel_order_tab);
+ }
+
return 0;
}
static int dca_subframe_footer(DCAContext *s, int base_channel)
{
- int aux_data_count = 0, i;
+ int in, out, aux_data_count, aux_data_end, reserved;
+ uint32_t nsyncaux;
/*
* Unpack optional information
if (s->timestamp)
skip_bits_long(&s->gb, 32);
- if (s->aux_data)
+ if (s->aux_data) {
aux_data_count = get_bits(&s->gb, 6);
- for (i = 0; i < aux_data_count; i++)
- get_bits(&s->gb, 8);
+ // align (32-bit)
+ skip_bits_long(&s->gb, (-get_bits_count(&s->gb)) & 31);
+
+ aux_data_end = 8 * aux_data_count + get_bits_count(&s->gb);
+
+ if ((nsyncaux = get_bits_long(&s->gb, 32)) != DCA_NSYNCAUX) {
+ av_log(s->avctx, AV_LOG_ERROR, "nSYNCAUX mismatch %#"PRIx32"\n",
+ nsyncaux);
+ return AVERROR_INVALIDDATA;
+ }
+
+ if (get_bits1(&s->gb)) { // bAUXTimeStampFlag
+ avpriv_request_sample(s->avctx,
+ "Auxiliary Decode Time Stamp Flag");
+ // align (4-bit)
+ skip_bits(&s->gb, (-get_bits_count(&s->gb)) & 4);
+ // 44 bits: nMSByte (8), nMarker (4), nLSByte (28), nMarker (4)
+ skip_bits_long(&s->gb, 44);
+ }
+
+ if ((s->core_downmix = get_bits1(&s->gb))) {
+ int am = get_bits(&s->gb, 3);
+ switch (am) {
+ case 0:
+ s->core_downmix_amode = DCA_MONO;
+ break;
+ case 1:
+ s->core_downmix_amode = DCA_STEREO;
+ break;
+ case 2:
+ s->core_downmix_amode = DCA_STEREO_TOTAL;
+ break;
+ case 3:
+ s->core_downmix_amode = DCA_3F;
+ break;
+ case 4:
+ s->core_downmix_amode = DCA_2F1R;
+ break;
+ case 5:
+ s->core_downmix_amode = DCA_2F2R;
+ break;
+ case 6:
+ s->core_downmix_amode = DCA_3F1R;
+ break;
+ default:
+ av_log(s->avctx, AV_LOG_ERROR,
+ "Invalid mode %d for embedded downmix coefficients\n",
+ am);
+ return AVERROR_INVALIDDATA;
+ }
+ for (out = 0; out < dca_channels[s->core_downmix_amode]; out++) {
+ for (in = 0; in < s->prim_channels + !!s->lfe; in++) {
+ uint16_t tmp = get_bits(&s->gb, 9);
+ if ((tmp & 0xFF) > 241) {
+ av_log(s->avctx, AV_LOG_ERROR,
+ "Invalid downmix coefficient code %"PRIu16"\n",
+ tmp);
+ return AVERROR_INVALIDDATA;
+ }
+ s->core_downmix_codes[in][out] = tmp;
+ }
+ }
+ }
+
+ align_get_bits(&s->gb); // byte align
+ skip_bits(&s->gb, 16); // nAUXCRC16
+
+ // additional data (reserved, cf. ETSI TS 102 114 V1.4.1)
+ if ((reserved = (aux_data_end - get_bits_count(&s->gb))) < 0) {
+ av_log(s->avctx, AV_LOG_ERROR,
+ "Overread auxiliary data by %d bits\n", -reserved);
+ return AVERROR_INVALIDDATA;
+ } else if (reserved) {
+ avpriv_request_sample(s->avctx,
+ "Core auxiliary data reserved content");
+ skip_bits_long(&s->gb, reserved);
+ }
+ }
if (s->crc_present && s->dynrange)
get_bits(&s->gb, 16);
#if FF_API_REQUEST_CHANNELS
FF_DISABLE_DEPRECATION_WARNINGS
- if (s->xch_present && (!avctx->request_channels ||
- avctx->request_channels > num_core_channels + !!s->lfe)) {
+ if (s->xch_present && !s->xch_disable &&
+ (!avctx->request_channels ||
+ avctx->request_channels > num_core_channels + !!s->lfe)) {
FF_ENABLE_DEPRECATION_WARNINGS
#else
- if (s->xch_present) {
+ if (s->xch_present && !s->xch_disable) {
#endif
avctx->channel_layout |= AV_CH_BACK_CENTER;
if (s->lfe) {
s->channel_order_tab[channels - 1 - !!s->lfe] < 0)
return AVERROR_INVALIDDATA;
- if (s->prim_channels > 2 &&
+ if (s->prim_channels + !!s->lfe > 2 &&
avctx->request_channel_layout == AV_CH_LAYOUT_STEREO) {
channels = 2;
s->output = DCA_STEREO;
avctx->channel_layout = AV_CH_LAYOUT_STEREO;
+
+ /* Stereo downmix coefficients
+ *
+ * The decoder can only downmix to 2-channel, so we need to ensure
+ * embedded downmix coefficients are actually targeting 2-channel.
+ */
+ if (s->core_downmix && (s->core_downmix_amode == DCA_STEREO ||
+ s->core_downmix_amode == DCA_STEREO_TOTAL)) {
+ int sign, code;
+ for (i = 0; i < s->prim_channels + !!s->lfe; i++) {
+ sign = s->core_downmix_codes[i][0] & 0x100 ? 1 : -1;
+ code = s->core_downmix_codes[i][0] & 0x0FF;
+ s->downmix_coef[i][0] = (!code ? 0.0f :
+ sign * dca_dmixtable[code - 1]);
+ sign = s->core_downmix_codes[i][1] & 0x100 ? 1 : -1;
+ code = s->core_downmix_codes[i][1] & 0x0FF;
+ s->downmix_coef[i][1] = (!code ? 0.0f :
+ sign * dca_dmixtable[code - 1]);
+ }
+ } else {
+ int am = s->amode & DCA_CHANNEL_MASK;
+ if (am >= FF_ARRAY_ELEMS(dca_default_coeffs)) {
+ av_log(s->avctx, AV_LOG_ERROR,
+ "Invalid channel mode %d\n", am);
+ return AVERROR_INVALIDDATA;
+ }
+ if (s->prim_channels + !!s->lfe >
+ FF_ARRAY_ELEMS(dca_default_coeffs[0])) {
+ avpriv_request_sample(s->avctx, "Downmixing %d channels",
+ s->prim_channels + !!s->lfe);
+ return AVERROR_PATCHWELCOME;
+ }
+ for (i = 0; i < s->prim_channels + !!s->lfe; i++) {
+ s->downmix_coef[i][0] = dca_default_coeffs[am][i][0];
+ s->downmix_coef[i][1] = dca_default_coeffs[am][i][1];
+ }
+ }
+ av_dlog(s->avctx, "Stereo downmix coeffs:\n");
+ for (i = 0; i < s->prim_channels + !!s->lfe; i++) {
+ av_dlog(s->avctx, "L, input channel %d = %f\n", i,
+ s->downmix_coef[i][0]);
+ av_dlog(s->avctx, "R, input channel %d = %f\n", i,
+ s->downmix_coef[i][1]);
+ }
+ av_dlog(s->avctx, "\n");
}
} else {
av_log(avctx, AV_LOG_ERROR, "Non standard configuration %d !\n", s->amode);
{ FF_PROFILE_UNKNOWN },
};
+static const AVOption options[] = {
+ { "disable_xch", "disable decoding of the XCh extension", offsetof(DCAContext, xch_disable), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, AV_OPT_FLAG_DECODING_PARAM|AV_OPT_FLAG_AUDIO_PARAM },
+ { NULL },
+};
+
+static const AVClass dca_decoder_class = {
+ .class_name = "DCA decoder",
+ .item_name = av_default_item_name,
+ .option = options,
+ .version = LIBAVUTIL_VERSION_INT,
+};
+
AVCodec ff_dca_decoder = {
.name = "dca",
.long_name = NULL_IF_CONFIG_SMALL("DCA (DTS Coherent Acoustics)"),
.sample_fmts = (const enum AVSampleFormat[]) { AV_SAMPLE_FMT_FLTP,
AV_SAMPLE_FMT_NONE },
.profiles = NULL_IF_CONFIG_SMALL(profiles),
+ .priv_class = &dca_decoder_class,
};
projects / ffmpeg / blobdiff