static int xwd_probe(AVProbeData *p)
{
const uint8_t *b = p->buf;
unsigned width, bpp, bpad, lsize;
if ( p->buf_size < XWD_HEADER_SIZE
|| AV_RB32(b ) < XWD_HEADER_SIZE // header size
|| AV_RB32(b + 4) != XWD_VERSION // version
|| AV_RB32(b + 8) != XWD_Z_PIXMAP // format
|| AV_RB32(b + 12) > 32 || !AV_RB32(b + 12) // depth
|| AV_RB32(b + 16) == 0 // width
|| AV_RB32(b + 20) == 0 // height
|| AV_RB32(b + 28) > 1 // byteorder
|| AV_RB32(b + 32) & ~56 || av_popcount(AV_RB32(b + 32)) != 1 // bitmap unit
|| AV_RB32(b + 36) > 1 // bitorder
|| AV_RB32(b + 40) & ~56 || av_popcount(AV_RB32(b + 40)) != 1 // padding
|| AV_RB32(b + 44) > 32 || !AV_RB32(b + 44) // bpp
|| AV_RB32(b + 68) > 256) // colours
return 0;
width = AV_RB32(b + 16);
bpad = AV_RB32(b + 40);
bpp = AV_RB32(b + 44);
lsize = AV_RB32(b + 48);
if (lsize < FFALIGN(width * bpp, bpad) >> 3)
return 0;
return AVPROBE_SCORE_MAX / 2 + 1;
}
int CMixer::Mixing(float* pOutput, int out_samples, BYTE* pInput, int in_samples)
{
int in_ch = av_popcount(m_in_layout);
int out_ch = av_popcount(m_out_layout);
float* buf = NULL;
if (m_in_avsf != m_in_avsf_used) { // need convert
buf = new float[in_samples * in_ch];
convert_to_float(m_in_avsf, (WORD)in_ch, in_samples, pInput, buf); // convert to float
pInput = (BYTE*)buf;
}
// int in_plane_size = in_samples * (av_sample_fmt_is_planar(in_avsf) ? 1 : in_ch) * av_get_bytes_per_sample(in_avsf);
int in_plane_size = in_samples * in_ch * av_get_bytes_per_sample(m_in_avsf_used);
int out_plane_size = out_samples * out_ch * sizeof(float);
out_samples = avresample_convert(m_pAVRCxt, (uint8_t**)&pOutput, in_plane_size, out_samples, (uint8_t**)&pInput, out_plane_size, in_samples);
if (buf) {
delete [] buf;
}
if (out_samples < 0) {
TRACE(_T("Mixer: avresample_convert failed\n"));
return 0;
}
return out_samples;
}
/**
* Return the number of channels in an ExSS speaker mask (HD)
*/
static int dca_exss_mask2count(int mask)
{
/* count bits that mean speaker pairs twice */
return av_popcount(mask) +
av_popcount(mask & (DCA_EXSS_CENTER_LEFT_RIGHT |
DCA_EXSS_FRONT_LEFT_RIGHT |
DCA_EXSS_FRONT_HIGH_LEFT_RIGHT |
DCA_EXSS_WIDE_LEFT_RIGHT |
DCA_EXSS_SIDE_LEFT_RIGHT |
DCA_EXSS_SIDE_HIGH_LEFT_RIGHT |
DCA_EXSS_SIDE_REAR_LEFT_RIGHT |
DCA_EXSS_REAR_LEFT_RIGHT |
DCA_EXSS_REAR_HIGH_LEFT_RIGHT));
}
void ff_dca_downmix_to_stereo_fixed(DCADSPContext *dcadsp, int32_t **samples,
int *coeff_l, int nsamples, int ch_mask)
{
int pos, spkr, max_spkr = av_log2(ch_mask);
int *coeff_r = coeff_l + av_popcount(ch_mask);
av_assert0(DCA_HAS_STEREO(ch_mask));
// Scale left and right channels
pos = (ch_mask & DCA_SPEAKER_MASK_C);
dcadsp->dmix_scale(samples[DCA_SPEAKER_L], coeff_l[pos ], nsamples);
dcadsp->dmix_scale(samples[DCA_SPEAKER_R], coeff_r[pos + 1], nsamples);
// Downmix remaining channels
for (spkr = 0; spkr <= max_spkr; spkr++) {
if (!(ch_mask & (1U << spkr)))
continue;
if (*coeff_l && spkr != DCA_SPEAKER_L)
dcadsp->dmix_add(samples[DCA_SPEAKER_L], samples[spkr],
*coeff_l, nsamples);
if (*coeff_r && spkr != DCA_SPEAKER_R)
dcadsp->dmix_add(samples[DCA_SPEAKER_R], samples[spkr],
*coeff_r, nsamples);
coeff_l++;
coeff_r++;
}
}
void ff_dca_downmix_to_stereo_float(AVFloatDSPContext *fdsp, float **samples,
int *coeff_l, int nsamples, int ch_mask)
{
int pos, spkr, max_spkr = av_log2(ch_mask);
int *coeff_r = coeff_l + av_popcount(ch_mask);
const float scale = 1.0f / (1 << 15);
av_assert0(DCA_HAS_STEREO(ch_mask));
// Scale left and right channels
pos = (ch_mask & DCA_SPEAKER_MASK_C);
fdsp->vector_fmul_scalar(samples[DCA_SPEAKER_L], samples[DCA_SPEAKER_L],
coeff_l[pos ] * scale, nsamples);
fdsp->vector_fmul_scalar(samples[DCA_SPEAKER_R], samples[DCA_SPEAKER_R],
coeff_r[pos + 1] * scale, nsamples);
// Downmix remaining channels
for (spkr = 0; spkr <= max_spkr; spkr++) {
if (!(ch_mask & (1U << spkr)))
continue;
if (*coeff_l && spkr != DCA_SPEAKER_L)
fdsp->vector_fmac_scalar(samples[DCA_SPEAKER_L], samples[spkr],
*coeff_l * scale, nsamples);
if (*coeff_r && spkr != DCA_SPEAKER_R)
fdsp->vector_fmac_scalar(samples[DCA_SPEAKER_R], samples[spkr],
*coeff_r * scale, nsamples);
coeff_l++;
coeff_r++;
}
}
DWORD CAC3Encoder::SelectLayout(DWORD layout)
{
// check supported layouts
if (m_pAVCodec && m_pAVCodec->channel_layouts) {
for (size_t i = 0; m_pAVCodec->channel_layouts[i] != 0; i++) {
if (layout == (DWORD)m_pAVCodec->channel_layouts[i]) {
return layout;
}
}
}
// select the suitable format
int channels = av_popcount(layout & ~AV_CH_LOW_FREQUENCY); // number of channels without lfe
DWORD new_layout = layout & AV_CH_LOW_FREQUENCY;
if (channels >= 5) {
new_layout |= AV_CH_LAYOUT_5POINT0;
} else if (channels == 4) {
new_layout |= AV_CH_LAYOUT_QUAD;
} else if (channels == 3) {
new_layout |= AV_CH_LAYOUT_2_1;
} else if (channels == 2) {
new_layout |= AV_CH_LAYOUT_STEREO;
} else if (channels == 1) {
new_layout |= AV_CH_LAYOUT_MONO;
}
return new_layout;
}
/**
* Skip mixing coefficients of a single mix out configuration (HD)
*/
static void dca_exss_skip_mix_coeffs(GetBitContext *gb, int channels, int out_ch)
{
int i;
for (i = 0; i < channels; i++) {
int mix_map_mask = get_bits(gb, out_ch);
int num_coeffs = av_popcount(mix_map_mask);
skip_bits_long(gb, num_coeffs * 6);
}
}
int CMixer::Mixing(float* pOutput, int out_samples, BYTE* pInput, int in_samples)
{
int in_ch = av_popcount(m_in_layout);
int out_ch = av_popcount(m_out_layout);
int in_plane_nb = av_sample_fmt_is_planar(m_in_avsf) ? in_ch : 1;
int in_plane_size = in_samples * (av_sample_fmt_is_planar(m_in_avsf) ? 1 : in_ch) * av_get_bytes_per_sample(m_in_avsf);
static BYTE* ppInput[AVRESAMPLE_MAX_CHANNELS];
for (int i = 0; i < in_plane_nb; i++) {
ppInput[i] = pInput + i * in_plane_size;
}
int out_plane_size = out_samples * out_ch * sizeof(float);
out_samples = avresample_convert(m_pAVRCxt, (uint8_t**)&pOutput, out_plane_size, out_samples, ppInput, in_plane_size, in_samples);
if (out_samples < 0) {
TRACE(_T("Mixer: avresample_convert failed\n"));
return 0;
}
return out_samples;
}
bool CAC3Encoder::Init(int sample_rate, DWORD channel_layout)
{
StreamFinish();
int ret;
m_pAVCtx = avcodec_alloc_context3(m_pAVCodec);
m_pAVCtx->bit_rate = 640000;
m_pAVCtx->sample_fmt = AV_SAMPLE_FMT_FLTP;
m_pAVCtx->sample_rate = SelectSamplerate(sample_rate);
m_pAVCtx->channel_layout = channel_layout;
m_pAVCtx->channels = av_popcount(channel_layout);
ret = avcodec_open2(m_pAVCtx, m_pAVCodec, NULL);
if (ret < 0) {
DbgLog((LOG_TRACE, 3, L"CAC3Encoder::Init() : avcodec_open2() failed"));
return false;
}
m_pFrame = av_frame_alloc();
if (!m_pFrame) {
DbgLog((LOG_TRACE, 3, L"CAC3Encoder::Init() : avcodec_alloc_frame() failed"));
return false;
}
m_pFrame->nb_samples = m_pAVCtx->frame_size;
m_pFrame->format = m_pAVCtx->sample_fmt;
m_pFrame->channel_layout = m_pAVCtx->channel_layout;
// the codec gives us the frame size, in samples,
// we calculate the size of the samples buffer in bytes
m_framesize = m_pAVCtx->frame_size * m_pAVCtx->channels * sizeof(float);
m_buffersize = av_samples_get_buffer_size(NULL, m_pAVCtx->channels, m_pAVCtx->frame_size, m_pAVCtx->sample_fmt, 0);
m_pSamples = (float*)av_malloc(m_buffersize);
if (!m_pSamples) {
DbgLog((LOG_TRACE, 3, L"CAC3Encoder::Init() : av_malloc(%d) failed", m_buffersize));
return false;
}
/* setup the data pointers in the AVFrame */
ret = avcodec_fill_audio_frame(m_pFrame, m_pAVCtx->channels, m_pAVCtx->sample_fmt, (const uint8_t*)m_pSamples, m_buffersize, 0);
if (ret < 0) {
DbgLog((LOG_TRACE, 3, L"CAC3Encoder::Init() : avcodec_fill_audio_frame() failed"));
return false;
}
return true;
}
static int pcx_probe(AVProbeData *p)
{
const uint8_t *b = p->buf;
if ( p->buf_size < 128
|| b[0] != 10
|| b[1] > 5
|| b[2] > 1
|| av_popcount(b[3]) != 1 || b[3] > 8
|| AV_RL16(&b[4]) > AV_RL16(&b[8])
|| AV_RL16(&b[6]) > AV_RL16(&b[10])
|| b[64])
return 0;
b += 73;
while (++b < p->buf + 128)
if (*b)
return AVPROBE_SCORE_EXTENSION / 4;
return AVPROBE_SCORE_EXTENSION + 1;
}
static int vmdaudio_decode_frame(AVCodecContext *avctx, void *data,
int *got_frame_ptr, AVPacket *avpkt)
{
AVFrame *frame = data;
const uint8_t *buf = avpkt->data;
const uint8_t *buf_end;
int buf_size = avpkt->size;
VmdAudioContext *s = avctx->priv_data;
int block_type, silent_chunks, audio_chunks;
int ret;
uint8_t *output_samples_u8;
int16_t *output_samples_s16;
if (buf_size < 16) {
av_log(avctx, AV_LOG_WARNING, "skipping small junk packet\n");
*got_frame_ptr = 0;
return buf_size;
}
block_type = buf[6];
if (block_type < BLOCK_TYPE_AUDIO || block_type > BLOCK_TYPE_SILENCE) {
av_log(avctx, AV_LOG_ERROR, "unknown block type: %d\n", block_type);
return AVERROR(EINVAL);
}
buf += 16;
buf_size -= 16;
/* get number of silent chunks */
silent_chunks = 0;
if (block_type == BLOCK_TYPE_INITIAL) {
uint32_t flags;
if (buf_size < 4) {
av_log(avctx, AV_LOG_ERROR, "packet is too small\n");
return AVERROR(EINVAL);
}
flags = AV_RB32(buf);
silent_chunks = av_popcount(flags);
buf += 4;
buf_size -= 4;
} else if (block_type == BLOCK_TYPE_SILENCE) {
silent_chunks = 1;
buf_size = 0; // should already be zero but set it just to be sure
}
/* ensure output buffer is large enough */
audio_chunks = buf_size / s->chunk_size;
/* drop incomplete chunks */
buf_size = audio_chunks * s->chunk_size;
/* get output buffer */
frame->nb_samples = ((silent_chunks + audio_chunks) * avctx->block_align) /
avctx->channels;
if ((ret = ff_get_buffer(avctx, frame, 0)) < 0)
return ret;
output_samples_u8 = frame->data[0];
output_samples_s16 = (int16_t *)frame->data[0];
/* decode silent chunks */
if (silent_chunks > 0) {
int silent_size = avctx->block_align * silent_chunks;
av_assert0(avctx->block_align * silent_chunks <= frame->nb_samples * avctx->channels);
if (s->out_bps == 2) {
memset(output_samples_s16, 0x00, silent_size * 2);
output_samples_s16 += silent_size;
} else {
memset(output_samples_u8, 0x80, silent_size);
output_samples_u8 += silent_size;
}
}
/* decode audio chunks */
if (audio_chunks > 0) {
buf_end = buf + buf_size;
av_assert0((buf_size & (avctx->channels > 1)) == 0);
while (buf_end - buf >= s->chunk_size) {
if (s->out_bps == 2) {
decode_audio_s16(output_samples_s16, buf, s->chunk_size,
avctx->channels);
output_samples_s16 += avctx->block_align;
} else {
memcpy(output_samples_u8, buf, s->chunk_size);
output_samples_u8 += avctx->block_align;
}
buf += s->chunk_size;
}
}
*got_frame_ptr = 1;
return avpkt->size;
}
static int parse_descriptor(DCAExssParser *s, DCAExssAsset *asset)
{
int i, j, drc_present, descr_size, descr_pos = get_bits_count(&s->gb);
// Size of audio asset descriptor in bytes
descr_size = get_bits(&s->gb, 9) + 1;
// Audio asset identifier
asset->asset_index = get_bits(&s->gb, 3);
//
// Per stream static metadata
//
if (s->static_fields_present) {
// Asset type descriptor presence
if (get_bits1(&s->gb))
// Asset type descriptor
skip_bits(&s->gb, 4);
// Language descriptor presence
if (get_bits1(&s->gb))
// Language descriptor
skip_bits(&s->gb, 24);
// Additional textual information presence
if (get_bits1(&s->gb)) {
// Byte size of additional text info
int text_size = get_bits(&s->gb, 10) + 1;
// Sanity check available size
if (get_bits_left(&s->gb) < text_size * 8)
return AVERROR_INVALIDDATA;
// Additional textual information string
skip_bits_long(&s->gb, text_size * 8);
}
// PCM bit resolution
asset->pcm_bit_res = get_bits(&s->gb, 5) + 1;
// Maximum sample rate
asset->max_sample_rate = ff_dca_sampling_freqs[get_bits(&s->gb, 4)];
// Total number of channels
asset->nchannels_total = get_bits(&s->gb, 8) + 1;
// One to one map channel to speakers
if (asset->one_to_one_map_ch_to_spkr = get_bits1(&s->gb)) {
int spkr_mask_nbits = 0;
int spkr_remap_nsets;
int nspeakers[8];
// Embedded stereo flag
asset->embedded_stereo = asset->nchannels_total > 2 && get_bits1(&s->gb);
// Embedded 6 channels flag
asset->embedded_6ch = asset->nchannels_total > 6 && get_bits1(&s->gb);
// Speaker mask enabled flag
if (asset->spkr_mask_enabled = get_bits1(&s->gb)) {
// Number of bits for speaker activity mask
spkr_mask_nbits = (get_bits(&s->gb, 2) + 1) << 2;
// Loudspeaker activity mask
asset->spkr_mask = get_bits(&s->gb, spkr_mask_nbits);
}
// Number of speaker remapping sets
if ((spkr_remap_nsets = get_bits(&s->gb, 3)) && !spkr_mask_nbits) {
av_log(s->avctx, AV_LOG_ERROR, "Speaker mask disabled yet there are remapping sets\n");
return AVERROR_INVALIDDATA;
}
// Standard loudspeaker layout mask
for (i = 0; i < spkr_remap_nsets; i++)
nspeakers[i] = ff_dca_count_chs_for_mask(get_bits(&s->gb, spkr_mask_nbits));
for (i = 0; i < spkr_remap_nsets; i++) {
// Number of channels to be decoded for speaker remapping
int nch_for_remaps = get_bits(&s->gb, 5) + 1;
for (j = 0; j < nspeakers[i]; j++) {
// Decoded channels to output speaker mapping mask
int remap_ch_mask = get_bits_long(&s->gb, nch_for_remaps);
// Loudspeaker remapping codes
skip_bits_long(&s->gb, av_popcount(remap_ch_mask) * 5);
}
}
} else {
asset->embedded_stereo = 0;
asset->embedded_6ch = 0;
asset->spkr_mask_enabled = 0;
asset->spkr_mask = 0;
// Representation type
asset->representation_type = get_bits(&s->gb, 3);
}
}
//
// DRC, DNC and mixing metadata
//
// Dynamic range coefficient presence flag
drc_present = get_bits1(&s->gb);
// Code for dynamic range coefficient
if (drc_present)
skip_bits(&s->gb, 8);
// Dialog normalization presence flag
if (get_bits1(&s->gb))
// Dialog normalization code
skip_bits(&s->gb, 5);
// DRC for stereo downmix
if (drc_present && asset->embedded_stereo)
skip_bits(&s->gb, 8);
// Mixing metadata presence flag
if (s->mix_metadata_enabled && get_bits1(&s->gb)) {
int nchannels_dmix;
// External mixing flag
skip_bits1(&s->gb);
// Post mixing / replacement gain adjustment
skip_bits(&s->gb, 6);
// DRC prior to mixing
if (get_bits(&s->gb, 2) == 3)
// Custom code for mixing DRC
skip_bits(&s->gb, 8);
else
// Limit for mixing DRC
skip_bits(&s->gb, 3);
// Scaling type for channels of main audio
// Scaling parameters of main audio
if (get_bits1(&s->gb))
for (i = 0; i < s->nmixoutconfigs; i++)
skip_bits_long(&s->gb, 6 * s->nmixoutchs[i]);
else
skip_bits_long(&s->gb, 6 * s->nmixoutconfigs);
nchannels_dmix = asset->nchannels_total;
if (asset->embedded_6ch)
nchannels_dmix += 6;
if (asset->embedded_stereo)
nchannels_dmix += 2;
for (i = 0; i < s->nmixoutconfigs; i++) {
if (!s->nmixoutchs[i]) {
av_log(s->avctx, AV_LOG_ERROR, "Invalid speaker layout mask for mixing configuration\n");
return AVERROR_INVALIDDATA;
}
for (j = 0; j < nchannels_dmix; j++) {
// Mix output mask
int mix_map_mask = get_bits(&s->gb, s->nmixoutchs[i]);
// Mixing coefficients
skip_bits_long(&s->gb, av_popcount(mix_map_mask) * 6);
}
}
}
//
// Decoder navigation data
//
// Coding mode for the asset
asset->coding_mode = get_bits(&s->gb, 2);
// Coding components used in asset
switch (asset->coding_mode) {
case 0: // Coding mode that may contain multiple coding components
asset->extension_mask = get_bits(&s->gb, 12);
if (asset->extension_mask & DCA_EXSS_CORE) {
// Size of core component in extension substream
asset->core_size = get_bits(&s->gb, 14) + 1;
// Core sync word present flag
if (get_bits1(&s->gb))
// Core sync distance
skip_bits(&s->gb, 2);
}
if (asset->extension_mask & DCA_EXSS_XBR)
// Size of XBR extension in extension substream
asset->xbr_size = get_bits(&s->gb, 14) + 1;
if (asset->extension_mask & DCA_EXSS_XXCH)
// Size of XXCH extension in extension substream
asset->xxch_size = get_bits(&s->gb, 14) + 1;
if (asset->extension_mask & DCA_EXSS_X96)
// Size of X96 extension in extension substream
asset->x96_size = get_bits(&s->gb, 12) + 1;
if (asset->extension_mask & DCA_EXSS_LBR)
parse_lbr_parameters(s, asset);
if (asset->extension_mask & DCA_EXSS_XLL)
parse_xll_parameters(s, asset);
if (asset->extension_mask & DCA_EXSS_RSV1)
skip_bits(&s->gb, 16);
if (asset->extension_mask & DCA_EXSS_RSV2)
skip_bits(&s->gb, 16);
break;
case 1: // Loss-less coding mode without CBR component
asset->extension_mask = DCA_EXSS_XLL;
parse_xll_parameters(s, asset);
break;
case 2: // Low bit rate mode
asset->extension_mask = DCA_EXSS_LBR;
parse_lbr_parameters(s, asset);
break;
case 3: // Auxiliary coding mode
asset->extension_mask = 0;
// Size of auxiliary coded data
skip_bits(&s->gb, 14);
// Auxiliary codec identification
skip_bits(&s->gb, 8);
// Aux sync word present flag
if (get_bits1(&s->gb))
// Aux sync distance
skip_bits(&s->gb, 3);
break;
}
if (asset->extension_mask & DCA_EXSS_XLL)
// DTS-HD stream ID
asset->hd_stream_id = get_bits(&s->gb, 3);
// One to one mixing flag
// Per channel main audio scaling flag
// Main audio scaling codes
// Decode asset in secondary decoder flag
// Revision 2 DRC metadata
// Reserved
// Zero pad
if (ff_dca_seek_bits(&s->gb, descr_pos + descr_size * 8)) {
av_log(s->avctx, AV_LOG_ERROR, "Read past end of EXSS asset descriptor\n");
return AVERROR_INVALIDDATA;
}
return 0;
}
int ff_dca_exss_parse(DCAExssParser *s, uint8_t *data, int size)
{
int i, ret, offset, wide_hdr, header_size;
if ((ret = init_get_bits8(&s->gb, data, size)) < 0)
return ret;
// Extension substream sync word
skip_bits_long(&s->gb, 32);
// User defined bits
skip_bits(&s->gb, 8);
// Extension substream index
s->exss_index = get_bits(&s->gb, 2);
// Flag indicating short or long header size
wide_hdr = get_bits1(&s->gb);
// Extension substream header length
header_size = get_bits(&s->gb, 8 + 4 * wide_hdr) + 1;
// Check CRC
if (ff_dca_check_crc(s->avctx, &s->gb, 32 + 8, header_size * 8)) {
av_log(s->avctx, AV_LOG_ERROR, "Invalid EXSS header checksum\n");
return AVERROR_INVALIDDATA;
}
s->exss_size_nbits = 16 + 4 * wide_hdr;
// Number of bytes of extension substream
s->exss_size = get_bits(&s->gb, s->exss_size_nbits) + 1;
if (s->exss_size > size) {
av_log(s->avctx, AV_LOG_ERROR, "Packet too short for EXSS frame\n");
return AVERROR_INVALIDDATA;
}
// Per stream static fields presence flag
if (s->static_fields_present = get_bits1(&s->gb)) {
int active_exss_mask[8];
// Reference clock code
skip_bits(&s->gb, 2);
// Extension substream frame duration
skip_bits(&s->gb, 3);
// Timecode presence flag
if (get_bits1(&s->gb))
// Timecode data
skip_bits_long(&s->gb, 36);
// Number of defined audio presentations
s->npresents = get_bits(&s->gb, 3) + 1;
if (s->npresents > 1) {
avpriv_request_sample(s->avctx, "%d audio presentations", s->npresents);
return AVERROR_PATCHWELCOME;
}
// Number of audio assets in extension substream
s->nassets = get_bits(&s->gb, 3) + 1;
if (s->nassets > 1) {
avpriv_request_sample(s->avctx, "%d audio assets", s->nassets);
return AVERROR_PATCHWELCOME;
}
// Active extension substream mask for audio presentation
for (i = 0; i < s->npresents; i++)
active_exss_mask[i] = get_bits(&s->gb, s->exss_index + 1);
// Active audio asset mask
for (i = 0; i < s->npresents; i++)
skip_bits_long(&s->gb, av_popcount(active_exss_mask[i]) * 8);
// Mixing metadata enable flag
if (s->mix_metadata_enabled = get_bits1(&s->gb)) {
int spkr_mask_nbits;
// Mixing metadata adjustment level
skip_bits(&s->gb, 2);
// Number of bits for mixer output speaker activity mask
spkr_mask_nbits = (get_bits(&s->gb, 2) + 1) << 2;
// Number of mixing configurations
s->nmixoutconfigs = get_bits(&s->gb, 2) + 1;
// Speaker layout mask for mixer output channels
for (i = 0; i < s->nmixoutconfigs; i++)
s->nmixoutchs[i] = ff_dca_count_chs_for_mask(get_bits(&s->gb, spkr_mask_nbits));
}
} else {
s->npresents = 1;
s->nassets = 1;
}
// Size of encoded asset data in bytes
offset = header_size;
for (i = 0; i < s->nassets; i++) {
s->assets[i].asset_offset = offset;
s->assets[i].asset_size = get_bits(&s->gb, s->exss_size_nbits) + 1;
offset += s->assets[i].asset_size;
if (offset > s->exss_size) {
av_log(s->avctx, AV_LOG_ERROR, "EXSS asset out of bounds\n");
return AVERROR_INVALIDDATA;
}
}
// Audio asset descriptor
for (i = 0; i < s->nassets; i++) {
if ((ret = parse_descriptor(s, &s->assets[i])) < 0)
return ret;
if ((ret = set_exss_offsets(&s->assets[i])) < 0) {
av_log(s->avctx, AV_LOG_ERROR, "Invalid extension size in EXSS asset descriptor\n");
return ret;
}
}
// Backward compatible core present
// Backward compatible core substream index
// Backward compatible core asset index
// Reserved
// Byte align
// CRC16 of extension substream header
if (ff_dca_seek_bits(&s->gb, header_size * 8)) {
av_log(s->avctx, AV_LOG_ERROR, "Read past end of EXSS header\n");
return AVERROR_INVALIDDATA;
}
return 0;
}
/**
* Parse extension substream asset header (HD)
*/
static int dca_exss_parse_asset_header(DCAContext *s)
{
int header_pos = get_bits_count(&s->gb);
int header_size;
int channels = 0;
int embedded_stereo = 0;
int embedded_6ch = 0;
int drc_code_present;
int extensions_mask = 0;
int i, j;
if (get_bits_left(&s->gb) < 16)
return AVERROR_INVALIDDATA;
/* We will parse just enough to get to the extensions bitmask with which
* we can set the profile value. */
header_size = get_bits(&s->gb, 9) + 1;
skip_bits(&s->gb, 3); // asset index
if (s->static_fields) {
if (get_bits1(&s->gb))
skip_bits(&s->gb, 4); // asset type descriptor
if (get_bits1(&s->gb))
skip_bits_long(&s->gb, 24); // language descriptor
if (get_bits1(&s->gb)) {
/* How can one fit 1024 bytes of text here if the maximum value
* for the asset header size field above was 512 bytes? */
int text_length = get_bits(&s->gb, 10) + 1;
if (get_bits_left(&s->gb) < text_length * 8)
return AVERROR_INVALIDDATA;
skip_bits_long(&s->gb, text_length * 8); // info text
}
skip_bits(&s->gb, 5); // bit resolution - 1
skip_bits(&s->gb, 4); // max sample rate code
channels = get_bits(&s->gb, 8) + 1;
if (get_bits1(&s->gb)) { // 1-to-1 channels to speakers
int spkr_remap_sets;
int spkr_mask_size = 16;
int num_spkrs[7];
if (channels > 2)
embedded_stereo = get_bits1(&s->gb);
if (channels > 6)
embedded_6ch = get_bits1(&s->gb);
if (get_bits1(&s->gb)) {
spkr_mask_size = (get_bits(&s->gb, 2) + 1) << 2;
skip_bits(&s->gb, spkr_mask_size); // spkr activity mask
}
spkr_remap_sets = get_bits(&s->gb, 3);
for (i = 0; i < spkr_remap_sets; i++) {
/* std layout mask for each remap set */
num_spkrs[i] = dca_exss_mask2count(get_bits(&s->gb, spkr_mask_size));
}
for (i = 0; i < spkr_remap_sets; i++) {
int num_dec_ch_remaps = get_bits(&s->gb, 5) + 1;
if (get_bits_left(&s->gb) < 0)
return AVERROR_INVALIDDATA;
for (j = 0; j < num_spkrs[i]; j++) {
int remap_dec_ch_mask = get_bits_long(&s->gb, num_dec_ch_remaps);
int num_dec_ch = av_popcount(remap_dec_ch_mask);
skip_bits_long(&s->gb, num_dec_ch * 5); // remap codes
}
}
} else {
skip_bits(&s->gb, 3); // representation type
}
}
drc_code_present = get_bits1(&s->gb);
if (drc_code_present)
get_bits(&s->gb, 8); // drc code
if (get_bits1(&s->gb))
skip_bits(&s->gb, 5); // dialog normalization code
if (drc_code_present && embedded_stereo)
get_bits(&s->gb, 8); // drc stereo code
if (s->mix_metadata && get_bits1(&s->gb)) {
skip_bits(&s->gb, 1); // external mix
skip_bits(&s->gb, 6); // post mix gain code
if (get_bits(&s->gb, 2) != 3) // mixer drc code
skip_bits(&s->gb, 3); // drc limit
else
skip_bits(&s->gb, 8); // custom drc code
if (get_bits1(&s->gb)) // channel specific scaling
for (i = 0; i < s->num_mix_configs; i++)
skip_bits_long(&s->gb, s->mix_config_num_ch[i] * 6); // scale codes
else
skip_bits_long(&s->gb, s->num_mix_configs * 6); // scale codes
for (i = 0; i < s->num_mix_configs; i++) {
if (get_bits_left(&s->gb) < 0)
return AVERROR_INVALIDDATA;
dca_exss_skip_mix_coeffs(&s->gb, channels, s->mix_config_num_ch[i]);
if (embedded_6ch)
dca_exss_skip_mix_coeffs(&s->gb, 6, s->mix_config_num_ch[i]);
if (embedded_stereo)
dca_exss_skip_mix_coeffs(&s->gb, 2, s->mix_config_num_ch[i]);
}
}
switch (get_bits(&s->gb, 2)) {
case 0:
extensions_mask = get_bits(&s->gb, 12);
break;
case 1:
extensions_mask = DCA_EXT_EXSS_XLL;
break;
case 2:
extensions_mask = DCA_EXT_EXSS_LBR;
break;
case 3:
extensions_mask = 0; /* aux coding */
break;
}
/* not parsed further, we were only interested in the extensions mask */
if (get_bits_left(&s->gb) < 0)
return AVERROR_INVALIDDATA;
if (get_bits_count(&s->gb) - header_pos > header_size * 8) {
av_log(s->avctx, AV_LOG_WARNING, "Asset header size mismatch.\n");
return AVERROR_INVALIDDATA;
}
skip_bits_long(&s->gb, header_pos + header_size * 8 - get_bits_count(&s->gb));
if (extensions_mask & DCA_EXT_EXSS_XLL)
s->profile = FF_PROFILE_DTS_HD_MA;
else if (extensions_mask & (DCA_EXT_EXSS_XBR | DCA_EXT_EXSS_X96 |
DCA_EXT_EXSS_XXCH))
s->profile = FF_PROFILE_DTS_HD_HRA;
if (!(extensions_mask & DCA_EXT_CORE))
av_log(s->avctx, AV_LOG_WARNING, "DTS core detection mismatch.\n");
if ((extensions_mask & DCA_CORE_EXTS) != s->core_ext_mask)
av_log(s->avctx, AV_LOG_WARNING,
"DTS extensions detection mismatch (%d, %d)\n",
extensions_mask & DCA_CORE_EXTS, s->core_ext_mask);
return 0;
}
void CMixer::Init(AVSampleFormat in_avsf, DWORD in_layout, DWORD out_layout, int in_samplerate, int out_samplerate)
{
// reset parameters
m_in_avsf = AV_SAMPLE_FMT_NONE;
m_in_layout = 0;
m_out_layout = 0;
m_in_samplerate = 0;
m_out_samplerate = 0;
av_free(m_matrix_dbl);
// Close Resample Context
avresample_close(m_pAVRCxt);
if (in_avsf >= AV_SAMPLE_FMT_U8P && in_avsf <= AV_SAMPLE_FMT_DBLP) { // planar audio is not supported (ffmpeg crashed)
m_in_avsf_used = AV_SAMPLE_FMT_FLT; // convert to float
} else {
m_in_avsf_used = in_avsf;
}
int ret = 0;
// Set options
av_opt_set_int(m_pAVRCxt, "in_sample_fmt", m_in_avsf_used, 0);
av_opt_set_int(m_pAVRCxt, "out_sample_fmt", AV_SAMPLE_FMT_FLT, 0); // forced float output
av_opt_set_int(m_pAVRCxt, "in_channel_layout", in_layout, 0);
av_opt_set_int(m_pAVRCxt, "out_channel_layout", out_layout, 0);
av_opt_set_int(m_pAVRCxt, "in_sample_rate", in_samplerate, 0);
av_opt_set_int(m_pAVRCxt, "out_sample_rate", out_samplerate, 0);
// Open Resample Context
ret = avresample_open(m_pAVRCxt);
if (ret < 0) {
TRACE(_T("Mixer: avresample_open failed\n"));
return;
}
// Create Matrix
int in_ch = av_popcount(in_layout);
int out_ch = av_popcount(out_layout);
m_matrix_dbl = (double*)av_mallocz(in_ch * out_ch * sizeof(*m_matrix_dbl));
// expand stereo
if (in_layout == AV_CH_LAYOUT_STEREO && (out_layout == AV_CH_LAYOUT_QUAD || out_layout == AV_CH_LAYOUT_5POINT1 || out_layout == AV_CH_LAYOUT_7POINT1)) {
m_matrix_dbl[0] = 1.0;
m_matrix_dbl[1] = 0.0;
m_matrix_dbl[2] = 0.0;
m_matrix_dbl[3] = 1.0;
if (out_layout == AV_CH_LAYOUT_QUAD) {
m_matrix_dbl[4] = 0.6666;
m_matrix_dbl[5] = (-0.2222);
m_matrix_dbl[6] = (-0.2222);
m_matrix_dbl[7] = 0.6666;
} else if (out_layout == AV_CH_LAYOUT_5POINT1 || out_layout == AV_CH_LAYOUT_7POINT1) {
m_matrix_dbl[4] = 0.5;
m_matrix_dbl[5] = 0.5;
m_matrix_dbl[6] = 0.0;
m_matrix_dbl[7] = 0.0;
m_matrix_dbl[8] = 0.6666;
m_matrix_dbl[9] = (-0.2222);
m_matrix_dbl[10] = (-0.2222);
m_matrix_dbl[11] = 0.6666;
if (out_layout == AV_CH_LAYOUT_7POINT1) {
m_matrix_dbl[12] = 0.6666;
m_matrix_dbl[13] = (-0.2222);
m_matrix_dbl[14] = (-0.2222);
m_matrix_dbl[15] = 0.6666;
}
}
} else {
const double center_mix_level = M_SQRT1_2;
const double surround_mix_level = M_SQRT1_2;
const double lfe_mix_level = M_SQRT1_2;
const int normalize = 0;
ret = avresample_build_matrix(in_layout, out_layout, center_mix_level, surround_mix_level, lfe_mix_level, normalize, m_matrix_dbl, in_ch, AV_MATRIX_ENCODING_NONE);
if (ret < 0) {
TRACE(_T("Mixer: avresample_build_matrix failed\n"));
av_free(m_matrix_dbl);
return;
}
}
#ifdef _DEBUG
CString matrix_str;
for (int j = 0; j < out_ch; j++) {
matrix_str.AppendFormat(_T("%d:"), j + 1);
for (int i = 0; i < in_ch; i++) {
double k = m_matrix_dbl[j * in_ch + i];
matrix_str.AppendFormat(_T(" %.4f"), k);
}
matrix_str += _T("\n");
}
TRACE(matrix_str);
#endif
// Set Matrix on the context
ret = avresample_set_matrix(m_pAVRCxt, m_matrix_dbl, in_ch);
if (ret < 0) {
TRACE(_T("Mixer: avresample_set_matrix failed\n"));
av_free(m_matrix_dbl);
return;
}
m_in_avsf = in_avsf;
m_in_layout = in_layout;
m_out_layout = out_layout;
m_in_samplerate = in_samplerate;
m_out_samplerate = out_samplerate;
}
void CMixer::Init(DWORD out_layout, DWORD in_layout, enum AVSampleFormat in_sf)
{
avresample_free(&m_pAVRCxt);
int ret = 0;
// Allocate Resample Context and set options.
m_pAVRCxt = avresample_alloc_context();
av_opt_set_int(m_pAVRCxt, "in_channel_layout", in_layout, 0);
av_opt_set_int(m_pAVRCxt, "in_sample_fmt", in_sf, 0);
av_opt_set_int(m_pAVRCxt, "out_channel_layout", out_layout, 0);
av_opt_set_int(m_pAVRCxt, "out_sample_fmt", AV_SAMPLE_FMT_FLT, 0); // forced float output
// Open Resample Context
ret = avresample_open(m_pAVRCxt);
if (ret < 0) {
TRACE(_T("Mixer: avresample_open failed\n"));
avresample_free(&m_pAVRCxt);
return;
}
// Create Matrix
int in_ch = av_popcount(in_layout);
int out_ch = av_popcount(out_layout);
double* matrix_dbl = (double*)av_mallocz(in_ch * out_ch * sizeof(*matrix_dbl));
// expand stereo
if (in_layout == AV_CH_LAYOUT_STEREO && (out_layout == AV_CH_LAYOUT_QUAD || out_layout == AV_CH_LAYOUT_5POINT1 || out_layout == AV_CH_LAYOUT_7POINT1)) {
matrix_dbl[0] = 1.0;
matrix_dbl[1] = 0.0;
matrix_dbl[2] = 0.0;
matrix_dbl[3] = 1.0;
if (out_layout == AV_CH_LAYOUT_QUAD) {
matrix_dbl[4] = 0.5;
matrix_dbl[5] = (-0.5);
matrix_dbl[6] = (-0.5);
matrix_dbl[7] = 0.5;
} else if (out_layout == AV_CH_LAYOUT_5POINT1 || out_layout == AV_CH_LAYOUT_7POINT1) {
matrix_dbl[4] = 0.5;
matrix_dbl[5] = 0.5;
matrix_dbl[6] = 0.0;
matrix_dbl[7] = 0.0;
matrix_dbl[8] = 0.5;
matrix_dbl[9] = (-0.5);
matrix_dbl[10] = (-0.5);
matrix_dbl[11] = 0.5;
if (out_layout == AV_CH_LAYOUT_7POINT1) {
matrix_dbl[12] = 0.5;
matrix_dbl[13] = (-0.5);
matrix_dbl[14] = (-0.5);
matrix_dbl[15] = 0.5;
}
}
} else {
const double center_mix_level = M_SQRT1_2;
const double surround_mix_level = M_SQRT1_2;
const double lfe_mix_level = M_SQRT1_2;
const int normalize = 0;
ret = avresample_build_matrix(in_layout, out_layout, center_mix_level, surround_mix_level, lfe_mix_level, normalize, matrix_dbl, in_ch, AV_MATRIX_ENCODING_NONE);
if (ret < 0) {
TRACE(_T("Mixer: avresample_build_matrix failed\n"));
av_free(matrix_dbl);
avresample_free(&m_pAVRCxt);
return;
}
}
#ifdef _DEBUG
CString matrix_str;
for (int j = 0; j < out_ch; j++) {
matrix_str.AppendFormat(_T("%d:"), j + 1);
for (int i = 0; i < in_ch; i++) {
double k = matrix_dbl[j * in_ch + i];
matrix_str.AppendFormat(_T(" %.4f"), k);
}
matrix_str += _T("\n");
}
TRACE(matrix_str);
#endif
// Set Matrix on the context
ret = avresample_set_matrix(m_pAVRCxt, matrix_dbl, in_ch);
av_free(matrix_dbl);
if (ret < 0) {
TRACE(_T("Mixer: avresample_set_matrix failed\n"));
avresample_free(&m_pAVRCxt);
return;
}
last_in_layout = in_layout;
last_out_layout = out_layout;
last_in_sf = in_sf;
}
void CMixer::Init(AVSampleFormat in_avsf, DWORD in_layout, DWORD out_layout, float matrix_norm, int in_samplerate, int out_samplerate)
{
// reset parameters
m_in_avsf = AV_SAMPLE_FMT_NONE;
m_in_layout = 0;
m_out_layout = 0;
m_matrix_norm = 0.0f;
m_in_samplerate = 0;
m_out_samplerate = 0;
av_free(m_matrix_dbl);
// Close Resample Context
avresample_close(m_pAVRCxt);
int ret = 0;
// Set options
av_opt_set_int(m_pAVRCxt, "in_sample_fmt", in_avsf, 0);
av_opt_set_int(m_pAVRCxt, "out_sample_fmt", AV_SAMPLE_FMT_FLT, 0); // forced float output
av_opt_set_int(m_pAVRCxt, "in_channel_layout", in_layout, 0);
av_opt_set_int(m_pAVRCxt, "out_channel_layout", out_layout, 0);
av_opt_set_int(m_pAVRCxt, "in_sample_rate", in_samplerate, 0);
av_opt_set_int(m_pAVRCxt, "out_sample_rate", out_samplerate, 0);
// Open Resample Context
ret = avresample_open(m_pAVRCxt);
if (ret < 0) {
TRACE(_T("Mixer: avresample_open failed\n"));
return;
}
// Create Matrix
int in_ch = av_popcount(in_layout);
int out_ch = av_popcount(out_layout);
m_matrix_dbl = (double*)av_mallocz(in_ch * out_ch * sizeof(*m_matrix_dbl));
// expand stereo
if (in_layout == AV_CH_LAYOUT_STEREO && (out_layout == AV_CH_LAYOUT_QUAD || out_layout == AV_CH_LAYOUT_5POINT1 || out_layout == AV_CH_LAYOUT_7POINT1)) {
m_matrix_dbl[0] = 1.0;
m_matrix_dbl[1] = 0.0;
m_matrix_dbl[2] = 0.0;
m_matrix_dbl[3] = 1.0;
if (out_layout == AV_CH_LAYOUT_QUAD) {
m_matrix_dbl[4] = 0.6666;
m_matrix_dbl[5] = (-0.2222);
m_matrix_dbl[6] = (-0.2222);
m_matrix_dbl[7] = 0.6666;
} else if (out_layout == AV_CH_LAYOUT_5POINT1 || out_layout == AV_CH_LAYOUT_7POINT1) {
m_matrix_dbl[4] = 0.5;
m_matrix_dbl[5] = 0.5;
m_matrix_dbl[6] = 0.0;
m_matrix_dbl[7] = 0.0;
m_matrix_dbl[8] = 0.6666;
m_matrix_dbl[9] = (-0.2222);
m_matrix_dbl[10] = (-0.2222);
m_matrix_dbl[11] = 0.6666;
if (out_layout == AV_CH_LAYOUT_7POINT1) {
m_matrix_dbl[12] = 0.6666;
m_matrix_dbl[13] = (-0.2222);
m_matrix_dbl[14] = (-0.2222);
m_matrix_dbl[15] = 0.6666;
}
}
} else {
const double center_mix_level = M_SQRT1_2;
const double surround_mix_level = 1.0;
const double lfe_mix_level = 1.0;
const int normalize = 0;
ret = avresample_build_matrix(in_layout, out_layout, center_mix_level, surround_mix_level, lfe_mix_level, normalize, m_matrix_dbl, in_ch, AV_MATRIX_ENCODING_NONE);
if (ret < 0) {
TRACE(_T("Mixer: avresample_build_matrix failed\n"));
av_free(m_matrix_dbl);
return;
}
// if back channels do not have sound, then divide side channels for the back and side
if (out_layout == AV_CH_LAYOUT_7POINT1) {
bool back_no_sound = true;
for (int i = 0; i < in_ch * 2; i++) {
if (m_matrix_dbl[4 * in_ch + i] != 0.0) {
back_no_sound = false;
}
}
if (back_no_sound) {
for (int i = 0; i < in_ch * 2; i++) {
m_matrix_dbl[4 * in_ch + i] = (m_matrix_dbl[6 * in_ch + i] *= M_SQRT1_2);
}
}
}
}
if (matrix_norm > 0.0f && matrix_norm <= 1.0f) { // 0.0 - normalize off; 1.0 - full normalize matrix
double max_peak = 0;
for (int j = 0; j < out_ch; j++) {
double peak = 0;
for (int i = 0; i < in_ch; i++) {
peak += fabs(m_matrix_dbl[j * in_ch + i]);
}
if (peak > max_peak) {
max_peak = peak;
}
}
if (max_peak > 1.0) {
double g = ((max_peak - 1.0) * (1.0 - matrix_norm) + 1.0) / max_peak;
for (int i = 0, n = in_ch * out_ch; i < n; i++) {
m_matrix_dbl[i] *= g;
}
}
}
#ifdef _DEBUG
CString matrix_str = _T("matrix:\n");
for (int j = 0; j < out_ch; j++) {
matrix_str.AppendFormat(_T("%d:"), j + 1);
for (int i = 0; i < in_ch; i++) {
double k = m_matrix_dbl[j * in_ch + i];
matrix_str.AppendFormat(_T(" %.4f"), k);
}
matrix_str += _T("\n");
}
TRACE(matrix_str);
#endif
// Set Matrix on the context
ret = avresample_set_matrix(m_pAVRCxt, m_matrix_dbl, in_ch);
if (ret < 0) {
TRACE(_T("Mixer: avresample_set_matrix failed\n"));
av_free(m_matrix_dbl);
return;
}
m_in_avsf = in_avsf;
m_in_layout = in_layout;
m_out_layout = out_layout;
m_matrix_norm = matrix_norm;
m_in_samplerate = in_samplerate;
m_out_samplerate = out_samplerate;
}