highperfocused/ffmpeg
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

opus: convert to new channel layout API

Browse Source 
Signed-off-by: James Almer <jamrial@gmail.com>
This commit is contained in:
Anton Khirnov 2019-05-28 14:38:10 +02:00 committed by James Almer
parent 045d6b9abf
commit cc37640a72
4 changed files with 61 additions and 35 deletions

57
libavcodec/opus.c

@ -296,14 +296,15 @@ av_cold int ff_opus_parse_extradata(AVCodecContext *avctx,
static const uint8_t default_channel_map[2] = { 0, 1 }; static const uint8_t default_channel_map[2] = { 0, 1 };
int (*channel_reorder)(int, int) = channel_reorder_unknown; int (*channel_reorder)(int, int) = channel_reorder_unknown;
int channels = avctx->ch_layout.nb_channels;
const uint8_t *extradata, *channel_map; const uint8_t *extradata, *channel_map;
int extradata_size; int extradata_size;
int version, channels, map_type, streams, stereo_streams, i, j; int version, map_type, streams, stereo_streams, i, j, ret;
uint64_t layout; AVChannelLayout layout = { 0 };
if (!avctx->extradata) { if (!avctx->extradata) {
if (avctx->channels > 2) { if (channels > 2) {
av_log(avctx, AV_LOG_ERROR, av_log(avctx, AV_LOG_ERROR,
"Multichannel configuration without extradata.\n"); "Multichannel configuration without extradata.\n");
return AVERROR(EINVAL); return AVERROR(EINVAL);
@ -331,7 +332,7 @@ av_cold int ff_opus_parse_extradata(AVCodecContext *avctx,
if (avctx->internal) if (avctx->internal)
avctx->internal->skip_samples = avctx->delay; avctx->internal->skip_samples = avctx->delay;
channels = avctx->extradata ? extradata[9] : (avctx->channels == 1) ? 1 : 2; channels = avctx->extradata ? extradata[9] : (channels == 1) ? 1 : 2;
if (!channels) { if (!channels) {
av_log(avctx, AV_LOG_ERROR, "Zero channel count specified in the extradata\n"); av_log(avctx, AV_LOG_ERROR, "Zero channel count specified in the extradata\n");
return AVERROR_INVALIDDATA; return AVERROR_INVALIDDATA;
@ -346,9 +347,11 @@ av_cold int ff_opus_parse_extradata(AVCodecContext *avctx,
if (channels > 2) { if (channels > 2) {
av_log(avctx, AV_LOG_ERROR, av_log(avctx, AV_LOG_ERROR,
"Channel mapping 0 is only specified for up to 2 channels\n"); "Channel mapping 0 is only specified for up to 2 channels\n");
return AVERROR_INVALIDDATA; ret = AVERROR_INVALIDDATA;
goto fail;
} }
layout = (channels == 1) ? AV_CH_LAYOUT_MONO : AV_CH_LAYOUT_STEREO; layout = (channels == 1) ? (AVChannelLayout)AV_CHANNEL_LAYOUT_MONO :
(AVChannelLayout)AV_CHANNEL_LAYOUT_STEREO;
streams = 1; streams = 1;
stereo_streams = channels - 1; stereo_streams = channels - 1;
channel_map = default_channel_map; channel_map = default_channel_map;
@ -356,7 +359,8 @@ av_cold int ff_opus_parse_extradata(AVCodecContext *avctx,
if (extradata_size < 21 + channels) { if (extradata_size < 21 + channels) {
av_log(avctx, AV_LOG_ERROR, "Invalid extradata size: %d\n", av_log(avctx, AV_LOG_ERROR, "Invalid extradata size: %d\n",
extradata_size); extradata_size);
return AVERROR_INVALIDDATA; ret = AVERROR_INVALIDDATA;
goto fail;
} }
streams = extradata[19]; streams = extradata[19];
@ -365,16 +369,18 @@ av_cold int ff_opus_parse_extradata(AVCodecContext *avctx,
streams + stereo_streams > 255) { streams + stereo_streams > 255) {
av_log(avctx, AV_LOG_ERROR, av_log(avctx, AV_LOG_ERROR,
"Invalid stream/stereo stream count: %d/%d\n", streams, stereo_streams); "Invalid stream/stereo stream count: %d/%d\n", streams, stereo_streams);
return AVERROR_INVALIDDATA; ret = AVERROR_INVALIDDATA;
goto fail;
} }
if (map_type == 1) { if (map_type == 1) {
if (channels > 8) { if (channels > 8) {
av_log(avctx, AV_LOG_ERROR, av_log(avctx, AV_LOG_ERROR,
"Channel mapping 1 is only specified for up to 8 channels\n"); "Channel mapping 1 is only specified for up to 8 channels\n");
return AVERROR_INVALIDDATA; ret = AVERROR_INVALIDDATA;
goto fail;
} }
layout = ff_vorbis_channel_layouts[channels - 1]; av_channel_layout_copy(&layout, &ff_vorbis_ch_layouts[channels - 1]);
channel_reorder = channel_reorder_vorbis; channel_reorder = channel_reorder_vorbis;
} else if (map_type == 2) { } else if (map_type == 2) {
int ambisonic_order = ff_sqrt(channels) - 1; int ambisonic_order = ff_sqrt(channels) - 1;
@ -384,15 +390,20 @@ av_cold int ff_opus_parse_extradata(AVCodecContext *avctx,
"Channel mapping 2 is only specified for channel counts" "Channel mapping 2 is only specified for channel counts"
" which can be written as (n + 1)^2 or (n + 1)^2 + 2" " which can be written as (n + 1)^2 or (n + 1)^2 + 2"
" for nonnegative integer n\n"); " for nonnegative integer n\n");
return AVERROR_INVALIDDATA; ret = AVERROR_INVALIDDATA;
goto fail;
} }
if (channels > 227) { if (channels > 227) {
av_log(avctx, AV_LOG_ERROR, "Too many channels\n"); av_log(avctx, AV_LOG_ERROR, "Too many channels\n");
return AVERROR_INVALIDDATA; ret = AVERROR_INVALIDDATA;
goto fail;
} }
layout = 0; layout.order = AV_CHANNEL_ORDER_UNSPEC;
} else layout.nb_channels = channels;
layout = 0; } else {
layout.order = AV_CHANNEL_ORDER_UNSPEC;
layout.nb_channels = channels;
}
channel_map = extradata + 21; channel_map = extradata + 21;
} else { } else {
@ -401,8 +412,10 @@ av_cold int ff_opus_parse_extradata(AVCodecContext *avctx,
} }
s->channel_maps = av_calloc(channels, sizeof(*s->channel_maps)); s->channel_maps = av_calloc(channels, sizeof(*s->channel_maps));
if (!s->channel_maps) if (!s->channel_maps) {
return AVERROR(ENOMEM); ret = AVERROR(ENOMEM);
goto fail;
}
for (i = 0; i < channels; i++) { for (i = 0; i < channels; i++) {
ChannelMap *map = &s->channel_maps[i]; ChannelMap *map = &s->channel_maps[i];
@ -415,7 +428,8 @@ av_cold int ff_opus_parse_extradata(AVCodecContext *avctx,
av_log(avctx, AV_LOG_ERROR, av_log(avctx, AV_LOG_ERROR,
"Invalid channel map for output channel %d: %d\n", i, idx); "Invalid channel map for output channel %d: %d\n", i, idx);
av_freep(&s->channel_maps); av_freep(&s->channel_maps);
return AVERROR_INVALIDDATA; ret = AVERROR_INVALIDDATA;
goto fail;
} }
/* check that we did not see this index yet */ /* check that we did not see this index yet */
@ -436,12 +450,15 @@ av_cold int ff_opus_parse_extradata(AVCodecContext *avctx,
} }
} }
avctx->channels = channels; av_channel_layout_uninit(&avctx->ch_layout);
avctx->channel_layout = layout; avctx->ch_layout = layout;
s->nb_streams = streams; s->nb_streams = streams;
s->nb_stereo_streams = stereo_streams; s->nb_stereo_streams = stereo_streams;
return 0; return 0;
fail:
av_channel_layout_uninit(&layout);
return ret;
} }
void ff_celt_quant_bands(CeltFrame *f, OpusRangeCoder *rc) void ff_celt_quant_bands(CeltFrame *f, OpusRangeCoder *rc)

4
libavcodec/opusdec.c

@ -458,7 +458,7 @@ static int opus_decode_packet(AVCodecContext *avctx, void *data,
return ret; return ret;
frame->nb_samples = 0; frame->nb_samples = 0;
for (i = 0; i < avctx->channels; i++) { for (i = 0; i < avctx->ch_layout.nb_channels; i++) {
ChannelMap *map = &c->channel_maps[i]; ChannelMap *map = &c->channel_maps[i];
if (!map->copy) if (!map->copy)
c->streams[map->stream_idx].out[map->channel_idx] = (float*)frame->extended_data[i]; c->streams[map->stream_idx].out[map->channel_idx] = (float*)frame->extended_data[i];
@ -541,7 +541,7 @@ static int opus_decode_packet(AVCodecContext *avctx, void *data,
} }
} }
for (i = 0; i < avctx->channels; i++) { for (i = 0; i < avctx->ch_layout.nb_channels; i++) {
ChannelMap *map = &c->channel_maps[i]; ChannelMap *map = &c->channel_maps[i];
/* handle copied channels */ /* handle copied channels */

15
libavcodec/opusenc.c

@ -66,7 +66,7 @@ static void opus_write_extradata(AVCodecContext *avctx)
bytestream_put_buffer(&bs, "OpusHead", 8); bytestream_put_buffer(&bs, "OpusHead", 8);
bytestream_put_byte (&bs, 0x1); bytestream_put_byte (&bs, 0x1);
bytestream_put_byte (&bs, avctx->channels); bytestream_put_byte (&bs, avctx->ch_layout.nb_channels);
bytestream_put_le16 (&bs, avctx->initial_padding); bytestream_put_le16 (&bs, avctx->initial_padding);
bytestream_put_le32 (&bs, avctx->sample_rate); bytestream_put_le32 (&bs, avctx->sample_rate);
bytestream_put_le16 (&bs, 0x0); bytestream_put_le16 (&bs, 0x0);
@ -518,11 +518,16 @@ static void opus_packet_assembler(OpusEncContext *s, AVPacket *avpkt)
static AVFrame *spawn_empty_frame(OpusEncContext *s) static AVFrame *spawn_empty_frame(OpusEncContext *s)
{ {
AVFrame *f = av_frame_alloc(); AVFrame *f = av_frame_alloc();
int ret;
if (!f) if (!f)
return NULL; return NULL;
f->format = s->avctx->sample_fmt; f->format = s->avctx->sample_fmt;
f->nb_samples = s->avctx->frame_size; f->nb_samples = s->avctx->frame_size;
f->channel_layout = s->avctx->channel_layout; ret = av_channel_layout_copy(&f->ch_layout, &s->avctx->ch_layout);
if (ret < 0) {
av_frame_free(&f);
return NULL;
}
if (av_frame_get_buffer(f, 4)) { if (av_frame_get_buffer(f, 4)) {
av_frame_free(&f); av_frame_free(&f);
return NULL; return NULL;
@ -626,7 +631,7 @@ static av_cold int opus_encode_init(AVCodecContext *avctx)
OpusEncContext *s = avctx->priv_data; OpusEncContext *s = avctx->priv_data;
s->avctx = avctx; s->avctx = avctx;
s->channels = avctx->channels; s->channels = avctx->ch_layout.nb_channels;
/* Opus allows us to change the framesize on each packet (and each packet may /* Opus allows us to change the framesize on each packet (and each packet may
* have multiple frames in it) but we can't change the codec's frame size on * have multiple frames in it) but we can't change the codec's frame size on
@ -734,8 +739,12 @@ const AVCodec ff_opus_encoder = {
.caps_internal = FF_CODEC_CAP_INIT_THREADSAFE | FF_CODEC_CAP_INIT_CLEANUP, .caps_internal = FF_CODEC_CAP_INIT_THREADSAFE | FF_CODEC_CAP_INIT_CLEANUP,
.capabilities = AV_CODEC_CAP_EXPERIMENTAL | AV_CODEC_CAP_SMALL_LAST_FRAME | AV_CODEC_CAP_DELAY, .capabilities = AV_CODEC_CAP_EXPERIMENTAL | AV_CODEC_CAP_SMALL_LAST_FRAME | AV_CODEC_CAP_DELAY,
.supported_samplerates = (const int []){ 48000, 0 }, .supported_samplerates = (const int []){ 48000, 0 },
#if FF_API_OLD_CHANNEL_LAYOUT
.channel_layouts = (const uint64_t []){ AV_CH_LAYOUT_MONO, .channel_layouts = (const uint64_t []){ AV_CH_LAYOUT_MONO,
AV_CH_LAYOUT_STEREO, 0 }, AV_CH_LAYOUT_STEREO, 0 },
#endif
.ch_layouts = (const AVChannelLayout []){ AV_CHANNEL_LAYOUT_MONO,
AV_CHANNEL_LAYOUT_STEREO, { 0 } },
.sample_fmts = (const enum AVSampleFormat[]){ AV_SAMPLE_FMT_FLTP, .sample_fmts = (const enum AVSampleFormat[]){ AV_SAMPLE_FMT_FLTP,
AV_SAMPLE_FMT_NONE }, AV_SAMPLE_FMT_NONE },
}; };

20
libavcodec/opusenc_psy.c

@ -83,7 +83,7 @@ static void step_collect_psy_metrics(OpusPsyContext *s, int index)
st->index = index; st->index = index;
for (ch = 0; ch < s->avctx->channels; ch++) { for (ch = 0; ch < s->avctx->ch_layout.nb_channels; ch++) {
const int lap_size = (1 << s->bsize_analysis); const int lap_size = (1 << s->bsize_analysis);
for (i = 1; i <= FFMIN(lap_size, index); i++) { for (i = 1; i <= FFMIN(lap_size, index); i++) {
const int offset = i*120; const int offset = i*120;
@ -105,7 +105,7 @@ static void step_collect_psy_metrics(OpusPsyContext *s, int index)
st->bands[ch][i] = &st->coeffs[ch][ff_celt_freq_bands[i] << s->bsize_analysis]; st->bands[ch][i] = &st->coeffs[ch][ff_celt_freq_bands[i] << s->bsize_analysis];
} }
for (ch = 0; ch < s->avctx->channels; ch++) { for (ch = 0; ch < s->avctx->ch_layout.nb_channels; ch++) {
for (i = 0; i < CELT_MAX_BANDS; i++) { for (i = 0; i < CELT_MAX_BANDS; i++) {
float avg_c_s, energy = 0.0f, dist_dev = 0.0f; float avg_c_s, energy = 0.0f, dist_dev = 0.0f;
const int range = ff_celt_freq_range[i] << s->bsize_analysis; const int range = ff_celt_freq_range[i] << s->bsize_analysis;
@ -128,7 +128,7 @@ static void step_collect_psy_metrics(OpusPsyContext *s, int index)
st->silence = !silence; st->silence = !silence;
if (s->avctx->channels > 1) { if (s->avctx->ch_layout.nb_channels > 1) {
for (i = 0; i < CELT_MAX_BANDS; i++) { for (i = 0; i < CELT_MAX_BANDS; i++) {
float incompat = 0.0f; float incompat = 0.0f;
const float *coeffs1 = st->bands[0][i]; const float *coeffs1 = st->bands[0][i];
@ -140,7 +140,7 @@ static void step_collect_psy_metrics(OpusPsyContext *s, int index)
} }
} }
for (ch = 0; ch < s->avctx->channels; ch++) { for (ch = 0; ch < s->avctx->ch_layout.nb_channels; ch++) {
for (i = 0; i < CELT_MAX_BANDS; i++) { for (i = 0; i < CELT_MAX_BANDS; i++) {
OpusBandExcitation *ex = &s->ex[ch][i]; OpusBandExcitation *ex = &s->ex[ch][i];
float bp_e = bessel_filter(&s->bfilter_lo[ch][i], st->energy[ch][i]); float bp_e = bessel_filter(&s->bfilter_lo[ch][i], st->energy[ch][i]);
@ -259,7 +259,7 @@ void ff_opus_psy_celt_frame_init(OpusPsyContext *s, CeltFrame *f, int index)
f->start_band = (s->p.mode == OPUS_MODE_HYBRID) ? 17 : 0; f->start_band = (s->p.mode == OPUS_MODE_HYBRID) ? 17 : 0;
f->end_band = ff_celt_band_end[s->p.bandwidth]; f->end_band = ff_celt_band_end[s->p.bandwidth];
f->channels = s->avctx->channels; f->channels = s->avctx->ch_layout.nb_channels;
f->size = s->p.framesize; f->size = s->p.framesize;
for (i = 0; i < (1 << f->size); i++) for (i = 0; i < (1 << f->size); i++)
@ -327,7 +327,7 @@ static void celt_gauge_psy_weight(OpusPsyContext *s, OpusPsyStep **start,
float tonal_contrib = 0.0f; float tonal_contrib = 0.0f;
for (f = 0; f < (1 << s->p.framesize); f++) { for (f = 0; f < (1 << s->p.framesize); f++) {
weight = start[f]->stereo[i]; weight = start[f]->stereo[i];
for (ch = 0; ch < s->avctx->channels; ch++) { for (ch = 0; ch < s->avctx->ch_layout.nb_channels; ch++) {
weight += start[f]->change_amp[ch][i] + start[f]->tone[ch][i] + start[f]->energy[ch][i]; weight += start[f]->change_amp[ch][i] + start[f]->tone[ch][i] + start[f]->energy[ch][i];
tonal_contrib += start[f]->tone[ch][i]; tonal_contrib += start[f]->tone[ch][i];
} }
@ -384,7 +384,7 @@ static void celt_search_for_dual_stereo(OpusPsyContext *s, CeltFrame *f)
float td1, td2; float td1, td2;
f->dual_stereo = 0; f->dual_stereo = 0;
if (s->avctx->channels < 2) if (s->avctx->ch_layout.nb_channels < 2)
return; return;
bands_dist(s, f, &td1); bands_dist(s, f, &td1);
@ -402,7 +402,7 @@ static void celt_search_for_intensity(OpusPsyContext *s, CeltFrame *f)
/* TODO: fix, make some heuristic up here using the lambda value */ /* TODO: fix, make some heuristic up here using the lambda value */
float end_band = 0; float end_band = 0;
if (s->avctx->channels < 2) if (s->avctx->ch_layout.nb_channels < 2)
return; return;
for (i = f->end_band; i >= end_band; i--) { for (i = f->end_band; i >= end_band; i--) {
@ -436,7 +436,7 @@ static int celt_search_for_tf(OpusPsyContext *s, OpusPsyStep **start, CeltFrame
float iscore0 = 0.0f; float iscore0 = 0.0f;
float iscore1 = 0.0f; float iscore1 = 0.0f;
for (j = 0; j < (1 << f->size); j++) { for (j = 0; j < (1 << f->size); j++) {
for (k = 0; k < s->avctx->channels; k++) { for (k = 0; k < s->avctx->ch_layout.nb_channels; k++) {
iscore0 += start[j]->tone[k][i]*start[j]->change_amp[k][i]/mag[0]; iscore0 += start[j]->tone[k][i]*start[j]->change_amp[k][i]/mag[0];
iscore1 += start[j]->tone[k][i]*start[j]->change_amp[k][i]/mag[1]; iscore1 += start[j]->tone[k][i]*start[j]->change_amp[k][i]/mag[1];
} }
@ -540,7 +540,7 @@ av_cold int ff_opus_psy_init(OpusPsyContext *s, AVCodecContext *avctx,
goto fail; goto fail;
} }
for (ch = 0; ch < s->avctx->channels; ch++) { for (ch = 0; ch < s->avctx->ch_layout.nb_channels; ch++) {
for (i = 0; i < CELT_MAX_BANDS; i++) { for (i = 0; i < CELT_MAX_BANDS; i++) {
bessel_init(&s->bfilter_hi[ch][i], 1.0f, 19.0f, 100.0f, 1); bessel_init(&s->bfilter_hi[ch][i], 1.0f, 19.0f, 100.0f, 1);
bessel_init(&s->bfilter_lo[ch][i], 1.0f, 20.0f, 100.0f, 0); bessel_init(&s->bfilter_lo[ch][i], 1.0f, 20.0f, 100.0f, 0);