2 Copyright (C) 2013 Paul Brossier <piem@aubio.org>
4 This file is part of aubio.
6 aubio is free software: you can redistribute it and/or modify
7 it under the terms of the GNU General Public License as published by
8 the Free Software Foundation, either version 3 of the License, or
9 (at your option) any later version.
11 aubio is distributed in the hope that it will be useful,
12 but WITHOUT ANY WARRANTY; without even the implied warranty of
13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 GNU General Public License for more details.
16 You should have received a copy of the GNU General Public License
17 along with aubio. If not, see <http://www.gnu.org/licenses/>.
21 #include "aubio_priv.h"
25 #include <libavcodec/avcodec.h>
26 #include <libavformat/avformat.h>
27 #if defined(HAVE_SWRESAMPLE)
28 #include <libswresample/swresample.h>
29 #elif defined(HAVE_AVRESAMPLE)
30 #include <libavresample/avresample.h>
32 #include <libavutil/opt.h>
34 // determine whether we use libavformat from ffmpeg or from libav
35 #define FFMPEG_LIBAVFORMAT (LIBAVFORMAT_VERSION_MICRO > 99 )
36 // max_analyze_duration2 was used from ffmpeg libavformat 55.43.100 -> 57.2.100
37 #define FFMPEG_LIBAVFORMAT_MAX_DUR2 FFMPEG_LIBAVFORMAT && ( \
38 (LIBAVFORMAT_VERSION_MAJOR == 55 && LIBAVFORMAT_VERSION_MINOR >= 43) \
39 || (LIBAVFORMAT_VERSION_MAJOR == 56) \
40 || (LIBAVFORMAT_VERSION_MAJOR == 57 && LIBAVFORMAT_VERSION_MINOR < 2) \
43 // backward compatibility with libavcodec55
44 #if LIBAVCODEC_VERSION_INT < AV_VERSION_INT(57,0,0)
45 #define HAVE_AUBIO_LIBAVCODEC_DEPRECATED 1
48 #if LIBAVCODEC_VERSION_INT < AV_VERSION_INT(58,3,102)
49 #define HAVE_AUBIO_LIBAVCODEC_TIMEBASE_FIX 1
52 #if LIBAVCODEC_VERSION_INT < AV_VERSION_INT(55,28,1)
53 #warning "libavcodec < 56 is deprecated"
54 #define av_frame_alloc avcodec_alloc_frame
55 #define av_frame_free avcodec_free_frame
56 #define av_packet_unref av_free_packet
59 #include "aubio_priv.h"
63 #include "source_avcodec.h"
65 #if LIBAVCODEC_VERSION_INT < AV_VERSION_INT(56, 56, 0)
66 #define AUBIO_AVCODEC_MAX_BUFFER_SIZE FF_MIN_BUFFER_SIZE
68 #define AUBIO_AVCODEC_MAX_BUFFER_SIZE AV_INPUT_BUFFER_MIN_SIZE
71 struct _aubio_source_avcodec_t {
76 // some data about the file
78 uint_t input_samplerate;
79 uint_t input_channels;
82 AVFormatContext *avFormatCtx;
83 AVCodecContext *avCodecCtx;
86 #ifdef HAVE_AVRESAMPLE
87 AVAudioResampleContext *avr;
88 #elif defined(HAVE_SWRESAMPLE)
94 sint_t selected_stream;
98 // create or re-create the context when _do or _do_multi is called
99 void aubio_source_avcodec_reset_resampler(aubio_source_avcodec_t * s);
100 // actually read a frame
101 void aubio_source_avcodec_readframe(aubio_source_avcodec_t *s,
102 uint_t * read_samples);
104 uint_t aubio_source_avcodec_has_network_url(aubio_source_avcodec_t *s);
106 uint_t aubio_source_avcodec_has_network_url(aubio_source_avcodec_t *s) {
107 char proto[20], authorization[256], hostname[128], uripath[256];
108 int proto_size = 20, authorization_size = 256, hostname_size = 128,
109 *port_ptr = 0, path_size = 256;
110 av_url_split(proto, proto_size, authorization, authorization_size, hostname,
111 hostname_size, port_ptr, uripath, path_size, s->path);
119 aubio_source_avcodec_t * new_aubio_source_avcodec(const char_t * path,
120 uint_t samplerate, uint_t hop_size) {
121 aubio_source_avcodec_t * s = AUBIO_NEW(aubio_source_avcodec_t);
122 AVFormatContext *avFormatCtx = NULL;
123 AVCodecContext *avCodecCtx = NULL;
124 AVFrame *avFrame = NULL;
125 sint_t selected_stream = -1;
126 #if FF_API_LAVF_AVCTX
127 AVCodecParameters *codecpar;
133 AUBIO_ERR("source_avcodec: Aborted opening null path\n");
136 if ((sint_t)samplerate < 0) {
137 AUBIO_ERR("source_avcodec: Can not open %s with samplerate %d\n",
141 if ((sint_t)hop_size <= 0) {
142 AUBIO_ERR("source_avcodec: Can not open %s with hop_size %d\n",
147 s->hop_size = hop_size;
150 s->path = AUBIO_ARRAY(char_t, strnlen(path, PATH_MAX) + 1);
151 strncpy(s->path, path, strnlen(path, PATH_MAX) + 1);
153 #if LIBAVFORMAT_VERSION_INT < AV_VERSION_INT(58,0,0)
154 // register all formats and codecs
158 if (aubio_source_avcodec_has_network_url(s)) {
159 avformat_network_init();
162 // try opening the file and get some info about it
164 if ( (err = avformat_open_input(&avFormatCtx, s->path, NULL, NULL) ) < 0 ) {
166 av_strerror (err, errorstr, sizeof(errorstr));
167 AUBIO_ERR("source_avcodec: Failed opening %s (%s)\n", s->path, errorstr);
171 // try to make sure max_analyze_duration is big enough for most songs
172 #if FFMPEG_LIBAVFORMAT_MAX_DUR2
173 avFormatCtx->max_analyze_duration2 *= 100;
175 avFormatCtx->max_analyze_duration *= 100;
178 // retrieve stream information
179 if ( (err = avformat_find_stream_info(avFormatCtx, NULL)) < 0 ) {
181 av_strerror (err, errorstr, sizeof(errorstr));
182 AUBIO_ERR("source_avcodec: Could not find stream information "
183 "for %s (%s)\n", s->path, errorstr);
187 // dump information about file onto standard error
188 //av_dump_format(avFormatCtx, 0, s->path, 0);
190 // look for the first audio stream
191 for (i = 0; i < avFormatCtx->nb_streams; i++) {
192 #if FF_API_LAVF_AVCTX
193 if (avFormatCtx->streams[i]->codecpar->codec_type == AVMEDIA_TYPE_AUDIO) {
195 if (avFormatCtx->streams[i]->codec->codec_type == AVMEDIA_TYPE_AUDIO) {
197 if (selected_stream == -1) {
200 AUBIO_WRN("source_avcodec: More than one audio stream in %s, "
201 "taking the first one\n", s->path);
205 if (selected_stream == -1) {
206 AUBIO_ERR("source_avcodec: No audio stream in %s\n", s->path);
209 //AUBIO_DBG("Taking stream %d in file %s\n", selected_stream, s->path);
210 s->selected_stream = selected_stream;
212 #if FF_API_LAVF_AVCTX
213 codecpar = avFormatCtx->streams[selected_stream]->codecpar;
214 if (codecpar == NULL) {
215 AUBIO_ERR("source_avcodec: Could not find decoder for %s", s->path);
218 codec = avcodec_find_decoder(codecpar->codec_id);
220 /* Allocate a codec context for the decoder */
221 avCodecCtx = avcodec_alloc_context3(codec);
223 AUBIO_ERR("source_avcodec: Failed to allocate the %s codec context "
224 "for path %s\n", av_get_media_type_string(AVMEDIA_TYPE_AUDIO),
229 avCodecCtx = avFormatCtx->streams[selected_stream]->codec;
230 codec = avcodec_find_decoder(avCodecCtx->codec_id);
233 AUBIO_ERR("source_avcodec: Could not find decoder for %s", s->path);
237 #if FF_API_LAVF_AVCTX
238 /* Copy codec parameters from input stream to output codec context */
239 if ((err = avcodec_parameters_to_context(avCodecCtx, codecpar)) < 0) {
240 AUBIO_ERR("source_avcodec: Failed to copy %s codec parameters to "
241 "decoder context for %s\n",
242 av_get_media_type_string(AVMEDIA_TYPE_AUDIO), s->path);
245 #if HAVE_AUBIO_LIBAVCODEC_TIMEBASE_FIX
246 // avoids 'skipped frames warning' with avecodec < 58, deprecated after
247 av_codec_set_pkt_timebase(avCodecCtx,
248 avFormatCtx->streams[selected_stream]->time_base);
252 if ( ( err = avcodec_open2(avCodecCtx, codec, NULL) ) < 0) {
254 av_strerror (err, errorstr, sizeof(errorstr));
255 AUBIO_ERR("source_avcodec: Could not load codec for %s (%s)\n", s->path,
260 /* get input specs */
261 s->input_samplerate = avCodecCtx->sample_rate;
262 s->input_channels = avCodecCtx->channels;
263 //AUBIO_DBG("input_samplerate: %d\n", s->input_samplerate);
264 //AUBIO_DBG("input_channels: %d\n", s->input_channels);
266 if (samplerate == 0) {
267 s->samplerate = s->input_samplerate;
269 s->samplerate = samplerate;
272 if (s->samplerate > s->input_samplerate) {
273 AUBIO_WRN("source_avcodec: upsampling %s from %d to %d\n", s->path,
274 s->input_samplerate, s->samplerate);
277 avFrame = av_frame_alloc();
279 AUBIO_ERR("source_avcodec: Could not allocate frame for (%s)\n", s->path);
282 /* allocate output for avr */
283 s->output = (smpl_t *)av_malloc(AUBIO_AVCODEC_MAX_BUFFER_SIZE
289 s->avFormatCtx = avFormatCtx;
290 s->avCodecCtx = avCodecCtx;
291 s->avFrame = avFrame;
293 aubio_source_avcodec_reset_resampler(s);
295 if (s->avr == NULL) goto beach;
299 //av_log_set_level(AV_LOG_QUIET);
304 //AUBIO_ERR("can not read %s at samplerate %dHz with a hop_size of %d\n",
305 // s->path, s->samplerate, s->hop_size);
306 del_aubio_source_avcodec(s);
310 void aubio_source_avcodec_reset_resampler(aubio_source_avcodec_t * s)
312 // create or reset resampler to/from mono/multi-channel
313 if ( s->avr == NULL ) {
315 int64_t input_layout = av_get_default_channel_layout(s->input_channels);
316 int64_t output_layout = av_get_default_channel_layout(s->input_channels);
317 #ifdef HAVE_AVRESAMPLE
318 AVAudioResampleContext *avr = avresample_alloc_context();
319 #elif defined(HAVE_SWRESAMPLE)
320 SwrContext *avr = swr_alloc();
321 #endif /* HAVE_AVRESAMPLE || HAVE_SWRESAMPLE */
323 av_opt_set_int(avr, "in_channel_layout", input_layout, 0);
324 av_opt_set_int(avr, "out_channel_layout", output_layout, 0);
325 av_opt_set_int(avr, "in_sample_rate", s->input_samplerate, 0);
326 av_opt_set_int(avr, "out_sample_rate", s->samplerate, 0);
327 av_opt_set_int(avr, "in_sample_fmt", s->avCodecCtx->sample_fmt, 0);
328 #if HAVE_AUBIO_DOUBLE
329 av_opt_set_int(avr, "out_sample_fmt", AV_SAMPLE_FMT_DBL, 0);
331 av_opt_set_int(avr, "out_sample_fmt", AV_SAMPLE_FMT_FLT, 0);
334 //av_opt_set_int(avr, "out_sample_fmt", AV_SAMPLE_FMT_FLTP, 0);
335 #ifdef HAVE_AVRESAMPLE
336 if ( ( err = avresample_open(avr) ) < 0)
337 #elif defined(HAVE_SWRESAMPLE)
338 if ( ( err = swr_init(avr) ) < 0)
339 #endif /* HAVE_AVRESAMPLE || HAVE_SWRESAMPLE */
342 av_strerror (err, errorstr, sizeof(errorstr));
343 AUBIO_ERR("source_avcodec: Could not open resampling context"
344 " for %s (%s)\n", s->path, errorstr);
351 void aubio_source_avcodec_readframe(aubio_source_avcodec_t *s,
352 uint_t * read_samples)
354 AVFormatContext *avFormatCtx = s->avFormatCtx;
355 AVCodecContext *avCodecCtx = s->avCodecCtx;
356 AVFrame *avFrame = s->avFrame;
357 AVPacket avPacket = s->avPacket;
358 #ifdef HAVE_AVRESAMPLE
359 AVAudioResampleContext *avr = s->avr;
360 #elif defined(HAVE_SWRESAMPLE)
361 SwrContext *avr = s->avr;
362 #endif /* HAVE_AVRESAMPLE || HAVE_SWRESAMPLE */
364 #ifdef HAVE_AVRESAMPLE
366 int in_samples = avFrame->nb_samples;
367 int out_linesize = 0;
368 int max_out_samples = AUBIO_AVCODEC_MAX_BUFFER_SIZE;
370 #elif defined(HAVE_SWRESAMPLE)
371 int in_samples = avFrame->nb_samples;
372 int max_out_samples = AUBIO_AVCODEC_MAX_BUFFER_SIZE / avCodecCtx->channels;
374 #endif /* HAVE_AVRESAMPLE || HAVE_SWRESAMPLE */
375 smpl_t *output = s->output;
376 #ifndef FF_API_LAVF_AVCTX
381 av_init_packet (&avPacket);
386 int err = av_read_frame (avFormatCtx, &avPacket);
387 if (err == AVERROR_EOF) {
393 av_strerror (err, errorstr, sizeof(errorstr));
394 AUBIO_ERR("source_avcodec: could not read frame in %s (%s)\n",
399 } while (avPacket.stream_index != s->selected_stream);
401 #if FF_API_LAVF_AVCTX
402 ret = avcodec_send_packet(avCodecCtx, &avPacket);
403 if (ret < 0 && ret != AVERROR_EOF) {
404 AUBIO_ERR("source_avcodec: error when sending packet for %s\n", s->path);
407 ret = avcodec_receive_frame(avCodecCtx, avFrame);
412 if (ret == AVERROR(EAGAIN)) {
413 //AUBIO_WRN("source_avcodec: output is not available right now - "
414 // "user must try to send new input\n");
416 } else if (ret == AVERROR_EOF) {
417 AUBIO_WRN("source_avcodec: the decoder has been fully flushed, "
418 "and there will be no more output frames\n");
420 AUBIO_ERR("source_avcodec: decoding errors on %s\n", s->path);
425 len = avcodec_decode_audio4(avCodecCtx, avFrame, &got_frame, &avPacket);
428 AUBIO_ERR("source_avcodec: error while decoding %s\n", s->path);
432 if (got_frame == 0) {
433 AUBIO_WRN("source_avcodec: did not get a frame when reading %s\n",
438 #if LIBAVUTIL_VERSION_MAJOR > 52
439 if (avFrame->channels != (sint_t)s->input_channels) {
440 AUBIO_WRN ("source_avcodec: trying to read from %d channel(s),"
441 "but configured for %d; is '%s' corrupt?\n",
442 avFrame->channels, s->input_channels, s->path);
446 #warning "avutil < 53 is deprecated, crashes might occur on corrupt files"
449 #ifdef HAVE_AVRESAMPLE
451 av_samples_get_buffer_size(&in_linesize, avCodecCtx->channels,
452 avFrame->nb_samples, avCodecCtx->sample_fmt, 1);
453 in_samples = avFrame->nb_samples;
455 max_out_samples = AUBIO_AVCODEC_MAX_BUFFER_SIZE;
456 out_samples = avresample_convert ( avr,
457 (uint8_t **)&output, out_linesize, max_out_samples,
458 (uint8_t **)avFrame->data, in_linesize, in_samples);
459 #elif defined(HAVE_SWRESAMPLE)
460 in_samples = avFrame->nb_samples;
461 max_out_samples = AUBIO_AVCODEC_MAX_BUFFER_SIZE / avCodecCtx->channels;
462 out_samples = swr_convert( avr,
463 (uint8_t **)&output, max_out_samples,
464 (const uint8_t **)avFrame->data, in_samples);
465 #endif /* HAVE_AVRESAMPLE || HAVE_SWRESAMPLE */
466 if (out_samples < 0) {
467 AUBIO_WRN("source_avcodec: error while resampling %s (%d)\n",
468 s->path, out_samples);
472 *read_samples = out_samples;
475 av_packet_unref(&avPacket);
478 void aubio_source_avcodec_do(aubio_source_avcodec_t * s, fvec_t * read_data,
482 uint_t total_wrote = 0;
483 uint_t length = aubio_source_validate_input_length("source_avcodec", s->path,
484 s->hop_size, read_data->length);
485 if (!s->avr || !s->avFormatCtx || !s->avCodecCtx) {
486 AUBIO_ERR("source_avcodec: could not read from %s (file was closed)\n",
491 while (total_wrote < length) {
492 end = MIN(s->read_samples - s->read_index, length - total_wrote);
493 for (i = 0; i < end; i++) {
494 read_data->data[i + total_wrote] = 0.;
495 for (j = 0; j < s->input_channels; j++) {
496 read_data->data[i + total_wrote] +=
497 s->output[(i + s->read_index) * s->input_channels + j];
499 read_data->data[i + total_wrote] *= 1./s->input_channels;
502 if (total_wrote < length) {
503 uint_t avcodec_read = 0;
504 aubio_source_avcodec_readframe(s, &avcodec_read);
505 s->read_samples = avcodec_read;
511 s->read_index += end;
515 aubio_source_pad_output(read_data, total_wrote);
520 void aubio_source_avcodec_do_multi(aubio_source_avcodec_t * s,
521 fmat_t * read_data, uint_t * read) {
524 uint_t total_wrote = 0;
525 uint_t length = aubio_source_validate_input_length("source_avcodec", s->path,
526 s->hop_size, read_data->length);
527 uint_t channels = aubio_source_validate_input_channels("source_avcodec",
528 s->path, s->input_channels, read_data->height);
529 if (!s->avr || !s->avFormatCtx || !s->avCodecCtx) {
530 AUBIO_ERR("source_avcodec: could not read from %s (file was closed)\n",
535 while (total_wrote < length) {
536 end = MIN(s->read_samples - s->read_index, length - total_wrote);
537 for (j = 0; j < channels; j++) {
538 for (i = 0; i < end; i++) {
539 read_data->data[j][i + total_wrote] =
540 s->output[(i + s->read_index) * s->input_channels + j];
544 if (total_wrote < length) {
545 uint_t avcodec_read = 0;
546 aubio_source_avcodec_readframe(s, &avcodec_read);
547 s->read_samples = avcodec_read;
553 s->read_index += end;
557 aubio_source_pad_multi_output(read_data, s->input_channels, total_wrote);
562 uint_t aubio_source_avcodec_get_samplerate(const aubio_source_avcodec_t * s) {
563 return s->samplerate;
566 uint_t aubio_source_avcodec_get_channels(const aubio_source_avcodec_t * s) {
567 return s->input_channels;
570 uint_t aubio_source_avcodec_seek (aubio_source_avcodec_t * s, uint_t pos) {
571 int64_t resampled_pos =
572 (uint_t)ROUND(pos * (s->input_samplerate * 1. / s->samplerate));
573 int64_t min_ts = MAX(resampled_pos - 2000, 0);
574 int64_t max_ts = MIN(resampled_pos + 2000, INT64_MAX);
575 int seek_flags = AVSEEK_FLAG_FRAME | AVSEEK_FLAG_ANY;
576 int ret = AUBIO_FAIL;
577 if (s->avFormatCtx != NULL && s->avr != NULL) {
580 AUBIO_ERR("source_avcodec: failed seeking in %s (file not opened?)",
584 if ((sint_t)pos < 0) {
585 AUBIO_ERR("source_avcodec: could not seek %s at %d (seeking position"
586 " should be >= 0)\n", s->path, pos);
589 ret = avformat_seek_file(s->avFormatCtx, s->selected_stream,
590 min_ts, resampled_pos, max_ts, seek_flags);
592 AUBIO_ERR("source_avcodec: failed seeking to %d in file %s",
599 #ifdef HAVE_AVRESAMPLE
600 // reset the AVAudioResampleContext
601 avresample_close(s->avr);
602 avresample_open(s->avr);
603 #elif defined(HAVE_SWRESAMPLE)
610 uint_t aubio_source_avcodec_get_duration (aubio_source_avcodec_t * s) {
611 if (s && &(s->avFormatCtx) != NULL) {
612 int64_t duration = s->avFormatCtx->duration;
613 return s->samplerate * ((uint_t)duration / 1e6 );
618 uint_t aubio_source_avcodec_close(aubio_source_avcodec_t * s) {
619 if (s->avr != NULL) {
620 #ifdef HAVE_AVRESAMPLE
621 avresample_close( s->avr );
622 #elif defined(HAVE_SWRESAMPLE)
623 swr_close ( s->avr );
628 if (s->avCodecCtx != NULL) {
629 #ifndef HAVE_AUBIO_LIBAVCODEC_DEPRECATED
630 avcodec_free_context( &s->avCodecCtx );
632 avcodec_close ( s->avCodecCtx );
635 s->avCodecCtx = NULL;
636 if (s->avFormatCtx != NULL) {
637 avformat_close_input(&s->avFormatCtx);
638 s->avFormatCtx = NULL;
640 av_packet_unref(&s->avPacket);
644 void del_aubio_source_avcodec(aubio_source_avcodec_t * s){
646 aubio_source_avcodec_close(s);
647 if (s->output != NULL) {
651 if (s->avFrame != NULL) {
652 av_frame_free( &(s->avFrame) );
662 #endif /* HAVE_LIBAV */