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>
35 // determine whether we use libavformat from ffmpeg or from libav
36 #define FFMPEG_LIBAVFORMAT (LIBAVFORMAT_VERSION_MICRO > 99 )
37 // max_analyze_duration2 was used from ffmpeg libavformat 55.43.100 -> 57.2.100
38 #define FFMPEG_LIBAVFORMAT_MAX_DUR2 FFMPEG_LIBAVFORMAT && ( \
39 (LIBAVFORMAT_VERSION_MAJOR == 55 && LIBAVFORMAT_VERSION_MINOR >= 43) \
40 || (LIBAVFORMAT_VERSION_MAJOR == 56) \
41 || (LIBAVFORMAT_VERSION_MAJOR == 57 && LIBAVFORMAT_VERSION_MINOR < 2) \
44 // backward compatibility with libavcodec55
45 #if LIBAVCODEC_VERSION_INT < AV_VERSION_INT(57,0,0)
46 #define HAVE_AUBIO_LIBAVCODEC_DEPRECATED 1
49 #if LIBAVCODEC_VERSION_INT < AV_VERSION_INT(58,3,102)
50 #define HAVE_AUBIO_LIBAVCODEC_TIMEBASE_FIX 1
53 #if LIBAVCODEC_VERSION_INT < AV_VERSION_INT(55,28,1)
54 #warning "libavcodec < 56 is deprecated"
55 #define av_frame_alloc avcodec_alloc_frame
56 #define av_frame_free avcodec_free_frame
57 #define av_packet_unref av_free_packet
60 #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 = s->avFormatCtx;
123 AVCodecContext *avCodecCtx = s->avCodecCtx;
124 AVFrame *avFrame = s->avFrame;
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 if (s->path) AUBIO_FREE(s->path);
151 s->path = AUBIO_ARRAY(char_t, strnlen(path, PATH_MAX) + 1);
152 strncpy(s->path, path, strnlen(path, PATH_MAX) + 1);
154 #if LIBAVFORMAT_VERSION_INT < AV_VERSION_INT(58,0,0)
155 // register all formats and codecs
159 if (aubio_source_avcodec_has_network_url(s)) {
160 avformat_network_init();
163 // try opening the file and get some info about it
165 if ( (err = avformat_open_input(&avFormatCtx, s->path, NULL, NULL) ) < 0 ) {
167 av_strerror (err, errorstr, sizeof(errorstr));
168 AUBIO_ERR("source_avcodec: Failed opening %s (%s)\n", s->path, errorstr);
172 // try to make sure max_analyze_duration is big enough for most songs
173 #if FFMPEG_LIBAVFORMAT_MAX_DUR2
174 avFormatCtx->max_analyze_duration2 *= 100;
176 avFormatCtx->max_analyze_duration *= 100;
179 // retrieve stream information
180 if ( (err = avformat_find_stream_info(avFormatCtx, NULL)) < 0 ) {
182 av_strerror (err, errorstr, sizeof(errorstr));
183 AUBIO_ERR("source_avcodec: Could not find stream information "
184 "for %s (%s)\n", s->path, errorstr);
188 // dump information about file onto standard error
189 //av_dump_format(avFormatCtx, 0, s->path, 0);
191 // look for the first audio stream
192 for (i = 0; i < avFormatCtx->nb_streams; i++) {
193 #if FF_API_LAVF_AVCTX
194 if (avFormatCtx->streams[i]->codecpar->codec_type == AVMEDIA_TYPE_AUDIO) {
196 if (avFormatCtx->streams[i]->codec->codec_type == AVMEDIA_TYPE_AUDIO) {
198 if (selected_stream == -1) {
201 AUBIO_WRN("source_avcodec: More than one audio stream in %s, "
202 "taking the first one\n", s->path);
206 if (selected_stream == -1) {
207 AUBIO_ERR("source_avcodec: No audio stream in %s\n", s->path);
210 //AUBIO_DBG("Taking stream %d in file %s\n", selected_stream, s->path);
211 s->selected_stream = selected_stream;
213 #if FF_API_LAVF_AVCTX
214 codecpar = avFormatCtx->streams[selected_stream]->codecpar;
215 if (codecpar == NULL) {
216 AUBIO_ERR("source_avcodec: Could not find decoder for %s", s->path);
219 codec = avcodec_find_decoder(codecpar->codec_id);
221 /* Allocate a codec context for the decoder */
222 avCodecCtx = avcodec_alloc_context3(codec);
224 AUBIO_ERR("source_avcodec: Failed to allocate the %s codec context "
225 "for path %s\n", av_get_media_type_string(AVMEDIA_TYPE_AUDIO),
230 avCodecCtx = avFormatCtx->streams[selected_stream]->codec;
231 codec = avcodec_find_decoder(avCodecCtx->codec_id);
234 AUBIO_ERR("source_avcodec: Could not find decoder for %s", s->path);
238 #if FF_API_LAVF_AVCTX
239 /* Copy codec parameters from input stream to output codec context */
240 if ((err = avcodec_parameters_to_context(avCodecCtx, codecpar)) < 0) {
241 AUBIO_ERR("source_avcodec: Failed to copy %s codec parameters to "
242 "decoder context for %s\n",
243 av_get_media_type_string(AVMEDIA_TYPE_AUDIO), s->path);
246 #if HAVE_AUBIO_LIBAVCODEC_TIMEBASE_FIX
247 // avoids 'skipped frames warning' with avecodec < 58, deprecated after
248 av_codec_set_pkt_timebase(avCodecCtx,
249 avFormatCtx->streams[selected_stream]->time_base);
253 if ( ( err = avcodec_open2(avCodecCtx, codec, NULL) ) < 0) {
255 av_strerror (err, errorstr, sizeof(errorstr));
256 AUBIO_ERR("source_avcodec: Could not load codec for %s (%s)\n", s->path,
261 /* get input specs */
262 s->input_samplerate = avCodecCtx->sample_rate;
263 s->input_channels = avCodecCtx->channels;
264 //AUBIO_DBG("input_samplerate: %d\n", s->input_samplerate);
265 //AUBIO_DBG("input_channels: %d\n", s->input_channels);
267 if (samplerate == 0) {
268 s->samplerate = s->input_samplerate;
270 s->samplerate = samplerate;
273 if (s->samplerate > s->input_samplerate) {
274 AUBIO_WRN("source_avcodec: upsampling %s from %d to %d\n", s->path,
275 s->input_samplerate, s->samplerate);
278 avFrame = av_frame_alloc();
280 AUBIO_ERR("source_avcodec: Could not allocate frame for (%s)\n", s->path);
283 /* allocate output for avr */
284 s->output = (smpl_t *)av_malloc(AUBIO_AVCODEC_MAX_BUFFER_SIZE
290 s->avFormatCtx = avFormatCtx;
291 s->avCodecCtx = avCodecCtx;
292 s->avFrame = avFrame;
294 aubio_source_avcodec_reset_resampler(s);
296 if (s->avr == NULL) goto beach;
300 //av_log_set_level(AV_LOG_QUIET);
305 //AUBIO_ERR("can not read %s at samplerate %dHz with a hop_size of %d\n",
306 // s->path, s->samplerate, s->hop_size);
307 del_aubio_source_avcodec(s);
311 void aubio_source_avcodec_reset_resampler(aubio_source_avcodec_t * s)
313 // create or reset resampler to/from mono/multi-channel
314 if ( s->avr == NULL ) {
316 int64_t input_layout = av_get_default_channel_layout(s->input_channels);
317 int64_t output_layout = av_get_default_channel_layout(s->input_channels);
318 #ifdef HAVE_AVRESAMPLE
319 AVAudioResampleContext *avr = avresample_alloc_context();
320 #elif defined(HAVE_SWRESAMPLE)
321 SwrContext *avr = swr_alloc();
322 #endif /* HAVE_AVRESAMPLE || HAVE_SWRESAMPLE */
324 av_opt_set_int(avr, "in_channel_layout", input_layout, 0);
325 av_opt_set_int(avr, "out_channel_layout", output_layout, 0);
326 av_opt_set_int(avr, "in_sample_rate", s->input_samplerate, 0);
327 av_opt_set_int(avr, "out_sample_rate", s->samplerate, 0);
328 av_opt_set_int(avr, "in_sample_fmt", s->avCodecCtx->sample_fmt, 0);
329 #if HAVE_AUBIO_DOUBLE
330 av_opt_set_int(avr, "out_sample_fmt", AV_SAMPLE_FMT_DBL, 0);
332 av_opt_set_int(avr, "out_sample_fmt", AV_SAMPLE_FMT_FLT, 0);
335 //av_opt_set_int(avr, "out_sample_fmt", AV_SAMPLE_FMT_FLTP, 0);
336 #ifdef HAVE_AVRESAMPLE
337 if ( ( err = avresample_open(avr) ) < 0)
338 #elif defined(HAVE_SWRESAMPLE)
339 if ( ( err = swr_init(avr) ) < 0)
340 #endif /* HAVE_AVRESAMPLE || HAVE_SWRESAMPLE */
343 av_strerror (err, errorstr, sizeof(errorstr));
344 AUBIO_ERR("source_avcodec: Could not open resampling context"
345 " for %s (%s)\n", s->path, errorstr);
352 void aubio_source_avcodec_readframe(aubio_source_avcodec_t *s,
353 uint_t * read_samples)
355 AVFormatContext *avFormatCtx = s->avFormatCtx;
356 AVCodecContext *avCodecCtx = s->avCodecCtx;
357 AVFrame *avFrame = s->avFrame;
358 AVPacket avPacket = s->avPacket;
359 #ifdef HAVE_AVRESAMPLE
360 AVAudioResampleContext *avr = s->avr;
361 #elif defined(HAVE_SWRESAMPLE)
362 SwrContext *avr = s->avr;
363 #endif /* HAVE_AVRESAMPLE || HAVE_SWRESAMPLE */
365 #ifdef HAVE_AVRESAMPLE
367 int in_samples = avFrame->nb_samples;
368 int out_linesize = 0;
369 int max_out_samples = AUBIO_AVCODEC_MAX_BUFFER_SIZE;
371 #elif defined(HAVE_SWRESAMPLE)
372 int in_samples = avFrame->nb_samples;
373 int max_out_samples = AUBIO_AVCODEC_MAX_BUFFER_SIZE / avCodecCtx->channels;
375 #endif /* HAVE_AVRESAMPLE || HAVE_SWRESAMPLE */
376 smpl_t *output = s->output;
377 #ifndef FF_API_LAVF_AVCTX
382 av_init_packet (&avPacket);
387 int err = av_read_frame (avFormatCtx, &avPacket);
388 if (err == AVERROR_EOF) {
394 av_strerror (err, errorstr, sizeof(errorstr));
395 AUBIO_ERR("source_avcodec: could not read frame in %s (%s)\n",
400 } while (avPacket.stream_index != s->selected_stream);
402 #if FF_API_LAVF_AVCTX
403 ret = avcodec_send_packet(avCodecCtx, &avPacket);
404 if (ret < 0 && ret != AVERROR_EOF) {
405 AUBIO_ERR("source_avcodec: error when sending packet for %s\n", s->path);
408 ret = avcodec_receive_frame(avCodecCtx, avFrame);
413 if (ret == AVERROR(EAGAIN)) {
414 //AUBIO_WRN("source_avcodec: output is not available right now - "
415 // "user must try to send new input\n");
417 } else if (ret == AVERROR_EOF) {
418 AUBIO_WRN("source_avcodec: the decoder has been fully flushed, "
419 "and there will be no more output frames\n");
421 AUBIO_ERR("source_avcodec: decoding errors on %s\n", s->path);
426 len = avcodec_decode_audio4(avCodecCtx, avFrame, &got_frame, &avPacket);
429 AUBIO_ERR("source_avcodec: error while decoding %s\n", s->path);
433 if (got_frame == 0) {
434 AUBIO_WRN("source_avcodec: did not get a frame when reading %s\n",
439 #if LIBAVUTIL_VERSION_MAJOR > 52
440 if (avFrame->channels != (sint_t)s->input_channels) {
441 AUBIO_WRN ("source_avcodec: trying to read from %d channel(s),"
442 "but configured for %d; is '%s' corrupt?\n",
443 avFrame->channels, s->input_channels, s->path);
447 #warning "avutil < 53 is deprecated, crashes might occur on corrupt files"
450 #ifdef HAVE_AVRESAMPLE
452 av_samples_get_buffer_size(&in_linesize, avCodecCtx->channels,
453 avFrame->nb_samples, avCodecCtx->sample_fmt, 1);
454 in_samples = avFrame->nb_samples;
456 max_out_samples = AUBIO_AVCODEC_MAX_BUFFER_SIZE;
457 out_samples = avresample_convert ( avr,
458 (uint8_t **)&output, out_linesize, max_out_samples,
459 (uint8_t **)avFrame->data, in_linesize, in_samples);
460 #elif defined(HAVE_SWRESAMPLE)
461 in_samples = avFrame->nb_samples;
462 max_out_samples = AUBIO_AVCODEC_MAX_BUFFER_SIZE / avCodecCtx->channels;
463 out_samples = swr_convert( avr,
464 (uint8_t **)&output, max_out_samples,
465 (const uint8_t **)avFrame->data, in_samples);
466 #endif /* HAVE_AVRESAMPLE || HAVE_SWRESAMPLE */
467 if (out_samples <= 0) {
468 AUBIO_WRN("source_avcodec: no sample found while converting frame (%s)\n",
473 *read_samples = out_samples;
476 s->avFormatCtx = avFormatCtx;
477 s->avCodecCtx = avCodecCtx;
478 s->avFrame = avFrame;
479 #if defined(HAVE_AVRESAMPLE) || defined(HAVE_SWRESAMPLE)
481 #endif /* HAVE_AVRESAMPLE || HAVE_SWRESAMPLE */
484 av_packet_unref(&avPacket);
487 void aubio_source_avcodec_do(aubio_source_avcodec_t * s, fvec_t * read_data,
491 uint_t total_wrote = 0;
492 while (total_wrote < s->hop_size) {
493 end = MIN(s->read_samples - s->read_index, s->hop_size - total_wrote);
494 for (i = 0; i < end; i++) {
495 read_data->data[i + total_wrote] = 0.;
496 for (j = 0; j < s->input_channels; j++) {
497 read_data->data[i + total_wrote] +=
498 s->output[(i + s->read_index) * s->input_channels + j];
500 read_data->data[i + total_wrote] *= 1./s->input_channels;
503 if (total_wrote < s->hop_size) {
504 uint_t avcodec_read = 0;
505 aubio_source_avcodec_readframe(s, &avcodec_read);
506 s->read_samples = avcodec_read;
512 s->read_index += end;
515 if (total_wrote < s->hop_size) {
516 for (i = total_wrote; i < s->hop_size; i++) {
517 read_data->data[i] = 0.;
523 void aubio_source_avcodec_do_multi(aubio_source_avcodec_t * s,
524 fmat_t * read_data, uint_t * read) {
527 uint_t total_wrote = 0;
528 while (total_wrote < s->hop_size) {
529 end = MIN(s->read_samples - s->read_index, s->hop_size - total_wrote);
530 for (j = 0; j < read_data->height; j++) {
531 for (i = 0; i < end; i++) {
532 read_data->data[j][i + total_wrote] =
533 s->output[(i + s->read_index) * s->input_channels + j];
537 if (total_wrote < s->hop_size) {
538 uint_t avcodec_read = 0;
539 aubio_source_avcodec_readframe(s, &avcodec_read);
540 s->read_samples = avcodec_read;
546 s->read_index += end;
549 if (total_wrote < s->hop_size) {
550 for (j = 0; j < read_data->height; j++) {
551 for (i = total_wrote; i < s->hop_size; i++) {
552 read_data->data[j][i] = 0.;
559 uint_t aubio_source_avcodec_get_samplerate(const aubio_source_avcodec_t * s) {
560 return s->samplerate;
563 uint_t aubio_source_avcodec_get_channels(const aubio_source_avcodec_t * s) {
564 return s->input_channels;
567 uint_t aubio_source_avcodec_seek (aubio_source_avcodec_t * s, uint_t pos) {
568 int64_t resampled_pos =
569 (uint_t)ROUND(pos * (s->input_samplerate * 1. / s->samplerate));
570 int64_t min_ts = MAX(resampled_pos - 2000, 0);
571 int64_t max_ts = MIN(resampled_pos + 2000, INT64_MAX);
572 int seek_flags = AVSEEK_FLAG_FRAME | AVSEEK_FLAG_ANY;
573 int ret = AUBIO_FAIL;
574 if (s->avFormatCtx != NULL && s->avr != NULL) {
577 AUBIO_ERR("source_avcodec: failed seeking in %s (file not opened?)",
581 if ((sint_t)pos < 0) {
582 AUBIO_ERR("source_avcodec: could not seek %s at %d (seeking position"
583 " should be >= 0)\n", s->path, pos);
586 ret = avformat_seek_file(s->avFormatCtx, s->selected_stream,
587 min_ts, resampled_pos, max_ts, seek_flags);
589 AUBIO_ERR("source_avcodec: failed seeking to %d in file %s",
596 #ifdef HAVE_AVRESAMPLE
597 // reset the AVAudioResampleContext
598 avresample_close(s->avr);
599 avresample_open(s->avr);
600 #elif defined(HAVE_SWRESAMPLE)
607 uint_t aubio_source_avcodec_get_duration (aubio_source_avcodec_t * s) {
608 if (s && &(s->avFormatCtx) != NULL) {
609 int64_t duration = s->avFormatCtx->duration;
610 return s->samplerate * ((uint_t)duration / 1e6 );
615 uint_t aubio_source_avcodec_close(aubio_source_avcodec_t * s) {
616 if (s->avr != NULL) {
617 #ifdef HAVE_AVRESAMPLE
618 avresample_close( s->avr );
619 #elif defined(HAVE_SWRESAMPLE)
620 swr_close ( s->avr );
625 if (s->avCodecCtx != NULL) {
626 #ifndef HAVE_AUBIO_LIBAVCODEC_DEPRECATED
627 avcodec_free_context( &s->avCodecCtx );
629 avcodec_close ( s->avCodecCtx );
632 s->avCodecCtx = NULL;
633 if (s->avFormatCtx != NULL) {
634 avformat_close_input(&s->avFormatCtx);
635 s->avFormatCtx = NULL;
637 av_packet_unref(&s->avPacket);
641 void del_aubio_source_avcodec(aubio_source_avcodec_t * s){
643 aubio_source_avcodec_close(s);
644 if (s->output != NULL) {
648 if (s->avFrame != NULL) {
649 av_frame_free( &(s->avFrame) );
659 #endif /* HAVE_LIBAV */