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 through 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(55,28,1)
46 #warning "libavcodec55 is deprecated"
47 #define HAVE_AUBIO_LIBAVCODEC_DEPRECATED 1
48 #define av_frame_alloc avcodec_alloc_frame
49 #define av_frame_free avcodec_free_frame
50 #define av_packet_unref av_free_packet
53 #include "aubio_priv.h"
56 #include "source_avcodec.h"
58 #define AUBIO_AVCODEC_MAX_BUFFER_SIZE FF_MIN_BUFFER_SIZE
60 struct _aubio_source_avcodec_t {
65 // some data about the file
67 uint_t input_samplerate;
68 uint_t input_channels;
71 AVFormatContext *avFormatCtx;
72 AVCodecContext *avCodecCtx;
75 #ifdef HAVE_AVRESAMPLE
76 AVAudioResampleContext *avr;
77 #elif defined(HAVE_SWRESAMPLE)
83 sint_t selected_stream;
88 // hack to create or re-create the context the first time _do or _do_multi is called
89 void aubio_source_avcodec_reset_resampler(aubio_source_avcodec_t * s, uint_t multi);
90 void aubio_source_avcodec_readframe(aubio_source_avcodec_t *s, uint_t * read_samples);
92 uint_t aubio_source_avcodec_has_network_url(aubio_source_avcodec_t *s);
94 uint_t aubio_source_avcodec_has_network_url(aubio_source_avcodec_t *s) {
95 char proto[20], authorization[256], hostname[128], uripath[256];
96 int proto_size = 20, authorization_size = 256, hostname_size = 128,
97 *port_ptr = 0, path_size = 256;
98 av_url_split(proto, proto_size, authorization, authorization_size, hostname,
99 hostname_size, port_ptr, uripath, path_size, s->path);
107 aubio_source_avcodec_t * new_aubio_source_avcodec(const char_t * path, uint_t samplerate, uint_t hop_size) {
108 aubio_source_avcodec_t * s = AUBIO_NEW(aubio_source_avcodec_t);
111 AUBIO_ERR("source_avcodec: Aborted opening null path\n");
114 if ((sint_t)samplerate < 0) {
115 AUBIO_ERR("source_avcodec: Can not open %s with samplerate %d\n", path, samplerate);
118 if ((sint_t)hop_size <= 0) {
119 AUBIO_ERR("source_avcodec: Can not open %s with hop_size %d\n", path, hop_size);
123 s->hop_size = hop_size;
126 if (s->path) AUBIO_FREE(s->path);
127 s->path = AUBIO_ARRAY(char_t, strnlen(path, PATH_MAX) + 1);
128 strncpy(s->path, path, strnlen(path, PATH_MAX) + 1);
130 // register all formats and codecs
133 if (aubio_source_avcodec_has_network_url(s)) {
134 avformat_network_init();
137 // try opening the file and get some info about it
138 AVFormatContext *avFormatCtx = s->avFormatCtx;
140 if ( (err = avformat_open_input(&avFormatCtx, s->path, NULL, NULL) ) < 0 ) {
142 av_strerror (err, errorstr, sizeof(errorstr));
143 AUBIO_ERR("source_avcodec: Failed opening %s (%s)\n", s->path, errorstr);
147 // try to make sure max_analyze_duration is big enough for most songs
148 #if FFMPEG_LIBAVFORMAT_MAX_DUR2
149 avFormatCtx->max_analyze_duration2 *= 100;
151 avFormatCtx->max_analyze_duration *= 100;
154 // retrieve stream information
155 if ( (err = avformat_find_stream_info(avFormatCtx, NULL)) < 0 ) {
157 av_strerror (err, errorstr, sizeof(errorstr));
158 AUBIO_ERR("source_avcodec: Could not find stream information " "for %s (%s)\n", s->path,
163 // dump information about file onto standard error
164 //av_dump_format(avFormatCtx, 0, s->path, 0);
166 // look for the first audio stream
168 sint_t selected_stream = -1;
169 for (i = 0; i < avFormatCtx->nb_streams; i++) {
170 #if FF_API_LAVF_AVCTX
171 if (avFormatCtx->streams[i]->codecpar->codec_type == AVMEDIA_TYPE_AUDIO) {
173 if (avFormatCtx->streams[i]->codec->codec_type == AVMEDIA_TYPE_AUDIO) {
175 if (selected_stream == -1) {
178 AUBIO_WRN("source_avcodec: More than one audio stream in %s, "
179 "taking the first one\n", s->path);
183 if (selected_stream == -1) {
184 AUBIO_ERR("source_avcodec: No audio stream in %s\n", s->path);
187 //AUBIO_DBG("Taking stream %d in file %s\n", selected_stream, s->path);
188 s->selected_stream = selected_stream;
190 AVCodecContext *avCodecCtx = s->avCodecCtx;
191 #if FF_API_LAVF_AVCTX
192 AVCodecParameters *codecpar = avFormatCtx->streams[selected_stream]->codecpar;
193 if (codecpar == NULL) {
194 AUBIO_ERR("source_avcodec: Could not find decoder for %s", s->path);
197 AVCodec *codec = avcodec_find_decoder(codecpar->codec_id);
199 /* Allocate a codec context for the decoder */
200 avCodecCtx = avcodec_alloc_context3(codec);
202 AUBIO_ERR("source_avcodec: Failed to allocate the %s codec context for path %s\n",
203 av_get_media_type_string(AVMEDIA_TYPE_AUDIO), s->path);
207 avCodecCtx = avFormatCtx->streams[selected_stream]->codec;
208 AVCodec *codec = avcodec_find_decoder(avCodecCtx->codec_id);
211 AUBIO_ERR("source_avcodec: Could not find decoder for %s", s->path);
215 #if FF_API_LAVF_AVCTX
216 /* Copy codec parameters from input stream to output codec context */
217 if ((err = avcodec_parameters_to_context(avCodecCtx, codecpar)) < 0) {
218 AUBIO_ERR("source_avcodec: Failed to copy %s codec parameters to decoder context for %s\n",
219 av_get_media_type_string(AVMEDIA_TYPE_AUDIO), s->path);
224 if ( ( err = avcodec_open2(avCodecCtx, codec, NULL) ) < 0) {
226 av_strerror (err, errorstr, sizeof(errorstr));
227 AUBIO_ERR("source_avcodec: Could not load codec for %s (%s)\n", s->path, errorstr);
231 /* get input specs */
232 s->input_samplerate = avCodecCtx->sample_rate;
233 s->input_channels = avCodecCtx->channels;
234 //AUBIO_DBG("input_samplerate: %d\n", s->input_samplerate);
235 //AUBIO_DBG("input_channels: %d\n", s->input_channels);
237 if (samplerate == 0) {
238 s->samplerate = s->input_samplerate;
240 s->samplerate = samplerate;
243 if (s->samplerate > s->input_samplerate) {
244 AUBIO_WRN("source_avcodec: upsampling %s from %d to %d\n", s->path,
245 s->input_samplerate, s->samplerate);
248 AVFrame *avFrame = s->avFrame;
249 avFrame = av_frame_alloc();
251 AUBIO_ERR("source_avcodec: Could not allocate frame for (%s)\n", s->path);
254 /* allocate output for avr */
255 s->output = (smpl_t *)av_malloc(AUBIO_AVCODEC_MAX_BUFFER_SIZE * sizeof(smpl_t));
260 s->avFormatCtx = avFormatCtx;
261 s->avCodecCtx = avCodecCtx;
262 s->avFrame = avFrame;
264 // default to mono output
265 aubio_source_avcodec_reset_resampler(s, 0);
270 //av_log_set_level(AV_LOG_QUIET);
275 //AUBIO_ERR("can not read %s at samplerate %dHz with a hop_size of %d\n",
276 // s->path, s->samplerate, s->hop_size);
277 del_aubio_source_avcodec(s);
281 void aubio_source_avcodec_reset_resampler(aubio_source_avcodec_t * s, uint_t multi) {
282 // create or reset resampler to/from mono/multi-channel
283 if ( (multi != s->multi) || (s->avr == NULL) ) {
284 int64_t input_layout = av_get_default_channel_layout(s->input_channels);
285 uint_t output_channels = multi ? s->input_channels : 1;
286 int64_t output_layout = av_get_default_channel_layout(output_channels);
287 #ifdef HAVE_AVRESAMPLE
288 AVAudioResampleContext *avr = avresample_alloc_context();
289 AVAudioResampleContext *oldavr = s->avr;
290 #elif defined(HAVE_SWRESAMPLE)
291 SwrContext *avr = swr_alloc();
292 SwrContext *oldavr = s->avr;
293 #endif /* HAVE_AVRESAMPLE || HAVE_SWRESAMPLE */
295 av_opt_set_int(avr, "in_channel_layout", input_layout, 0);
296 av_opt_set_int(avr, "out_channel_layout", output_layout, 0);
297 av_opt_set_int(avr, "in_sample_rate", s->input_samplerate, 0);
298 av_opt_set_int(avr, "out_sample_rate", s->samplerate, 0);
299 av_opt_set_int(avr, "in_sample_fmt", s->avCodecCtx->sample_fmt, 0);
300 #if HAVE_AUBIO_DOUBLE
301 av_opt_set_int(avr, "out_sample_fmt", AV_SAMPLE_FMT_DBL, 0);
303 av_opt_set_int(avr, "out_sample_fmt", AV_SAMPLE_FMT_FLT, 0);
306 //av_opt_set_int(avr, "out_sample_fmt", AV_SAMPLE_FMT_FLTP, 0);
308 #ifdef HAVE_AVRESAMPLE
309 if ( ( err = avresample_open(avr) ) < 0) {
310 #elif defined(HAVE_SWRESAMPLE)
311 if ( ( err = swr_init(avr) ) < 0) {
312 #endif /* HAVE_AVRESAMPLE || HAVE_SWRESAMPLE */
314 av_strerror (err, errorstr, sizeof(errorstr));
315 AUBIO_ERR("source_avcodec: Could not open AVAudioResampleContext for %s (%s)\n",
321 if (oldavr != NULL) {
322 #ifdef HAVE_AVRESAMPLE
323 avresample_close( oldavr );
324 #elif defined(HAVE_SWRESAMPLE)
325 swr_close ( oldavr );
326 #endif /* HAVE_AVRESAMPLE || HAVE_SWRESAMPLE */
334 void aubio_source_avcodec_readframe(aubio_source_avcodec_t *s, uint_t * read_samples) {
335 AVFormatContext *avFormatCtx = s->avFormatCtx;
336 AVCodecContext *avCodecCtx = s->avCodecCtx;
337 AVFrame *avFrame = s->avFrame;
338 AVPacket avPacket = s->avPacket;
339 av_init_packet (&avPacket);
340 #ifdef HAVE_AVRESAMPLE
341 AVAudioResampleContext *avr = s->avr;
342 #elif defined(HAVE_SWRESAMPLE)
343 SwrContext *avr = s->avr;
344 #endif /* HAVE_AVRESAMPLE || HAVE_SWRESAMPLE */
345 smpl_t *output = s->output;
350 int err = av_read_frame (avFormatCtx, &avPacket);
351 if (err == AVERROR_EOF) {
357 av_strerror (err, errorstr, sizeof(errorstr));
358 AUBIO_ERR("source_avcodec: could not read frame in %s (%s)\n", s->path, errorstr);
362 } while (avPacket.stream_index != s->selected_stream);
365 #if FF_API_LAVF_AVCTX
366 int ret = avcodec_send_packet(avCodecCtx, &avPacket);
367 if (ret < 0 && ret != AVERROR_EOF) {
368 AUBIO_ERR("source_avcodec: error when sending packet for %s\n", s->path);
371 ret = avcodec_receive_frame(avCodecCtx, avFrame);
376 if (ret == AVERROR(EAGAIN)) {
377 //AUBIO_WRN("source_avcodec: output is not available right now - user must try to send new input\n");
379 } else if (ret == AVERROR_EOF) {
380 AUBIO_WRN("source_avcodec: the decoder has been fully flushed, and there will be no more output frames\n");
382 AUBIO_ERR("source_avcodec: decoding errors on %s\n", s->path);
387 int len = avcodec_decode_audio4(avCodecCtx, avFrame, &got_frame, &avPacket);
390 AUBIO_ERR("source_avcodec: error while decoding %s\n", s->path);
394 if (got_frame == 0) {
395 AUBIO_WRN("source_avcodec: did not get a frame when reading %s\n", s->path);
399 #ifdef HAVE_AVRESAMPLE
401 av_samples_get_buffer_size(&in_linesize, avCodecCtx->channels,
402 avFrame->nb_samples, avCodecCtx->sample_fmt, 1);
403 int in_samples = avFrame->nb_samples;
404 int out_linesize = 0;
405 int max_out_samples = AUBIO_AVCODEC_MAX_BUFFER_SIZE;
406 int out_samples = avresample_convert ( avr,
407 (uint8_t **)&output, out_linesize, max_out_samples,
408 (uint8_t **)avFrame->data, in_linesize, in_samples);
409 #elif defined(HAVE_SWRESAMPLE)
410 int in_samples = avFrame->nb_samples;
411 int max_out_samples = AUBIO_AVCODEC_MAX_BUFFER_SIZE / avCodecCtx->channels;
412 int out_samples = swr_convert( avr,
413 (uint8_t **)&output, max_out_samples,
414 (const uint8_t **)avFrame->data, in_samples);
415 #endif /* HAVE_AVRESAMPLE || HAVE_SWRESAMPLE */
416 if (out_samples <= 0) {
417 AUBIO_WRN("source_avcodec: no sample found while converting frame (%s)\n", s->path);
421 *read_samples = out_samples;
424 s->avFormatCtx = avFormatCtx;
425 s->avCodecCtx = avCodecCtx;
426 s->avFrame = avFrame;
427 #if defined(HAVE_AVRESAMPLE) || defined(HAVE_SWRESAMPLE)
429 #endif /* HAVE_AVRESAMPLE || HAVE_SWRESAMPLE */
432 av_packet_unref(&avPacket);
435 void aubio_source_avcodec_do(aubio_source_avcodec_t * s, fvec_t * read_data, uint_t * read){
438 uint_t total_wrote = 0;
440 if (s->multi == 1) aubio_source_avcodec_reset_resampler(s, 0);
441 while (total_wrote < s->hop_size) {
442 end = MIN(s->read_samples - s->read_index, s->hop_size - total_wrote);
443 for (i = 0; i < end; i++) {
444 read_data->data[i + total_wrote] = s->output[i + s->read_index];
447 if (total_wrote < s->hop_size) {
448 uint_t avcodec_read = 0;
449 aubio_source_avcodec_readframe(s, &avcodec_read);
450 s->read_samples = avcodec_read;
456 s->read_index += end;
459 if (total_wrote < s->hop_size) {
460 for (i = total_wrote; i < s->hop_size; i++) {
461 read_data->data[i] = 0.;
467 void aubio_source_avcodec_do_multi(aubio_source_avcodec_t * s, fmat_t * read_data, uint_t * read){
470 uint_t total_wrote = 0;
472 if (s->multi == 0) aubio_source_avcodec_reset_resampler(s, 1);
473 while (total_wrote < s->hop_size) {
474 end = MIN(s->read_samples - s->read_index, s->hop_size - total_wrote);
475 for (j = 0; j < read_data->height; j++) {
476 for (i = 0; i < end; i++) {
477 read_data->data[j][i + total_wrote] =
478 s->output[(i + s->read_index) * s->input_channels + j];
482 if (total_wrote < s->hop_size) {
483 uint_t avcodec_read = 0;
484 aubio_source_avcodec_readframe(s, &avcodec_read);
485 s->read_samples = avcodec_read;
491 s->read_index += end;
494 if (total_wrote < s->hop_size) {
495 for (j = 0; j < read_data->height; j++) {
496 for (i = total_wrote; i < s->hop_size; i++) {
497 read_data->data[j][i] = 0.;
504 uint_t aubio_source_avcodec_get_samplerate(const aubio_source_avcodec_t * s) {
505 return s->samplerate;
508 uint_t aubio_source_avcodec_get_channels(const aubio_source_avcodec_t * s) {
509 return s->input_channels;
512 uint_t aubio_source_avcodec_seek (aubio_source_avcodec_t * s, uint_t pos) {
513 int64_t resampled_pos = (uint_t)ROUND(pos * (s->input_samplerate * 1. / s->samplerate));
514 int64_t min_ts = MAX(resampled_pos - 2000, 0);
515 int64_t max_ts = MIN(resampled_pos + 2000, INT64_MAX);
516 int seek_flags = AVSEEK_FLAG_FRAME | AVSEEK_FLAG_ANY;
517 int ret = AUBIO_FAIL;
518 if (s->avFormatCtx != NULL && s->avr != NULL) {
521 AUBIO_ERR("source_avcodec: failed seeking in %s (file not opened?)", s->path);
524 if ((sint_t)pos < 0) {
525 AUBIO_ERR("source_avcodec: could not seek %s at %d (seeking position"
526 " should be >= 0)\n", s->path, pos);
529 ret = avformat_seek_file(s->avFormatCtx, s->selected_stream,
530 min_ts, resampled_pos, max_ts, seek_flags);
532 AUBIO_ERR("source_avcodec: failed seeking to %d in file %s", pos, s->path);
538 #ifdef HAVE_AVRESAMPLE
539 // reset the AVAudioResampleContext
540 avresample_close(s->avr);
541 avresample_open(s->avr);
542 #elif defined(HAVE_SWRESAMPLE)
549 uint_t aubio_source_avcodec_get_duration (aubio_source_avcodec_t * s) {
550 if (s && &(s->avFormatCtx) != NULL) {
551 int64_t duration = s->avFormatCtx->duration;
552 return s->samplerate * ((uint_t)duration / 1e6 );
557 uint_t aubio_source_avcodec_close(aubio_source_avcodec_t * s) {
558 if (s->avr != NULL) {
559 #ifdef HAVE_AVRESAMPLE
560 avresample_close( s->avr );
561 #elif defined(HAVE_SWRESAMPLE)
562 swr_close ( s->avr );
567 if (s->avCodecCtx != NULL) {
568 #ifndef HAVE_AUBIO_LIBAVCODEC_DEPRECATED
569 avcodec_free_context( &s->avCodecCtx );
571 avcodec_close ( s->avCodecCtx );
574 s->avCodecCtx = NULL;
575 if (s->avFormatCtx != NULL) {
576 avformat_close_input(&s->avFormatCtx);
577 #ifndef HAVE_AUBIO_LIBAVCODEC_DEPRECATED // avoid crash on old libavcodec54
578 avformat_free_context(s->avFormatCtx);
580 s->avFormatCtx = NULL;
582 av_packet_unref(&s->avPacket);
586 void del_aubio_source_avcodec(aubio_source_avcodec_t * s){
588 aubio_source_avcodec_close(s);
589 if (s->output != NULL) {
593 if (s->avFrame != NULL) {
594 av_frame_free( &(s->avFrame) );
604 #endif /* HAVE_LIBAV */