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(55,28,1)
50 #warning "libavcodec < 56 is deprecated"
51 #define av_frame_alloc avcodec_alloc_frame
52 #define av_frame_free avcodec_free_frame
53 #define av_packet_unref av_free_packet
56 #include "aubio_priv.h"
59 #include "source_avcodec.h"
61 #if LIBAVCODEC_VERSION_INT < AV_VERSION_INT(56, 56, 0)
62 #define AUBIO_AVCODEC_MAX_BUFFER_SIZE FF_MIN_BUFFER_SIZE
64 #define AUBIO_AVCODEC_MAX_BUFFER_SIZE AV_INPUT_BUFFER_MIN_SIZE
67 struct _aubio_source_avcodec_t {
72 // some data about the file
74 uint_t input_samplerate;
75 uint_t input_channels;
78 AVFormatContext *avFormatCtx;
79 AVCodecContext *avCodecCtx;
82 #ifdef HAVE_AVRESAMPLE
83 AVAudioResampleContext *avr;
84 #elif defined(HAVE_SWRESAMPLE)
90 sint_t selected_stream;
94 // create or re-create the context when _do or _do_multi is called
95 void aubio_source_avcodec_reset_resampler(aubio_source_avcodec_t * s);
96 // actually read a frame
97 void aubio_source_avcodec_readframe(aubio_source_avcodec_t *s,
98 uint_t * read_samples);
100 uint_t aubio_source_avcodec_has_network_url(aubio_source_avcodec_t *s);
102 uint_t aubio_source_avcodec_has_network_url(aubio_source_avcodec_t *s) {
103 char proto[20], authorization[256], hostname[128], uripath[256];
104 int proto_size = 20, authorization_size = 256, hostname_size = 128,
105 *port_ptr = 0, path_size = 256;
106 av_url_split(proto, proto_size, authorization, authorization_size, hostname,
107 hostname_size, port_ptr, uripath, path_size, s->path);
115 aubio_source_avcodec_t * new_aubio_source_avcodec(const char_t * path,
116 uint_t samplerate, uint_t hop_size) {
117 aubio_source_avcodec_t * s = AUBIO_NEW(aubio_source_avcodec_t);
118 AVFormatContext *avFormatCtx = s->avFormatCtx;
119 AVCodecContext *avCodecCtx = s->avCodecCtx;
120 AVFrame *avFrame = s->avFrame;
121 sint_t selected_stream = -1;
122 #if FF_API_LAVF_AVCTX
123 AVCodecParameters *codecpar;
129 AUBIO_ERR("source_avcodec: Aborted opening null path\n");
132 if ((sint_t)samplerate < 0) {
133 AUBIO_ERR("source_avcodec: Can not open %s with samplerate %d\n",
137 if ((sint_t)hop_size <= 0) {
138 AUBIO_ERR("source_avcodec: Can not open %s with hop_size %d\n",
143 s->hop_size = hop_size;
146 if (s->path) AUBIO_FREE(s->path);
147 s->path = AUBIO_ARRAY(char_t, strnlen(path, PATH_MAX) + 1);
148 strncpy(s->path, path, strnlen(path, PATH_MAX) + 1);
150 #if LIBAVFORMAT_VERSION_INT < AV_VERSION_INT(58,0,0)
151 // register all formats and codecs
155 if (aubio_source_avcodec_has_network_url(s)) {
156 avformat_network_init();
159 // try opening the file and get some info about it
161 if ( (err = avformat_open_input(&avFormatCtx, s->path, NULL, NULL) ) < 0 ) {
163 av_strerror (err, errorstr, sizeof(errorstr));
164 AUBIO_ERR("source_avcodec: Failed opening %s (%s)\n", s->path, errorstr);
168 // try to make sure max_analyze_duration is big enough for most songs
169 #if FFMPEG_LIBAVFORMAT_MAX_DUR2
170 avFormatCtx->max_analyze_duration2 *= 100;
172 avFormatCtx->max_analyze_duration *= 100;
175 // retrieve stream information
176 if ( (err = avformat_find_stream_info(avFormatCtx, NULL)) < 0 ) {
178 av_strerror (err, errorstr, sizeof(errorstr));
179 AUBIO_ERR("source_avcodec: Could not find stream information "
180 "for %s (%s)\n", s->path, errorstr);
184 // dump information about file onto standard error
185 //av_dump_format(avFormatCtx, 0, s->path, 0);
187 // look for the first audio stream
188 for (i = 0; i < avFormatCtx->nb_streams; i++) {
189 #if FF_API_LAVF_AVCTX
190 if (avFormatCtx->streams[i]->codecpar->codec_type == AVMEDIA_TYPE_AUDIO) {
192 if (avFormatCtx->streams[i]->codec->codec_type == AVMEDIA_TYPE_AUDIO) {
194 if (selected_stream == -1) {
197 AUBIO_WRN("source_avcodec: More than one audio stream in %s, "
198 "taking the first one\n", s->path);
202 if (selected_stream == -1) {
203 AUBIO_ERR("source_avcodec: No audio stream in %s\n", s->path);
206 //AUBIO_DBG("Taking stream %d in file %s\n", selected_stream, s->path);
207 s->selected_stream = selected_stream;
209 #if FF_API_LAVF_AVCTX
210 codecpar = avFormatCtx->streams[selected_stream]->codecpar;
211 if (codecpar == NULL) {
212 AUBIO_ERR("source_avcodec: Could not find decoder for %s", s->path);
215 codec = avcodec_find_decoder(codecpar->codec_id);
217 /* Allocate a codec context for the decoder */
218 avCodecCtx = avcodec_alloc_context3(codec);
220 AUBIO_ERR("source_avcodec: Failed to allocate the %s codec context "
221 "for path %s\n", av_get_media_type_string(AVMEDIA_TYPE_AUDIO),
226 avCodecCtx = avFormatCtx->streams[selected_stream]->codec;
227 codec = avcodec_find_decoder(avCodecCtx->codec_id);
230 AUBIO_ERR("source_avcodec: Could not find decoder for %s", s->path);
234 #if FF_API_LAVF_AVCTX
235 /* Copy codec parameters from input stream to output codec context */
236 if ((err = avcodec_parameters_to_context(avCodecCtx, codecpar)) < 0) {
237 AUBIO_ERR("source_avcodec: Failed to copy %s codec parameters to "
238 "decoder context for %s\n",
239 av_get_media_type_string(AVMEDIA_TYPE_AUDIO), s->path);
242 #if FFMPEG_LIBAVFORMAT // deprecated with ffmpeg, required with libav
243 av_codec_set_pkt_timebase(avCodecCtx,
244 avFormatCtx->streams[selected_stream]->time_base);
248 if ( ( err = avcodec_open2(avCodecCtx, codec, NULL) ) < 0) {
250 av_strerror (err, errorstr, sizeof(errorstr));
251 AUBIO_ERR("source_avcodec: Could not load codec for %s (%s)\n", s->path,
256 /* get input specs */
257 s->input_samplerate = avCodecCtx->sample_rate;
258 s->input_channels = avCodecCtx->channels;
259 //AUBIO_DBG("input_samplerate: %d\n", s->input_samplerate);
260 //AUBIO_DBG("input_channels: %d\n", s->input_channels);
262 if (samplerate == 0) {
263 s->samplerate = s->input_samplerate;
265 s->samplerate = samplerate;
268 if (s->samplerate > s->input_samplerate) {
269 AUBIO_WRN("source_avcodec: upsampling %s from %d to %d\n", s->path,
270 s->input_samplerate, s->samplerate);
273 avFrame = av_frame_alloc();
275 AUBIO_ERR("source_avcodec: Could not allocate frame for (%s)\n", s->path);
278 /* allocate output for avr */
279 s->output = (smpl_t *)av_malloc(AUBIO_AVCODEC_MAX_BUFFER_SIZE
285 s->avFormatCtx = avFormatCtx;
286 s->avCodecCtx = avCodecCtx;
287 s->avFrame = avFrame;
289 aubio_source_avcodec_reset_resampler(s);
291 if (s->avr == NULL) goto beach;
295 //av_log_set_level(AV_LOG_QUIET);
300 //AUBIO_ERR("can not read %s at samplerate %dHz with a hop_size of %d\n",
301 // s->path, s->samplerate, s->hop_size);
302 del_aubio_source_avcodec(s);
306 void aubio_source_avcodec_reset_resampler(aubio_source_avcodec_t * s)
308 // create or reset resampler to/from mono/multi-channel
309 if ( s->avr == NULL ) {
311 int64_t input_layout = av_get_default_channel_layout(s->input_channels);
312 int64_t output_layout = av_get_default_channel_layout(s->input_channels);
313 #ifdef HAVE_AVRESAMPLE
314 AVAudioResampleContext *avr = avresample_alloc_context();
315 #elif defined(HAVE_SWRESAMPLE)
316 SwrContext *avr = swr_alloc();
317 #endif /* HAVE_AVRESAMPLE || HAVE_SWRESAMPLE */
319 av_opt_set_int(avr, "in_channel_layout", input_layout, 0);
320 av_opt_set_int(avr, "out_channel_layout", output_layout, 0);
321 av_opt_set_int(avr, "in_sample_rate", s->input_samplerate, 0);
322 av_opt_set_int(avr, "out_sample_rate", s->samplerate, 0);
323 av_opt_set_int(avr, "in_sample_fmt", s->avCodecCtx->sample_fmt, 0);
324 #if HAVE_AUBIO_DOUBLE
325 av_opt_set_int(avr, "out_sample_fmt", AV_SAMPLE_FMT_DBL, 0);
327 av_opt_set_int(avr, "out_sample_fmt", AV_SAMPLE_FMT_FLT, 0);
330 //av_opt_set_int(avr, "out_sample_fmt", AV_SAMPLE_FMT_FLTP, 0);
331 #ifdef HAVE_AVRESAMPLE
332 if ( ( err = avresample_open(avr) ) < 0)
333 #elif defined(HAVE_SWRESAMPLE)
334 if ( ( err = swr_init(avr) ) < 0)
335 #endif /* HAVE_AVRESAMPLE || HAVE_SWRESAMPLE */
338 av_strerror (err, errorstr, sizeof(errorstr));
339 AUBIO_ERR("source_avcodec: Could not open resampling context"
340 " for %s (%s)\n", s->path, errorstr);
347 void aubio_source_avcodec_readframe(aubio_source_avcodec_t *s,
348 uint_t * read_samples)
350 AVFormatContext *avFormatCtx = s->avFormatCtx;
351 AVCodecContext *avCodecCtx = s->avCodecCtx;
352 AVFrame *avFrame = s->avFrame;
353 AVPacket avPacket = s->avPacket;
354 #ifdef HAVE_AVRESAMPLE
355 AVAudioResampleContext *avr = s->avr;
356 #elif defined(HAVE_SWRESAMPLE)
357 SwrContext *avr = s->avr;
358 #endif /* HAVE_AVRESAMPLE || HAVE_SWRESAMPLE */
360 #ifdef HAVE_AVRESAMPLE
362 int in_samples = avFrame->nb_samples;
363 int out_linesize = 0;
364 int max_out_samples = AUBIO_AVCODEC_MAX_BUFFER_SIZE;
366 #elif defined(HAVE_SWRESAMPLE)
367 int in_samples = avFrame->nb_samples;
368 int max_out_samples = AUBIO_AVCODEC_MAX_BUFFER_SIZE / avCodecCtx->channels;
370 #endif /* HAVE_AVRESAMPLE || HAVE_SWRESAMPLE */
371 smpl_t *output = s->output;
372 #ifndef FF_API_LAVF_AVCTX
377 av_init_packet (&avPacket);
382 int err = av_read_frame (avFormatCtx, &avPacket);
383 if (err == AVERROR_EOF) {
389 av_strerror (err, errorstr, sizeof(errorstr));
390 AUBIO_ERR("source_avcodec: could not read frame in %s (%s)\n",
395 } while (avPacket.stream_index != s->selected_stream);
397 #if FF_API_LAVF_AVCTX
398 ret = avcodec_send_packet(avCodecCtx, &avPacket);
399 if (ret < 0 && ret != AVERROR_EOF) {
400 AUBIO_ERR("source_avcodec: error when sending packet for %s\n", s->path);
403 ret = avcodec_receive_frame(avCodecCtx, avFrame);
408 if (ret == AVERROR(EAGAIN)) {
409 //AUBIO_WRN("source_avcodec: output is not available right now - "
410 // "user must try to send new input\n");
412 } else if (ret == AVERROR_EOF) {
413 AUBIO_WRN("source_avcodec: the decoder has been fully flushed, "
414 "and there will be no more output frames\n");
416 AUBIO_ERR("source_avcodec: decoding errors on %s\n", s->path);
421 len = avcodec_decode_audio4(avCodecCtx, avFrame, &got_frame, &avPacket);
424 AUBIO_ERR("source_avcodec: error while decoding %s\n", s->path);
428 if (got_frame == 0) {
429 AUBIO_WRN("source_avcodec: did not get a frame when reading %s\n",
434 #if LIBAVUTIL_VERSION_MAJOR > 52
435 if (avFrame->channels != (sint_t)s->input_channels) {
436 AUBIO_WRN ("source_avcodec: trying to read from %d channel(s),"
437 "but configured for %d; is '%s' corrupt?\n",
438 avFrame->channels, s->input_channels, s->path);
442 #warning "avutil < 53 is deprecated, crashes might occur on corrupt files"
445 #ifdef HAVE_AVRESAMPLE
447 av_samples_get_buffer_size(&in_linesize, avCodecCtx->channels,
448 avFrame->nb_samples, avCodecCtx->sample_fmt, 1);
449 in_samples = avFrame->nb_samples;
451 max_out_samples = AUBIO_AVCODEC_MAX_BUFFER_SIZE;
452 out_samples = avresample_convert ( avr,
453 (uint8_t **)&output, out_linesize, max_out_samples,
454 (uint8_t **)avFrame->data, in_linesize, in_samples);
455 #elif defined(HAVE_SWRESAMPLE)
456 in_samples = avFrame->nb_samples;
457 max_out_samples = AUBIO_AVCODEC_MAX_BUFFER_SIZE / avCodecCtx->channels;
458 out_samples = swr_convert( avr,
459 (uint8_t **)&output, max_out_samples,
460 (const uint8_t **)avFrame->data, in_samples);
461 #endif /* HAVE_AVRESAMPLE || HAVE_SWRESAMPLE */
462 if (out_samples <= 0) {
463 AUBIO_WRN("source_avcodec: no sample found while converting frame (%s)\n",
468 *read_samples = out_samples;
471 s->avFormatCtx = avFormatCtx;
472 s->avCodecCtx = avCodecCtx;
473 s->avFrame = avFrame;
474 #if defined(HAVE_AVRESAMPLE) || defined(HAVE_SWRESAMPLE)
476 #endif /* HAVE_AVRESAMPLE || HAVE_SWRESAMPLE */
479 av_packet_unref(&avPacket);
482 void aubio_source_avcodec_do(aubio_source_avcodec_t * s, fvec_t * read_data,
486 uint_t total_wrote = 0;
487 while (total_wrote < s->hop_size) {
488 end = MIN(s->read_samples - s->read_index, s->hop_size - total_wrote);
489 for (i = 0; i < end; i++) {
490 read_data->data[i + total_wrote] = 0.;
491 for (j = 0; j < s->input_channels; j++) {
492 read_data->data[i + total_wrote] +=
493 s->output[(i + s->read_index) * s->input_channels + j];
495 read_data->data[i + total_wrote] *= 1./s->input_channels;
498 if (total_wrote < s->hop_size) {
499 uint_t avcodec_read = 0;
500 aubio_source_avcodec_readframe(s, &avcodec_read);
501 s->read_samples = avcodec_read;
507 s->read_index += end;
510 if (total_wrote < s->hop_size) {
511 for (i = total_wrote; i < s->hop_size; i++) {
512 read_data->data[i] = 0.;
518 void aubio_source_avcodec_do_multi(aubio_source_avcodec_t * s,
519 fmat_t * read_data, uint_t * read) {
522 uint_t total_wrote = 0;
523 while (total_wrote < s->hop_size) {
524 end = MIN(s->read_samples - s->read_index, s->hop_size - total_wrote);
525 for (j = 0; j < read_data->height; j++) {
526 for (i = 0; i < end; i++) {
527 read_data->data[j][i + total_wrote] =
528 s->output[(i + s->read_index) * s->input_channels + j];
532 if (total_wrote < s->hop_size) {
533 uint_t avcodec_read = 0;
534 aubio_source_avcodec_readframe(s, &avcodec_read);
535 s->read_samples = avcodec_read;
541 s->read_index += end;
544 if (total_wrote < s->hop_size) {
545 for (j = 0; j < read_data->height; j++) {
546 for (i = total_wrote; i < s->hop_size; i++) {
547 read_data->data[j][i] = 0.;
554 uint_t aubio_source_avcodec_get_samplerate(const aubio_source_avcodec_t * s) {
555 return s->samplerate;
558 uint_t aubio_source_avcodec_get_channels(const aubio_source_avcodec_t * s) {
559 return s->input_channels;
562 uint_t aubio_source_avcodec_seek (aubio_source_avcodec_t * s, uint_t pos) {
563 int64_t resampled_pos =
564 (uint_t)ROUND(pos * (s->input_samplerate * 1. / s->samplerate));
565 int64_t min_ts = MAX(resampled_pos - 2000, 0);
566 int64_t max_ts = MIN(resampled_pos + 2000, INT64_MAX);
567 int seek_flags = AVSEEK_FLAG_FRAME | AVSEEK_FLAG_ANY;
568 int ret = AUBIO_FAIL;
569 if (s->avFormatCtx != NULL && s->avr != NULL) {
572 AUBIO_ERR("source_avcodec: failed seeking in %s (file not opened?)",
576 if ((sint_t)pos < 0) {
577 AUBIO_ERR("source_avcodec: could not seek %s at %d (seeking position"
578 " should be >= 0)\n", s->path, pos);
581 ret = avformat_seek_file(s->avFormatCtx, s->selected_stream,
582 min_ts, resampled_pos, max_ts, seek_flags);
584 AUBIO_ERR("source_avcodec: failed seeking to %d in file %s",
591 #ifdef HAVE_AVRESAMPLE
592 // reset the AVAudioResampleContext
593 avresample_close(s->avr);
594 avresample_open(s->avr);
595 #elif defined(HAVE_SWRESAMPLE)
602 uint_t aubio_source_avcodec_get_duration (aubio_source_avcodec_t * s) {
603 if (s && &(s->avFormatCtx) != NULL) {
604 int64_t duration = s->avFormatCtx->duration;
605 return s->samplerate * ((uint_t)duration / 1e6 );
610 uint_t aubio_source_avcodec_close(aubio_source_avcodec_t * s) {
611 if (s->avr != NULL) {
612 #ifdef HAVE_AVRESAMPLE
613 avresample_close( s->avr );
614 #elif defined(HAVE_SWRESAMPLE)
615 swr_close ( s->avr );
620 if (s->avCodecCtx != NULL) {
621 #ifndef HAVE_AUBIO_LIBAVCODEC_DEPRECATED
622 avcodec_free_context( &s->avCodecCtx );
624 avcodec_close ( s->avCodecCtx );
627 s->avCodecCtx = NULL;
628 if (s->avFormatCtx != NULL) {
629 avformat_close_input(&s->avFormatCtx);
630 s->avFormatCtx = NULL;
632 av_packet_unref(&s->avPacket);
636 void del_aubio_source_avcodec(aubio_source_avcodec_t * s){
638 aubio_source_avcodec_close(s);
639 if (s->output != NULL) {
643 if (s->avFrame != NULL) {
644 av_frame_free( &(s->avFrame) );
654 #endif /* HAVE_LIBAV */