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/>.
26 // determine whether we use libavformat from ffmpeg or from libav
27 #define FFMPEG_LIBAVFORMAT (LIBAVFORMAT_VERSION_MICRO > 99 )
28 // max_analyze_duration2 was used from ffmpeg libavformat 55.43.100 through 57.2.100
29 #define FFMPEG_LIBAVFORMAT_MAX_DUR2 FFMPEG_LIBAVFORMAT && ( \
30 (LIBAVFORMAT_VERSION_MAJOR == 55 && LIBAVFORMAT_VERSION_MINOR >= 43) \
31 || (LIBAVFORMAT_VERSION_MAJOR == 56) \
32 || (LIBAVFORMAT_VERSION_MAJOR == 57 && LIBAVFORMAT_VERSION_MINOR < 2) \
35 #include <libavcodec/avcodec.h>
36 #include <libavformat/avformat.h>
37 #include <libavresample/avresample.h>
38 #include <libavutil/opt.h>
41 #include "aubio_priv.h"
44 #include "source_avcodec.h"
46 #define AUBIO_AVCODEC_MAX_BUFFER_SIZE FF_MIN_BUFFER_SIZE
48 struct _aubio_source_avcodec_t {
53 // some data about the file
55 uint_t input_samplerate;
56 uint_t input_channels;
59 AVFormatContext *avFormatCtx;
60 AVCodecContext *avCodecCtx;
62 AVAudioResampleContext *avr;
66 sint_t selected_stream;
71 // hack to create or re-create the context the first time _do or _do_multi is called
72 void aubio_source_avcodec_reset_resampler(aubio_source_avcodec_t * s, uint_t multi);
73 void aubio_source_avcodec_readframe(aubio_source_avcodec_t *s, uint_t * read_samples);
75 uint_t aubio_source_avcodec_has_network_url(aubio_source_avcodec_t *s);
77 uint_t aubio_source_avcodec_has_network_url(aubio_source_avcodec_t *s) {
78 char proto[20], authorization[256], hostname[128], uripath[256];
79 int proto_size = 20, authorization_size = 256, hostname_size = 128,
80 *port_ptr = 0, path_size = 256;
81 av_url_split(proto, proto_size, authorization, authorization_size, hostname,
82 hostname_size, port_ptr, uripath, path_size, s->path);
90 aubio_source_avcodec_t * new_aubio_source_avcodec(const char_t * path, uint_t samplerate, uint_t hop_size) {
91 aubio_source_avcodec_t * s = AUBIO_NEW(aubio_source_avcodec_t);
94 AUBIO_ERR("source_avcodec: Aborted opening null path\n");
97 if ((sint_t)samplerate < 0) {
98 AUBIO_ERR("source_avcodec: Can not open %s with samplerate %d\n", path, samplerate);
101 if ((sint_t)hop_size <= 0) {
102 AUBIO_ERR("source_avcodec: Can not open %s with hop_size %d\n", path, hop_size);
106 s->hop_size = hop_size;
109 if (s->path) AUBIO_FREE(s->path);
110 s->path = AUBIO_ARRAY(char_t, strnlen(path, PATH_MAX) + 1);
111 strncpy(s->path, path, strnlen(path, PATH_MAX) + 1);
113 // register all formats and codecs
116 if (aubio_source_avcodec_has_network_url(s)) {
117 avformat_network_init();
120 // try opening the file and get some info about it
121 AVFormatContext *avFormatCtx = s->avFormatCtx;
123 if ( (err = avformat_open_input(&avFormatCtx, s->path, NULL, NULL) ) < 0 ) {
125 av_strerror (err, errorstr, sizeof(errorstr));
126 AUBIO_ERR("source_avcodec: Failed opening %s (%s)\n", s->path, errorstr);
130 // try to make sure max_analyze_duration is big enough for most songs
131 #if FFMPEG_LIBAVFORMAT_MAX_DUR2
132 avFormatCtx->max_analyze_duration2 *= 100;
134 avFormatCtx->max_analyze_duration *= 100;
137 // retrieve stream information
138 if ( (err = avformat_find_stream_info(avFormatCtx, NULL)) < 0 ) {
140 av_strerror (err, errorstr, sizeof(errorstr));
141 AUBIO_ERR("source_avcodec: Could not find stream information " "for %s (%s)\n", s->path,
146 // dump information about file onto standard error
147 //av_dump_format(avFormatCtx, 0, s->path, 0);
149 // look for the first audio stream
151 sint_t selected_stream = -1;
152 for (i = 0; i < avFormatCtx->nb_streams; i++) {
153 #if FF_API_LAVF_AVCTX
154 if (avFormatCtx->streams[i]->codecpar->codec_type == AVMEDIA_TYPE_AUDIO) {
156 if (avFormatCtx->streams[i]->codec->codec_type == AVMEDIA_TYPE_AUDIO) {
158 if (selected_stream == -1) {
161 AUBIO_WRN("source_avcodec: More than one audio stream in %s, "
162 "taking the first one\n", s->path);
166 if (selected_stream == -1) {
167 AUBIO_ERR("source_avcodec: No audio stream in %s\n", s->path);
170 //AUBIO_DBG("Taking stream %d in file %s\n", selected_stream, s->path);
171 s->selected_stream = selected_stream;
173 AVCodecContext *avCodecCtx = s->avCodecCtx;
174 #if FF_API_LAVF_AVCTX
175 AVCodecParameters *codecpar = avFormatCtx->streams[selected_stream]->codecpar;
176 if (codecpar == NULL) {
177 AUBIO_ERR("source_avcodec: Could not find decoder for %s", s->path);
180 AVCodec *codec = avcodec_find_decoder(codecpar->codec_id);
182 /* Allocate a codec context for the decoder */
183 avCodecCtx = avcodec_alloc_context3(codec);
185 AUBIO_ERR("source_avcodec: Failed to allocate the %s codec context for path %s\n",
186 av_get_media_type_string(AVMEDIA_TYPE_AUDIO), s->path);
190 avCodecCtx = avFormatCtx->streams[selected_stream]->codec;
191 AVCodec *codec = avcodec_find_decoder(avCodecCtx->codec_id);
194 AUBIO_ERR("source_avcodec: Could not find decoder for %s", s->path);
198 #if FF_API_LAVF_AVCTX
199 /* Copy codec parameters from input stream to output codec context */
200 if ((err = avcodec_parameters_to_context(avCodecCtx, codecpar)) < 0) {
201 AUBIO_ERR("source_avcodec: Failed to copy %s codec parameters to decoder context for %s\n",
202 av_get_media_type_string(AVMEDIA_TYPE_AUDIO), s->path);
207 if ( ( err = avcodec_open2(avCodecCtx, codec, NULL) ) < 0) {
209 av_strerror (err, errorstr, sizeof(errorstr));
210 AUBIO_ERR("source_avcodec: Could not load codec for %s (%s)\n", s->path, errorstr);
214 /* get input specs */
215 s->input_samplerate = avCodecCtx->sample_rate;
216 s->input_channels = avCodecCtx->channels;
217 //AUBIO_DBG("input_samplerate: %d\n", s->input_samplerate);
218 //AUBIO_DBG("input_channels: %d\n", s->input_channels);
220 if (samplerate == 0) {
221 s->samplerate = s->input_samplerate;
223 s->samplerate = samplerate;
226 if (s->samplerate > s->input_samplerate) {
227 AUBIO_WRN("source_avcodec: upsampling %s from %d to %d\n", s->path,
228 s->input_samplerate, s->samplerate);
231 AVFrame *avFrame = s->avFrame;
232 avFrame = av_frame_alloc();
234 AUBIO_ERR("source_avcodec: Could not allocate frame for (%s)\n", s->path);
237 /* allocate output for avr */
238 s->output = (smpl_t *)av_malloc(AUBIO_AVCODEC_MAX_BUFFER_SIZE * sizeof(smpl_t));
243 s->avFormatCtx = avFormatCtx;
244 s->avCodecCtx = avCodecCtx;
245 s->avFrame = avFrame;
247 // default to mono output
248 aubio_source_avcodec_reset_resampler(s, 0);
253 //av_log_set_level(AV_LOG_QUIET);
258 //AUBIO_ERR("can not read %s at samplerate %dHz with a hop_size of %d\n",
259 // s->path, s->samplerate, s->hop_size);
260 del_aubio_source_avcodec(s);
264 void aubio_source_avcodec_reset_resampler(aubio_source_avcodec_t * s, uint_t multi) {
265 // create or reset resampler to/from mono/multi-channel
266 if ( (multi != s->multi) || (s->avr == NULL) ) {
267 int64_t input_layout = av_get_default_channel_layout(s->input_channels);
268 uint_t output_channels = multi ? s->input_channels : 1;
269 int64_t output_layout = av_get_default_channel_layout(output_channels);
270 AVAudioResampleContext *avr = avresample_alloc_context();
271 AVAudioResampleContext *oldavr = s->avr;
273 av_opt_set_int(avr, "in_channel_layout", input_layout, 0);
274 av_opt_set_int(avr, "out_channel_layout", output_layout, 0);
275 av_opt_set_int(avr, "in_sample_rate", s->input_samplerate, 0);
276 av_opt_set_int(avr, "out_sample_rate", s->samplerate, 0);
277 av_opt_set_int(avr, "in_sample_fmt", s->avCodecCtx->sample_fmt, 0);
278 #if HAVE_AUBIO_DOUBLE
279 av_opt_set_int(avr, "out_sample_fmt", AV_SAMPLE_FMT_DBL, 0);
281 av_opt_set_int(avr, "out_sample_fmt", AV_SAMPLE_FMT_FLT, 0);
284 //av_opt_set_int(avr, "out_sample_fmt", AV_SAMPLE_FMT_FLTP, 0);
286 if ( ( err = avresample_open(avr) ) < 0) {
288 av_strerror (err, errorstr, sizeof(errorstr));
289 AUBIO_ERR("source_avcodec: Could not open AVAudioResampleContext for %s (%s)\n",
295 if (oldavr != NULL) {
296 avresample_close( oldavr );
304 void aubio_source_avcodec_readframe(aubio_source_avcodec_t *s, uint_t * read_samples) {
305 AVFormatContext *avFormatCtx = s->avFormatCtx;
306 AVCodecContext *avCodecCtx = s->avCodecCtx;
307 AVFrame *avFrame = s->avFrame;
309 av_init_packet (&avPacket);
310 AVAudioResampleContext *avr = s->avr;
311 smpl_t *output = s->output;
316 int err = av_read_frame (avFormatCtx, &avPacket);
317 if (err == AVERROR_EOF) {
323 av_strerror (err, errorstr, sizeof(errorstr));
324 AUBIO_ERR("source_avcodec: could not read frame in %s (%s)\n", s->path, errorstr);
327 } while (avPacket.stream_index != s->selected_stream);
330 #if FF_API_LAVF_AVCTX
331 int ret = avcodec_send_packet(avCodecCtx, &avPacket);
332 if (ret < 0 && ret != AVERROR_EOF) {
333 AUBIO_ERR("source_avcodec: error when sending packet for %s\n", s->path);
336 ret = avcodec_receive_frame(avCodecCtx, avFrame);
341 if (ret == AVERROR(EAGAIN)) {
342 AUBIO_WRN("source_avcodec: output is not available right now - user must try to send new input\n");
343 } else if (ret == AVERROR_EOF) {
344 AUBIO_WRN("source_avcodec: the decoder has been fully flushed, and there will be no more output frames\n");
346 AUBIO_ERR("source_avcodec: decoding errors on %s\n", s->path);
351 int len = avcodec_decode_audio4(avCodecCtx, avFrame, &got_frame, &avPacket);
354 AUBIO_ERR("source_avcodec: error while decoding %s\n", s->path);
358 if (got_frame == 0) {
359 AUBIO_WRN("source_avcodec: did not get a frame when reading %s\n", s->path);
364 av_samples_get_buffer_size(&in_linesize, avCodecCtx->channels,
365 avFrame->nb_samples, avCodecCtx->sample_fmt, 1);
366 int in_samples = avFrame->nb_samples;
367 int out_linesize = 0;
368 int max_out_samples = AUBIO_AVCODEC_MAX_BUFFER_SIZE;
369 int out_samples = avresample_convert ( avr,
370 (uint8_t **)&output, out_linesize, max_out_samples,
371 (uint8_t **)avFrame->data, in_linesize, in_samples);
372 if (out_samples <= 0) {
373 AUBIO_WRN("source_avcodec: no sample found while converting frame (%s)\n", s->path);
377 *read_samples = out_samples;
380 s->avFormatCtx = avFormatCtx;
381 s->avCodecCtx = avCodecCtx;
382 s->avFrame = avFrame;
386 av_packet_unref(&avPacket);
389 void aubio_source_avcodec_do(aubio_source_avcodec_t * s, fvec_t * read_data, uint_t * read){
392 uint_t total_wrote = 0;
394 if (s->multi == 1) aubio_source_avcodec_reset_resampler(s, 0);
395 while (total_wrote < s->hop_size) {
396 end = MIN(s->read_samples - s->read_index, s->hop_size - total_wrote);
397 for (i = 0; i < end; i++) {
398 read_data->data[i + total_wrote] = s->output[i + s->read_index];
401 if (total_wrote < s->hop_size) {
402 uint_t avcodec_read = 0;
403 aubio_source_avcodec_readframe(s, &avcodec_read);
404 s->read_samples = avcodec_read;
410 s->read_index += end;
413 if (total_wrote < s->hop_size) {
414 for (i = total_wrote; i < s->hop_size; i++) {
415 read_data->data[i] = 0.;
421 void aubio_source_avcodec_do_multi(aubio_source_avcodec_t * s, fmat_t * read_data, uint_t * read){
424 uint_t total_wrote = 0;
426 if (s->multi == 0) aubio_source_avcodec_reset_resampler(s, 1);
427 while (total_wrote < s->hop_size) {
428 end = MIN(s->read_samples - s->read_index, s->hop_size - total_wrote);
429 for (j = 0; j < read_data->height; j++) {
430 for (i = 0; i < end; i++) {
431 read_data->data[j][i + total_wrote] =
432 s->output[(i + s->read_index) * s->input_channels + j];
436 if (total_wrote < s->hop_size) {
437 uint_t avcodec_read = 0;
438 aubio_source_avcodec_readframe(s, &avcodec_read);
439 s->read_samples = avcodec_read;
445 s->read_index += end;
448 if (total_wrote < s->hop_size) {
449 for (j = 0; j < read_data->height; j++) {
450 for (i = total_wrote; i < s->hop_size; i++) {
451 read_data->data[j][i] = 0.;
458 uint_t aubio_source_avcodec_get_samplerate(const aubio_source_avcodec_t * s) {
459 return s->samplerate;
462 uint_t aubio_source_avcodec_get_channels(const aubio_source_avcodec_t * s) {
463 return s->input_channels;
466 uint_t aubio_source_avcodec_seek (aubio_source_avcodec_t * s, uint_t pos) {
467 int64_t resampled_pos = (uint_t)ROUND(pos * (s->input_samplerate * 1. / s->samplerate));
468 int64_t min_ts = MAX(resampled_pos - 2000, 0);
469 int64_t max_ts = MIN(resampled_pos + 2000, INT64_MAX);
470 int seek_flags = AVSEEK_FLAG_FRAME | AVSEEK_FLAG_ANY;
471 int ret = avformat_seek_file(s->avFormatCtx, s->selected_stream,
472 min_ts, resampled_pos, max_ts, seek_flags);
474 AUBIO_ERR("Failed seeking to %d in file %s", pos, s->path);
480 // reset the AVAudioResampleContext
481 avresample_close(s->avr);
482 avresample_open(s->avr);
486 uint_t aubio_source_avcodec_get_duration (aubio_source_avcodec_t * s) {
487 if (s && &(s->avFormatCtx) != NULL) {
488 int64_t duration = s->avFormatCtx->duration;
489 return s->samplerate * ((uint_t)duration / 1e6 );
494 uint_t aubio_source_avcodec_close(aubio_source_avcodec_t * s) {
495 if (s->avr != NULL) {
496 avresample_close( s->avr );
500 if (s->avCodecCtx != NULL) {
501 avcodec_close ( s->avCodecCtx );
503 s->avCodecCtx = NULL;
504 if (s->avFormatCtx != NULL) {
505 avformat_close_input ( &(s->avFormatCtx) );
507 s->avFormatCtx = NULL;
511 void del_aubio_source_avcodec(aubio_source_avcodec_t * s){
513 aubio_source_avcodec_close(s);
514 if (s->output != NULL) {
518 if (s->avFrame != NULL) {
519 av_frame_free( &(s->avFrame) );
521 if (s->path) AUBIO_FREE(s->path);
526 #endif /* HAVE_LIBAV */