diff options
| author | Carl Hetherington <cth@carlh.net> | 2013-03-01 23:36:22 +0000 |
|---|---|---|
| committer | Carl Hetherington <cth@carlh.net> | 2013-03-01 23:36:22 +0000 |
| commit | ab1a666e724911c41dfe08fc96748b38ace839c1 (patch) | |
| tree | 3b9bd2f9fe80156e5dfaf4e5cb72080e21ba9816 /src/lib | |
| parent | b1b3e2e30ce8152b03c9cd3502a00b1814c654ea (diff) | |
| parent | ff950a024d0bfd892354d53e851c3915a68d2c89 (diff) | |
Merge.
Diffstat (limited to 'src/lib')
| -rw-r--r-- | src/lib/analyse_audio_job.cc | 110 | ||||
| -rw-r--r-- | src/lib/analyse_audio_job.h | 44 | ||||
| -rw-r--r-- | src/lib/audio_analysis.cc | 125 | ||||
| -rw-r--r-- | src/lib/audio_analysis.h | 67 | ||||
| -rw-r--r-- | src/lib/encoder.cc | 26 | ||||
| -rw-r--r-- | src/lib/ffmpeg_decoder.cc | 14 | ||||
| -rw-r--r-- | src/lib/film.cc | 46 | ||||
| -rw-r--r-- | src/lib/film.h | 8 | ||||
| -rw-r--r-- | src/lib/job.cc | 1 | ||||
| -rw-r--r-- | src/lib/job.h | 1 | ||||
| -rw-r--r-- | src/lib/options.h | 6 | ||||
| -rw-r--r-- | src/lib/util.cc | 85 | ||||
| -rw-r--r-- | src/lib/util.h | 16 | ||||
| -rw-r--r-- | src/lib/writer.cc | 6 | ||||
| -rw-r--r-- | src/lib/wscript | 2 |
15 files changed, 513 insertions, 44 deletions
diff --git a/src/lib/analyse_audio_job.cc b/src/lib/analyse_audio_job.cc new file mode 100644 index 000000000..ca316f70e --- /dev/null +++ b/src/lib/analyse_audio_job.cc @@ -0,0 +1,110 @@ +/* + Copyright (C) 2012 Carl Hetherington <cth@carlh.net> + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation; either version 2 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program; if not, write to the Free Software + Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + +*/ + +#include "audio_analysis.h" +#include "analyse_audio_job.h" +#include "compose.hpp" +#include "film.h" +#include "options.h" +#include "decoder_factory.h" +#include "audio_decoder.h" + +#include "i18n.h" + +using std::string; +using std::max; +using std::cout; +using boost::shared_ptr; + +int const AnalyseAudioJob::_num_points = 1024; + +AnalyseAudioJob::AnalyseAudioJob (shared_ptr<Film> f) + : Job (f) + , _done (0) + , _samples_per_point (1) +{ + +} + +string +AnalyseAudioJob::name () const +{ + return String::compose (_("Analyse audio of %1"), _film->name()); +} + +void +AnalyseAudioJob::run () +{ + if (!_film->audio_stream () || !_film->length()) { + set_progress (1); + set_state (FINISHED_ERROR); + return; + } + + DecodeOptions options; + options.decode_video = false; + + Decoders decoders = decoder_factory (_film, options); + assert (decoders.audio); + + decoders.audio->set_audio_stream (_film->audio_stream ()); + decoders.audio->Audio.connect (bind (&AnalyseAudioJob::audio, this, _1)); + + int64_t total_audio_frames = video_frames_to_audio_frames (_film->length().get(), _film->audio_stream()->sample_rate(), _film->frames_per_second()); + _samples_per_point = total_audio_frames / _num_points; + + _current.resize (_film->audio_stream()->channels ()); + _analysis.reset (new AudioAnalysis (_film->audio_stream()->channels())); + + while (!decoders.audio->pass()) { + set_progress (float (_done) / total_audio_frames); + } + + _analysis->write (_film->audio_analysis_path ()); + + set_progress (1); + set_state (FINISHED_OK); +} + +void +AnalyseAudioJob::audio (shared_ptr<AudioBuffers> b) +{ + for (int i = 0; i < b->frames(); ++i) { + for (int j = 0; j < b->channels(); ++j) { + float s = b->data(j)[i]; + if (fabsf (s) < 10e-7) { + /* stringstream can't serialise and recover inf or -inf, so prevent such + values by replacing with this (140dB down) */ + s = 10e-7; + } + _current[j][AudioPoint::RMS] += pow (s, 2); + _current[j][AudioPoint::PEAK] = max (_current[j][AudioPoint::PEAK], fabsf (s)); + + if ((_done % _samples_per_point) == 0) { + _current[j][AudioPoint::RMS] = sqrt (_current[j][AudioPoint::RMS] / _samples_per_point); + _analysis->add_point (j, _current[j]); + + _current[j] = AudioPoint (); + } + } + + ++_done; + } +} + diff --git a/src/lib/analyse_audio_job.h b/src/lib/analyse_audio_job.h new file mode 100644 index 000000000..dc1e073ee --- /dev/null +++ b/src/lib/analyse_audio_job.h @@ -0,0 +1,44 @@ +/* + Copyright (C) 2012 Carl Hetherington <cth@carlh.net> + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation; either version 2 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program; if not, write to the Free Software + Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + +*/ + +#include "job.h" +#include "audio_analysis.h" + +class AudioBuffers; + +class AnalyseAudioJob : public Job +{ +public: + AnalyseAudioJob (boost::shared_ptr<Film> f); + + std::string name () const; + void run (); + +private: + void audio (boost::shared_ptr<AudioBuffers>); + + int64_t _done; + int64_t _samples_per_point; + std::vector<AudioPoint> _current; + + boost::shared_ptr<AudioAnalysis> _analysis; + + static const int _num_points; +}; + diff --git a/src/lib/audio_analysis.cc b/src/lib/audio_analysis.cc new file mode 100644 index 000000000..9d708bbfd --- /dev/null +++ b/src/lib/audio_analysis.cc @@ -0,0 +1,125 @@ +/* + Copyright (C) 2012 Carl Hetherington <cth@carlh.net> + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation; either version 2 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program; if not, write to the Free Software + Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + +*/ + +#include <stdint.h> +#include <cmath> +#include <cassert> +#include <fstream> +#include <boost/filesystem.hpp> +#include "audio_analysis.h" + +using std::ostream; +using std::istream; +using std::string; +using std::ofstream; +using std::ifstream; +using std::vector; +using std::cout; +using std::max; +using std::list; + +AudioPoint::AudioPoint () +{ + for (int i = 0; i < COUNT; ++i) { + _data[i] = 0; + } +} + +AudioPoint::AudioPoint (istream& s) +{ + for (int i = 0; i < COUNT; ++i) { + s >> _data[i]; + } +} + +void +AudioPoint::write (ostream& s) const +{ + for (int i = 0; i < COUNT; ++i) { + s << _data[i] << "\n"; + } +} + + +AudioAnalysis::AudioAnalysis (int channels) +{ + _data.resize (channels); +} + +AudioAnalysis::AudioAnalysis (string filename) +{ + ifstream f (filename.c_str ()); + + int channels; + f >> channels; + _data.resize (channels); + + for (int i = 0; i < channels; ++i) { + int points; + f >> points; + for (int j = 0; j < points; ++j) { + _data[i].push_back (AudioPoint (f)); + } + } +} + +void +AudioAnalysis::add_point (int c, AudioPoint const & p) +{ + assert (c < channels ()); + _data[c].push_back (p); +} + +AudioPoint +AudioAnalysis::get_point (int c, int p) const +{ + assert (p < points (c)); + return _data[c][p]; +} + +int +AudioAnalysis::channels () const +{ + return _data.size (); +} + +int +AudioAnalysis::points (int c) const +{ + assert (c < channels ()); + return _data[c].size (); +} + +void +AudioAnalysis::write (string filename) +{ + string tmp = filename + ".tmp"; + + ofstream f (tmp.c_str ()); + f << _data.size() << "\n"; + for (vector<vector<AudioPoint> >::iterator i = _data.begin(); i != _data.end(); ++i) { + f << i->size () << "\n"; + for (vector<AudioPoint>::iterator j = i->begin(); j != i->end(); ++j) { + j->write (f); + } + } + + f.close (); + boost::filesystem::rename (tmp, filename); +} diff --git a/src/lib/audio_analysis.h b/src/lib/audio_analysis.h new file mode 100644 index 000000000..6e0e2b78a --- /dev/null +++ b/src/lib/audio_analysis.h @@ -0,0 +1,67 @@ +/* + Copyright (C) 2012 Carl Hetherington <cth@carlh.net> + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation; either version 2 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program; if not, write to the Free Software + Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + +*/ + +#ifndef DVDOMATIC_AUDIO_ANALYSIS_H +#define DVDOMATIC_AUDIO_ANALYSIS_H + +#include <iostream> +#include <vector> +#include <list> + +class AudioPoint +{ +public: + enum Type { + PEAK, + RMS, + COUNT + }; + + AudioPoint (); + AudioPoint (std::istream &); + + void write (std::ostream &) const; + + float& operator[] (int t) { + return _data[t]; + } + +private: + float _data[COUNT]; +}; + +class AudioAnalysis +{ +public: + AudioAnalysis (int c); + AudioAnalysis (std::string); + + void add_point (int c, AudioPoint const & p); + + AudioPoint get_point (int c, int p) const; + int points (int c) const; + int channels () const; + + void write (std::string); + +private: + std::vector<std::vector<AudioPoint> > _data; +}; + +#endif diff --git a/src/lib/encoder.cc b/src/lib/encoder.cc index d4a27d01b..3cc643cd6 100644 --- a/src/lib/encoder.cc +++ b/src/lib/encoder.cc @@ -425,18 +425,20 @@ Encoder::encoder_thread (ServerDescription* server) void Encoder::write_audio (shared_ptr<const AudioBuffers> data) { - if (_film->audio_channels() == 1) { - /* We need to switch things around so that the mono channel is on - the centre channel of a 5.1 set (with other channels silent). - */ - - shared_ptr<AudioBuffers> b (new AudioBuffers (6, data->frames ())); - b->make_silent (libdcp::LEFT); - b->make_silent (libdcp::RIGHT); - memcpy (b->data()[libdcp::CENTRE], data->data()[0], data->frames() * sizeof(float)); - b->make_silent (libdcp::LFE); - b->make_silent (libdcp::LS); - b->make_silent (libdcp::RS); + AudioMapping m (_film->audio_channels ()); + if (m.dcp_channels() != _film->audio_channels()) { + + /* Remap (currently just for mono -> 5.1) */ + + shared_ptr<AudioBuffers> b (new AudioBuffers (m.dcp_channels(), data->frames ())); + for (int i = 0; i < m.dcp_channels(); ++i) { + optional<int> s = m.dcp_to_source (static_cast<libdcp::Channel> (i)); + if (!s) { + b->make_silent (i); + } else { + memcpy (b->data()[i], data->data()[s.get()], data->frames() * sizeof(float)); + } + } data = b; } diff --git a/src/lib/ffmpeg_decoder.cc b/src/lib/ffmpeg_decoder.cc index 8834f28ed..462db283a 100644 --- a/src/lib/ffmpeg_decoder.cc +++ b/src/lib/ffmpeg_decoder.cc @@ -199,7 +199,7 @@ FFmpegDecoder::setup_audio () void FFmpegDecoder::setup_subtitle () { - if (!_subtitle_stream) { + if (!_subtitle_stream || _subtitle_stream->id() >= int (_format_context->nb_streams)) { return; } @@ -238,8 +238,10 @@ FFmpegDecoder::pass () int frame_finished; - while (avcodec_decode_video2 (_video_codec_context, _frame, &frame_finished, &_packet) >= 0 && frame_finished) { - filter_and_emit_video (_frame); + if (_opt.decode_video) { + while (avcodec_decode_video2 (_video_codec_context, _frame, &frame_finished, &_packet) >= 0 && frame_finished) { + filter_and_emit_video (_frame); + } } if (_audio_stream && _opt.decode_audio) { @@ -260,7 +262,7 @@ FFmpegDecoder::pass () shared_ptr<FFmpegAudioStream> ffa = dynamic_pointer_cast<FFmpegAudioStream> (_audio_stream); - if (_packet.stream_index == _video_stream) { + if (_packet.stream_index == _video_stream && _opt.decode_video) { int frame_finished; int const r = avcodec_decode_video2 (_video_codec_context, _frame, &frame_finished, &_packet); @@ -290,9 +292,9 @@ FFmpegDecoder::pass () was before this packet. Until then audio is thrown away. */ - if (_first_video && _first_video.get() <= source_pts_seconds) { + if ((_first_video && _first_video.get() <= source_pts_seconds) || !_opt.decode_video) { - if (!_first_audio) { + if (!_first_audio && _opt.decode_video) { _first_audio = source_pts_seconds; /* This is our first audio frame, and if we've arrived here we must have had our diff --git a/src/lib/film.cc b/src/lib/film.cc index c119f1515..510158e94 100644 --- a/src/lib/film.cc +++ b/src/lib/film.cc @@ -51,6 +51,7 @@ #include "video_decoder.h" #include "audio_decoder.h" #include "external_audio_decoder.h" +#include "analyse_audio_job.h" #include "i18n.h" @@ -239,6 +240,15 @@ Film::video_mxf_filename () const return video_state_identifier() + N_(".mxf"); } +string +Film::audio_analysis_path () const +{ + boost::filesystem::path p; + p /= "analysis"; + p /= content_digest(); + return file (p.string ()); +} + /** Add suitable Jobs to the JobManager to create a DCP for this Film */ void Film::make_dcp () @@ -305,6 +315,19 @@ Film::make_dcp () } } +/** Start a job to analyse the audio of our content file */ +void +Film::analyse_audio () +{ + if (_analyse_audio_job) { + return; + } + + _analyse_audio_job.reset (new AnalyseAudioJob (shared_from_this())); + _analyse_audio_job->Finished.connect (bind (&Film::analyse_audio_finished, this)); + JobManager::instance()->add (_analyse_audio_job); +} + /** Start a job to examine our content file */ void Film::examine_content () @@ -319,6 +342,15 @@ Film::examine_content () } void +Film::analyse_audio_finished () +{ + ensure_ui_thread (); + _analyse_audio_job.reset (); + + AudioAnalysisFinished (); +} + +void Film::examine_content_finished () { _examine_content_job.reset (); @@ -872,6 +904,13 @@ Film::set_content (string c) set_content_audio_streams (d.audio->audio_streams ()); } + { + boost::mutex::scoped_lock lm (_state_mutex); + _content = c; + } + + signal_changed (CONTENT); + /* Start off with the first audio and subtitle streams */ if (d.audio && !d.audio->audio_streams().empty()) { set_content_audio_stream (d.audio->audio_streams().front()); @@ -881,13 +920,6 @@ Film::set_content (string c) set_subtitle_stream (d.video->subtitle_streams().front()); } - { - boost::mutex::scoped_lock lm (_state_mutex); - _content = c; - } - - signal_changed (CONTENT); - examine_content (); } catch (...) { diff --git a/src/lib/film.h b/src/lib/film.h index 04a483998..847ab434e 100644 --- a/src/lib/film.h +++ b/src/lib/film.h @@ -45,6 +45,7 @@ class Job; class Filter; class Log; class ExamineContentJob; +class AnalyseAudioJob; class ExternalAudioStream; /** @class Film @@ -65,8 +66,10 @@ public: std::string info_path (int f) const; std::string video_mxf_dir () const; std::string video_mxf_filename () const; + std::string audio_analysis_path () const; void examine_content (); + void analyse_audio (); void send_dcp_to_tms (); void make_dcp (); @@ -364,6 +367,8 @@ public: /** Emitted when some property has changed */ mutable boost::signals2::signal<void (Property)> Changed; + boost::signals2::signal<void ()> AudioAnalysisFinished; + /** Current version number of the state file */ static int const state_version; @@ -374,9 +379,12 @@ private: /** Any running ExamineContentJob, or 0 */ boost::shared_ptr<ExamineContentJob> _examine_content_job; + /** Any running AnalyseAudioJob, or 0 */ + boost::shared_ptr<AnalyseAudioJob> _analyse_audio_job; void signal_changed (Property); void examine_content_finished (); + void analyse_audio_finished (); std::string video_state_identifier () const; /** Complete path to directory containing the film metadata; diff --git a/src/lib/job.cc b/src/lib/job.cc index 77d367136..8c1612a55 100644 --- a/src/lib/job.cc +++ b/src/lib/job.cc @@ -150,7 +150,6 @@ Job::set_state (State s) if (_state == FINISHED_OK || _state == FINISHED_ERROR) { _ran_for = elapsed_time (); - Finished (); } } diff --git a/src/lib/job.h b/src/lib/job.h index 1538e2779..c98dbaea1 100644 --- a/src/lib/job.h +++ b/src/lib/job.h @@ -65,6 +65,7 @@ public: void descend (float); float overall_progress () const; + /** Emitted by the JobManagerView from the UI thread */ boost::signals2::signal<void()> Finished; protected: diff --git a/src/lib/options.h b/src/lib/options.h index 2cd7dffde..0d2c07fd5 100644 --- a/src/lib/options.h +++ b/src/lib/options.h @@ -28,11 +28,13 @@ class DecodeOptions { public: DecodeOptions () - : decode_audio (true) + : decode_video (true) + , decode_audio (true) , decode_subtitles (false) , video_sync (true) {} - + + bool decode_video; bool decode_audio; bool decode_subtitles; bool video_sync; diff --git a/src/lib/util.cc b/src/lib/util.cc index 3d70a3122..de69636da 100644 --- a/src/lib/util.cc +++ b/src/lib/util.cc @@ -450,19 +450,6 @@ dcp_audio_sample_rate (int fs) return 96000; } -int -dcp_audio_channels (int f) -{ - if (f == 1) { - /* The source is mono, so to put the mono channel into - the centre we need to generate a 5.1 soundtrack. - */ - return 6; - } - - return f; -} - bool operator== (Crop const & a, Crop const & b) { return (a.left == b.left && a.right == b.right && a.top == b.top && a.bottom == b.bottom); @@ -903,3 +890,75 @@ cpu_info () return info; } + +string +audio_channel_name (int c) +{ + assert (MAX_AUDIO_CHANNELS == 6); + + /* TRANSLATORS: these are the names of audio channels; Lfe (sub) is the low-frequency + enhancement channel (sub-woofer)./ + */ + string const channels[] = { + "Left", + "Right", + "Centre", + "Lfe (sub)", + "Left surround", + "Right surround", + }; + + return channels[c]; +} + +AudioMapping::AudioMapping (int c) + : _source_channels (c) +{ + +} + +optional<libdcp::Channel> +AudioMapping::source_to_dcp (int c) const +{ + if (c >= _source_channels) { + return optional<libdcp::Channel> (); + } + + if (_source_channels == 1) { + /* mono sources to centre */ + return libdcp::CENTRE; + } + + return static_cast<libdcp::Channel> (c); +} + +optional<int> +AudioMapping::dcp_to_source (libdcp::Channel c) const +{ + if (_source_channels == 1) { + if (c == libdcp::CENTRE) { + return 0; + } else { + return optional<int> (); + } + } + + if (static_cast<int> (c) >= _source_channels) { + return optional<int> (); + } + + return static_cast<int> (c); +} + +int +AudioMapping::dcp_channels () const +{ + if (_source_channels == 1) { + /* The source is mono, so to put the mono channel into + the centre we need to generate a 5.1 soundtrack. + */ + return 6; + } + + return _source_channels; +} diff --git a/src/lib/util.h b/src/lib/util.h index 87735ea8e..22c6ea95b 100644 --- a/src/lib/util.h +++ b/src/lib/util.h @@ -29,6 +29,7 @@ #include <vector> #include <boost/shared_ptr.hpp> #include <boost/asio.hpp> +#include <boost/optional.hpp> #include <libdcp/util.h> extern "C" { #include <libavcodec/avcodec.h> @@ -57,6 +58,7 @@ extern std::vector<std::string> split_at_spaces_considering_quotes (std::string) extern std::string md5_digest (std::string); extern std::string md5_digest (void const *, int); extern void ensure_ui_thread (); +extern std::string audio_channel_name (int); typedef int SourceFrame; @@ -178,7 +180,6 @@ struct Rect extern std::string crop_string (Position, libdcp::Size); extern int dcp_audio_sample_rate (int); -extern int dcp_audio_channels (int); extern std::string colour_lut_index_to_name (int index); extern int stride_round_up (int, int const *, int); extern int stride_lookup (int c, int const * stride); @@ -268,6 +269,19 @@ private: float** _data; }; +class AudioMapping +{ +public: + AudioMapping (int); + + boost::optional<libdcp::Channel> source_to_dcp (int c) const; + boost::optional<int> dcp_to_source (libdcp::Channel c) const; + int dcp_channels () const; + +private: + int _source_channels; +}; + extern int64_t video_frames_to_audio_frames (SourceFrame v, float audio_sample_rate, float frames_per_second); extern bool still_image_file (std::string); extern std::pair<std::string, int> cpu_info (); diff --git a/src/lib/writer.cc b/src/lib/writer.cc index d480d502a..334ecec65 100644 --- a/src/lib/writer.cc +++ b/src/lib/writer.cc @@ -74,13 +74,15 @@ Writer::Writer (shared_ptr<Film> f) _picture_asset_writer = _picture_asset->start_write (_first_nonexistant_frame > 0); - if (dcp_audio_channels (_film->audio_channels()) > 0) { + AudioMapping m (_film->audio_channels ()); + + if (m.dcp_channels() > 0) { _sound_asset.reset ( new libdcp::SoundAsset ( _film->dir (_film->dcp_name()), N_("audio.mxf"), DCPFrameRate (_film->frames_per_second()).frames_per_second, - dcp_audio_channels (_film->audio_channels()), + m.dcp_channels (), dcp_audio_sample_rate (_film->audio_stream()->sample_rate()) ) ); diff --git a/src/lib/wscript b/src/lib/wscript index 6ddb94851..d36a24e7a 100644 --- a/src/lib/wscript +++ b/src/lib/wscript @@ -4,6 +4,8 @@ import i18n sources = """ ab_transcode_job.cc ab_transcoder.cc + analyse_audio_job.cc + audio_analysis.cc audio_decoder.cc audio_source.cc config.cc |
