X-Git-Url: https://main.carlh.net/gitweb/?p=dcpomatic.git;a=blobdiff_plain;f=src%2Flib%2Fanalyse_audio_job.cc;h=6ba6b5ecfe62999243f84f4b1516bafcaccb49c7;hp=af58e77ac59058097be12614c8c8499656ccff93;hb=c04fec82d25127fafa73c3daff87bece9aa8c8e8;hpb=d2137ac5db409e686b4d9b3fa567935a5e416d41 diff --git a/src/lib/analyse_audio_job.cc b/src/lib/analyse_audio_job.cc index af58e77ac..6ba6b5ecf 100644 --- a/src/lib/analyse_audio_job.cc +++ b/src/lib/analyse_audio_job.cc @@ -1,52 +1,140 @@ /* - Copyright (C) 2012 Carl Hetherington + Copyright (C) 2012-2018 Carl Hetherington - This program is free software; you can redistribute it and/or modify + This file is part of DCP-o-matic. + + DCP-o-matic 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, + DCP-o-matic 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. + along with DCP-o-matic. If not, see . */ #include "audio_analysis.h" #include "audio_buffers.h" #include "analyse_audio_job.h" +#include "audio_content.h" #include "compose.hpp" +#include "dcpomatic_log.h" #include "film.h" #include "player.h" +#include "playlist.h" +#include "filter.h" +#include "audio_filter_graph.h" +#include "config.h" +extern "C" { +#include +#include +#ifdef DCPOMATIC_HAVE_EBUR128_PATCHED_FFMPEG +#include +#endif +} +#include +#include #include "i18n.h" using std::string; +using std::vector; using std::max; using std::min; using std::cout; using boost::shared_ptr; +using boost::dynamic_pointer_cast; +using namespace dcpomatic; +#if BOOST_VERSION >= 106100 +using namespace boost::placeholders; +#endif int const AnalyseAudioJob::_num_points = 1024; -AnalyseAudioJob::AnalyseAudioJob (shared_ptr f, shared_ptr c) - : Job (f) - , _content (c) +static void add_if_required(vector& v, size_t i, double db) +{ + if (v.size() > i) { + v[i] = pow(10, db / 20); + } +} + +/** @param from_zero true to analyse audio from time 0 in the playlist, otherwise begin at Playlist::start */ +AnalyseAudioJob::AnalyseAudioJob (shared_ptr film, shared_ptr playlist, bool from_zero) + : Job (film) + , _playlist (playlist) + , _path (film->audio_analysis_path(playlist)) + , _from_zero (from_zero) , _done (0) , _samples_per_point (1) + , _current (0) + , _sample_peak (new float[film->audio_channels()]) + , _sample_peak_frame (new Frame[film->audio_channels()]) +#ifdef DCPOMATIC_HAVE_EBUR128_PATCHED_FFMPEG + , _ebur128 (new AudioFilterGraph (film->audio_frame_rate(), film->audio_channels())) +#endif { + LOG_DEBUG_AUDIO_ANALYSIS_NC("AnalyseAudioJob::AnalyseAudioJob"); + +#ifdef DCPOMATIC_HAVE_EBUR128_PATCHED_FFMPEG + _filters.push_back (new Filter ("ebur128", "ebur128", "audio", "ebur128=peak=true")); + _ebur128->setup (_filters); +#endif + + for (int i = 0; i < film->audio_channels(); ++i) { + _sample_peak[i] = 0; + _sample_peak_frame[i] = 0; + } + + if (!_from_zero) { + _start = _playlist->start().get_value_or(DCPTime()); + } + + /* XXX: is this right? Especially for more than 5.1? */ + vector channel_corrections(film->audio_channels(), 1); + add_if_required (channel_corrections, 4, -3); // Ls + add_if_required (channel_corrections, 5, -3); // Rs + add_if_required (channel_corrections, 6, -144); // HI + add_if_required (channel_corrections, 7, -144); // VI + add_if_required (channel_corrections, 8, -3); // Lc + add_if_required (channel_corrections, 9, -3); // Rc + add_if_required (channel_corrections, 10, -3); // Lc + add_if_required (channel_corrections, 11, -3); // Rc + add_if_required (channel_corrections, 12, -144); // DBox + add_if_required (channel_corrections, 13, -144); // Sync + add_if_required (channel_corrections, 14, -144); // Sign Language + add_if_required (channel_corrections, 15, -144); // Unused + _leqm.reset(new leqm_nrt::Calculator( + film->audio_channels(), + film->audio_frame_rate(), + 24, + channel_corrections, + 850, // suggested by leqm_nrt CLI source + 64, // suggested by leqm_nrt CLI source + boost::thread::hardware_concurrency() + )); +} + +AnalyseAudioJob::~AnalyseAudioJob () +{ + stop_thread (); + BOOST_FOREACH (Filter const * i, _filters) { + delete const_cast (i); + } + delete[] _current; + delete[] _sample_peak; + delete[] _sample_peak_frame; } string AnalyseAudioJob::name () const { - return _("Analyse audio"); + return _("Analysing audio"); } string @@ -58,57 +146,132 @@ AnalyseAudioJob::json_name () const void AnalyseAudioJob::run () { - shared_ptr content = _content.lock (); - if (!content) { - return; - } + LOG_DEBUG_AUDIO_ANALYSIS_NC("AnalyseAudioJob::run"); + + shared_ptr player (new Player(_film, _playlist)); + player->set_ignore_video (); + player->set_ignore_text (); + player->set_fast (); + player->set_play_referenced (); + player->Audio.connect (bind (&AnalyseAudioJob::analyse, this, _1, _2)); - shared_ptr playlist (new Playlist); - playlist->add (content); - shared_ptr player (new Player (_film, playlist)); - - int64_t const len = _film->length().frames (_film->audio_frame_rate()); + DCPTime const length = _playlist->length (_film); + + Frame const len = DCPTime (length - _start).frames_round (_film->audio_frame_rate()); _samples_per_point = max (int64_t (1), len / _num_points); - _current.resize (_film->audio_channels ()); + delete[] _current; + _current = new AudioPoint[_film->audio_channels ()]; _analysis.reset (new AudioAnalysis (_film->audio_channels ())); - _done = 0; - DCPTime const block = DCPTime::from_seconds (1.0 / 8); - for (DCPTime t; t < _film->length(); t += block) { - analyse (player->get_audio (t, block, false)); - set_progress (t.seconds() / _film->length().seconds()); + bool has_any_audio = false; + BOOST_FOREACH (shared_ptr c, _playlist->content ()) { + if (c->audio) { + has_any_audio = true; + } + } + + if (has_any_audio) { + LOG_DEBUG_AUDIO_ANALYSIS("Seeking to %1", to_string(_start)); + player->seek (_start, true); + _done = 0; + LOG_DEBUG_AUDIO_ANALYSIS("Starting loop for playlist of length %1", to_string(length)); + while (!player->pass ()) {} + } + + LOG_DEBUG_AUDIO_ANALYSIS_NC("Loop complete"); + + vector sample_peak; + for (int i = 0; i < _film->audio_channels(); ++i) { + sample_peak.push_back ( + AudioAnalysis::PeakTime (_sample_peak[i], DCPTime::from_frames (_sample_peak_frame[i], _film->audio_frame_rate ())) + ); + } + _analysis->set_sample_peak (sample_peak); + +#ifdef DCPOMATIC_HAVE_EBUR128_PATCHED_FFMPEG + if (Config::instance()->analyse_ebur128 ()) { + void* eb = _ebur128->get("Parsed_ebur128_0")->priv; + vector true_peak; + for (int i = 0; i < _film->audio_channels(); ++i) { + true_peak.push_back (av_ebur128_get_true_peaks(eb)[i]); + } + _analysis->set_true_peak (true_peak); + _analysis->set_integrated_loudness (av_ebur128_get_integrated_loudness(eb)); + _analysis->set_loudness_range (av_ebur128_get_loudness_range(eb)); + } +#endif + + if (_playlist->content().size() == 1) { + /* If there was only one piece of content in this analysis we may later need to know what its + gain was when we analysed it. + */ + shared_ptr ac = _playlist->content().front()->audio; + if (ac) { + _analysis->set_analysis_gain (ac->gain()); + } } - _analysis->write (content->audio_analysis_path ()); - + _analysis->set_samples_per_point (_samples_per_point); + _analysis->set_sample_rate (_film->audio_frame_rate ()); + _analysis->set_leqm (_leqm->leq_m()); + _analysis->write (_path); + + LOG_DEBUG_AUDIO_ANALYSIS_NC("Job finished"); set_progress (1); set_state (FINISHED_OK); } void -AnalyseAudioJob::analyse (shared_ptr b) +AnalyseAudioJob::analyse (shared_ptr b, DCPTime time) { - 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 + LOG_DEBUG_AUDIO_ANALYSIS("Received %1 frames at %2", b->frames(), to_string(time)); + DCPOMATIC_ASSERT (time >= _start); + +#ifdef DCPOMATIC_HAVE_EBUR128_PATCHED_FFMPEG + if (Config::instance()->analyse_ebur128 ()) { + _ebur128->process (b); + } +#endif + + int const frames = b->frames (); + int const channels = b->channels (); + vector interleaved(frames * channels); + + for (int j = 0; j < channels; ++j) { + float* data = b->data(j); + for (int i = 0; i < frames; ++i) { + float s = data[i]; + + interleaved[i * channels + j] = s; + + float as = fabsf (s); + if (as < 10e-7) { + /* We may struggle to serialise and recover inf or -inf, so prevent such values by replacing with this (140dB down) */ - s = 10e-7; + s = as = 10e-7; } _current[j][AudioPoint::RMS] += pow (s, 2); - _current[j][AudioPoint::PEAK] = max (_current[j][AudioPoint::PEAK], fabsf (s)); + _current[j][AudioPoint::PEAK] = max (_current[j][AudioPoint::PEAK], as); - if ((_done % _samples_per_point) == 0) { + if (as > _sample_peak[j]) { + _sample_peak[j] = as; + _sample_peak_frame[j] = _done + i; + } + + if (((_done + i) % _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; } -} + _leqm->add(interleaved); + + _done += frames; + + DCPTime const length = _playlist->length (_film); + set_progress ((time.seconds() - _start.seconds()) / (length.seconds() - _start.seconds())); + LOG_DEBUG_AUDIO_ANALYSIS_NC("Frames processed"); +}