/*\r
-* copyright (c) 2010 Sveriges Television AB <info@casparcg.com>\r
+* Copyright (c) 2011 Sveriges Television AB <info@casparcg.com>\r
*\r
-* This file is part of CasparCG.\r
+* This file is part of CasparCG (www.casparcg.com).\r
*\r
-* CasparCG is free software: you can redistribute it and/or modify\r
-* it under the terms of the GNU General Public License as published by\r
-* the Free Software Foundation, either version 3 of the License, or\r
-* (at your option) any later version.\r
+* CasparCG is free software: you can redistribute it and/or modify\r
+* it under the terms of the GNU General Public License as published by\r
+* the Free Software Foundation, either version 3 of the License, or\r
+* (at your option) any later version.\r
*\r
-* CasparCG is distributed in the hope that it will be useful,\r
-* but WITHOUT ANY WARRANTY; without even the implied warranty of\r
-* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\r
-* GNU General Public License for more details.\r
-\r
-* You should have received a copy of the GNU General Public License\r
-* along with CasparCG. If not, see <http://www.gnu.org/licenses/>.\r
+* CasparCG is distributed in the hope that it will be useful,\r
+* but WITHOUT ANY WARRANTY; without even the implied warranty of\r
+* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\r
+* GNU General Public License for more details.\r
+*\r
+* You should have received a copy of the GNU General Public License\r
+* along with CasparCG. If not, see <http://www.gnu.org/licenses/>.\r
*\r
+* Author: Robert Nagy, ronag89@gmail.com\r
*/\r
+\r
#include "../../stdafx.h"\r
\r
#include "audio_mixer.h"\r
\r
#include <core/mixer/write_frame.h>\r
-#include <core/producer/frame/audio_transform.h>\r
+#include <core/producer/frame/frame_transform.h>\r
+#include <common/diagnostics/graph.h>\r
+\r
+#include <tbb/cache_aligned_allocator.h>\r
+\r
+#include <boost/range/adaptors.hpp>\r
+#include <boost/range/distance.hpp>\r
+\r
+#include <map>\r
+#include <stack>\r
+#include <vector>\r
\r
namespace caspar { namespace core {\r
- \r
-struct audio_mixer::implementation\r
+\r
+struct audio_item\r
{\r
- std::deque<std::vector<int16_t>> audio_data_;\r
- std::stack<core::audio_transform> transform_stack_;\r
+ const void* tag;\r
+ frame_transform transform;\r
+ audio_buffer audio_data;\r
+\r
+ audio_item()\r
+ {\r
+ }\r
\r
- std::map<int, core::audio_transform> prev_audio_transforms_;\r
- std::map<int, core::audio_transform> next_audio_transforms_;\r
+ audio_item(audio_item&& other)\r
+ : tag(std::move(other.tag))\r
+ , transform(std::move(other.transform))\r
+ , audio_data(std::move(other.audio_data))\r
+ {\r
+ }\r
+};\r
\r
+typedef std::vector<float, tbb::cache_aligned_allocator<float>> audio_buffer_ps;\r
+ \r
+struct audio_stream\r
+{\r
+ frame_transform prev_transform;\r
+ audio_buffer_ps audio_data;\r
+};\r
+\r
+struct audio_mixer::implementation\r
+{\r
+ safe_ptr<diagnostics::graph> graph_;\r
+ std::stack<core::frame_transform> transform_stack_;\r
+ std::map<const void*, audio_stream> audio_streams_;\r
+ std::vector<audio_item> items_;\r
+ std::vector<size_t> audio_cadence_;\r
+ video_format_desc format_desc_;\r
+ \r
public:\r
- implementation()\r
+ implementation(const safe_ptr<diagnostics::graph>& graph)\r
+ : graph_(graph)\r
+ , format_desc_(video_format_desc::get(video_format::invalid))\r
{\r
- transform_stack_.push(core::audio_transform());\r
- audio_data_.push_back(std::vector<int16_t>()); // One frame delay\r
+ graph_->set_color("volume", diagnostics::color(1.0f, 0.8f, 0.1f));\r
+ transform_stack_.push(core::frame_transform());\r
}\r
\r
void begin(core::basic_frame& frame)\r
{\r
- transform_stack_.push(transform_stack_.top()*frame.get_audio_transform());\r
+ transform_stack_.push(transform_stack_.top()*frame.get_frame_transform());\r
}\r
\r
- void visit(const core::write_frame& frame)\r
+ void visit(core::write_frame& frame)\r
{\r
- if(!transform_stack_.top().get_has_audio())\r
- return;\r
-\r
- const auto& audio_data = frame.audio_data();\r
- const auto tag = frame.tag(); // Get the identifier for the audio-stream.\r
-\r
- if(audio_data_.back().empty())\r
- audio_data_.back().resize(audio_data.size(), 0);\r
- \r
- const auto next = transform_stack_.top();\r
- auto prev = next;\r
-\r
- const auto it = prev_audio_transforms_.find(tag);\r
- if(it != prev_audio_transforms_.end())\r
- prev = it->second;\r
- \r
- next_audio_transforms_[tag] = next; // Store all active tags, inactive tags will be removed in end_pass.\r
- \r
- if(next.get_gain() < 0.001 && prev.get_gain() < 0.001)\r
+ if(transform_stack_.top().volume < 0.002 || frame.audio_data().empty())\r
return;\r
- \r
- static const int BASE = 1<<15;\r
\r
- const auto next_gain = static_cast<int>(next.get_gain()*BASE);\r
- const auto prev_gain = static_cast<int>(prev.get_gain()*BASE);\r
+ audio_item item;\r
+ item.tag = frame.tag();\r
+ item.transform = transform_stack_.top();\r
+ item.audio_data = std::move(frame.audio_data()); // Note: We don't need to care about upper/lower since audio_data is removed/moved from the last field.\r
\r
- const int n_samples = audio_data_.back().size();\r
-\r
- tbb::parallel_for\r
- (\r
- tbb::blocked_range<size_t>(0, audio_data.size()),\r
- [&](const tbb::blocked_range<size_t>& r)\r
- {\r
- for(size_t n = r.begin(); n < r.end(); ++n)\r
- {\r
- const int sample_gain = (prev_gain - (prev_gain * n)/n_samples) + (next_gain * n)/n_samples;\r
- const int sample = (static_cast<int>(audio_data[n])*sample_gain)/BASE;\r
- audio_data_.back()[n] = static_cast<int16_t>((static_cast<int>(audio_data_.back()[n]) + sample) & 0xFFFF);\r
- }\r
- }\r
- );\r
+ items_.push_back(std::move(item)); \r
}\r
\r
- void begin(const core::audio_transform& transform)\r
+ void begin(const core::frame_transform& transform)\r
{\r
transform_stack_.push(transform_stack_.top()*transform);\r
}\r
transform_stack_.pop();\r
}\r
\r
- std::vector<int16_t> mix()\r
- {\r
- prev_audio_transforms_ = std::move(next_audio_transforms_); \r
- auto result = std::move(audio_data_.front());\r
- audio_data_.pop_front();\r
- audio_data_.push_back(std::vector<int16_t>());\r
- return std::move(result);\r
+ audio_buffer mix(const video_format_desc& format_desc)\r
+ { \r
+ if(format_desc_ != format_desc)\r
+ {\r
+ audio_streams_.clear();\r
+ audio_cadence_ = format_desc.audio_cadence;\r
+ format_desc_ = format_desc;\r
+ } \r
+ \r
+ std::map<const void*, audio_stream> next_audio_streams;\r
+\r
+ BOOST_FOREACH(auto& item, items_)\r
+ { \r
+ audio_buffer_ps next_audio;\r
+\r
+ auto next_transform = item.transform;\r
+ auto prev_transform = next_transform;\r
+\r
+ const auto it = audio_streams_.find(item.tag);\r
+ if(it != audio_streams_.end())\r
+ { \r
+ prev_transform = it->second.prev_transform;\r
+ next_audio = std::move(it->second.audio_data);\r
+ }\r
+\r
+ if(prev_transform.volume < 0.001 && next_transform.volume < 0.001)\r
+ continue;\r
+ \r
+ const float prev_volume = static_cast<float>(prev_transform.volume);\r
+ const float next_volume = static_cast<float>(next_transform.volume);\r
+ \r
+ auto alpha = (next_volume-prev_volume)/static_cast<float>(item.audio_data.size()/format_desc.audio_channels);\r
+ \r
+ for(size_t n = 0; n < item.audio_data.size(); ++n)\r
+ next_audio.push_back(item.audio_data[n] * (prev_volume + (n/format_desc_.audio_channels) * alpha));\r
+ \r
+ next_audio_streams[item.tag].prev_transform = std::move(next_transform); // Store all active tags, inactive tags will be removed at the end.\r
+ next_audio_streams[item.tag].audio_data = std::move(next_audio); \r
+ } \r
+\r
+ items_.clear();\r
+\r
+ audio_streams_ = std::move(next_audio_streams);\r
+ \r
+ if(audio_streams_.empty()) \r
+ audio_streams_[nullptr].audio_data = audio_buffer_ps(audio_cadence_.front(), 0.0f);\r
+ \r
+ { // sanity check\r
+\r
+ auto nb_invalid_streams = boost::count_if(audio_streams_ | boost::adaptors::map_values, [&](const audio_stream& x)\r
+ {\r
+ return x.audio_data.size() < audio_cadence_.front();\r
+ });\r
+\r
+ if(nb_invalid_streams > 0) \r
+ CASPAR_LOG(trace) << "[audio_mixer] Incorrect frame audio cadence detected."; \r
+ }\r
+\r
+ std::vector<float> result_ps(audio_cadence_.front(), 0.0f);\r
+ BOOST_FOREACH(auto& stream, audio_streams_ | boost::adaptors::map_values)\r
+ {\r
+ //CASPAR_LOG(debug) << stream.audio_data.size() << L" : " << result_ps.size();\r
+\r
+ if(stream.audio_data.size() < result_ps.size())\r
+ {\r
+ stream.audio_data.resize(result_ps.size(), 0.0f);\r
+ CASPAR_LOG(trace) << L"[audio_mixer] Appended zero samples";\r
+ }\r
+\r
+ auto out = boost::range::transform(result_ps, stream.audio_data, std::begin(result_ps), std::plus<float>());\r
+ stream.audio_data.erase(std::begin(stream.audio_data), std::begin(stream.audio_data) + std::distance(std::begin(result_ps), out));\r
+ } \r
+ \r
+ boost::range::rotate(audio_cadence_, std::begin(audio_cadence_)+1);\r
+ \r
+ audio_buffer result;\r
+ result.reserve(result_ps.size());\r
+ boost::range::transform(result_ps, std::back_inserter(result), [](float sample){return static_cast<int32_t>(sample);}); \r
+\r
+ auto max = boost::range::max_element(result);\r
+\r
+ graph_->set_value("volume", static_cast<double>(std::abs(*max))/std::numeric_limits<int32_t>::max());\r
+\r
+ return result;\r
}\r
};\r
\r
-audio_mixer::audio_mixer() : impl_(new implementation()){}\r
+audio_mixer::audio_mixer(const safe_ptr<diagnostics::graph>& graph) : impl_(new implementation(graph)){}\r
void audio_mixer::begin(core::basic_frame& frame){impl_->begin(frame);}\r
void audio_mixer::visit(core::write_frame& frame){impl_->visit(frame);}\r
void audio_mixer::end(){impl_->end();}\r
-std::vector<int16_t> audio_mixer::mix(){return impl_->mix();}\r
-audio_mixer& audio_mixer::operator=(audio_mixer&& other)\r
-{\r
- impl_ = std::move(other.impl_);\r
- return *this;\r
-}\r
+audio_buffer audio_mixer::operator()(const video_format_desc& format_desc){return impl_->mix(format_desc);}\r
\r
}}
\ No newline at end of file
projects / casparcg / blobdiff