Merge branch 'main' into full-webrtc

# Conflicts:
#	src/WebRTCDSP.cpp
This commit is contained in:
Kirill Kirilenko 2020-11-12 18:00:31 +03:00
commit 45e6daf262
6 changed files with 190 additions and 24 deletions

View file

@ -1,6 +1,7 @@
#include "AudioProcessor.h"
#include "SpeexDSP.h"
#include "Timer.h"
#include "WebRTCDSP.h"
#include <QAudioBuffer>
@ -9,7 +10,7 @@
namespace SpeexWebRTCTest {
namespace {
Q_LOGGING_CATEGORY(AudioProcessor, "processor")
Q_LOGGING_CATEGORY(processor, "processor")
}
QVector<qreal> calculateAudioLevels(const QAudioBuffer& buffer);
@ -78,6 +79,8 @@ void AudioProcessor::process()
{
while (doWork_)
{
auto waitUntil = std::chrono::high_resolution_clock::now() + std::chrono::milliseconds(5);
const std::size_t bytesToRead =
bufferSize_ * format_.sampleSize() / 8 * format_.channelCount();
const std::size_t monitorToRead =
@ -122,12 +125,14 @@ void AudioProcessor::process()
emit readyRead();
}
std::this_thread::sleep_for(std::chrono::milliseconds(5));
std::this_thread::sleep_until(waitUntil);
}
}
void AudioProcessor::processBuffer(QAudioBuffer& inputBuffer, const QAudioBuffer& monitorBuffer)
{
TIMER(qDebug(processor))
QVector<qreal> inputLevels = calculateAudioLevels(inputBuffer);
if (dsp_)
@ -157,6 +162,10 @@ qint64 AudioProcessor::bytesAvailable() const
bool AudioProcessor::open(QIODevice::OpenMode mode)
{
std::unique_lock<std::mutex> lock1(inputMutex_);
std::unique_lock<std::mutex> lock2(outputMutex_);
std::unique_lock<std::mutex> lock3(monitorMutex_);
inputBuffer_.clear();
outputBuffer_.clear();
monitorBuffer_.clear();
@ -195,6 +204,14 @@ Backend AudioProcessor::getCurrentBackend() const
void AudioProcessor::switchBackend(Backend backend)
{
std::unique_lock<std::mutex> lock1(inputMutex_);
std::unique_lock<std::mutex> lock2(outputMutex_);
std::unique_lock<std::mutex> lock3(monitorMutex_);
inputBuffer_.clear();
outputBuffer_.clear();
monitorBuffer_.clear();
if (backend == Backend::Speex)
dsp_.reset(new SpeexDSP(format_, monitorFormat_));
else

View file

@ -12,11 +12,11 @@ file(GLOB SOURCES *.cpp)
file(GLOB HEADERS *.h)
file(GLOB_RECURSE RESOURCES *.qrc)
if (WIN32 AND CMAKE_BUILD_TYPE STREQUAL "Release")
add_executable(${TARGET_NAME} WIN32 ${SOURCES} ${HEADERS} ${RESOURCES})
else()
#if (WIN32 AND CMAKE_BUILD_TYPE STREQUAL "Release")
# add_executable(${TARGET_NAME} WIN32 ${SOURCES} ${HEADERS} ${RESOURCES})
#else()
add_executable(${TARGET_NAME} ${SOURCES} ${HEADERS} ${RESOURCES})
endif()
#endif()
target_link_libraries(${TARGET_NAME}
speexdsp

View file

@ -85,6 +85,18 @@ MainWindow::~MainWindow()
stopRecording();
}
void fixFormatForDevice(QAudioFormat& format, const QAudioDeviceInfo& info)
{
if (!info.isFormatSupported(format))
{
QAudioFormat newFormat = info.nearestFormat(format);
qWarning(Gui).nospace() << "Preferred format " << format << " is not supported by device "
<< info.deviceName() << ".";
qWarning(Gui) << "Trying to use nearest format" << newFormat;
format = newFormat;
}
}
void MainWindow::initializeAudio(const QAudioDeviceInfo& inputDeviceInfo,
const QAudioDeviceInfo& outputDeviceInfo,
const QAudioDeviceInfo& monitorDeviceInfo)
@ -94,24 +106,9 @@ void MainWindow::initializeAudio(const QAudioDeviceInfo& inputDeviceInfo,
auto outputFormat = getOutputFormat();
auto monitorFormat = getMonitorFormat();
if (!inputDeviceInfo.isFormatSupported(captureFormat))
{
qWarning(Gui)
<< "Preferred format is not supported by input device - trying to use nearest";
captureFormat = inputDeviceInfo.nearestFormat(captureFormat);
}
if (!monitorDeviceInfo.isFormatSupported(monitorFormat))
{
qWarning(Gui)
<< "Preferred format is not supported by monitor device - trying to use nearest";
monitorFormat = inputDeviceInfo.nearestFormat(monitorFormat);
}
if (!outputDeviceInfo.isFormatSupported(outputFormat))
{
qWarning(Gui)
<< "Preferred format is not supported by output device - trying to use nearest";
outputFormat = inputDeviceInfo.nearestFormat(outputFormat);
}
fixFormatForDevice(captureFormat, inputDeviceInfo);
fixFormatForDevice(outputFormat, outputDeviceInfo);
fixFormatForDevice(monitorFormat, monitorDeviceInfo);
audioInput_.reset(new QAudioInput(inputDeviceInfo, captureFormat));
audioOutput_.reset(new QAudioOutput(outputDeviceInfo, outputFormat));

View file

@ -1,5 +1,7 @@
#include "SpeexDSP.h"
#include "Timer.h"
#include <speex/speex_echo.h>
#include <speex/speex_preprocess.h>
@ -36,6 +38,8 @@ SpeexDSP::~SpeexDSP()
void SpeexDSP::processFrame(QAudioBuffer& mainBuffer, const QAudioBuffer& auxBuffer)
{
TIMER(qDebug(Speex))
qDebug(Speex).noquote() << QString(
"got %1 near-end samples (%2ms) and %3 far-end samples (%4ms)")
.arg(mainBuffer.frameCount())

144
src/Timer.h Normal file
View file

@ -0,0 +1,144 @@
#ifndef _SPEEX_WEBRTC_TEST_TIMER_H_
#define _SPEEX_WEBRTC_TEST_TIMER_H_
#include <QDebug>
#include <chrono>
#include <stdexcept>
#include <utility>
namespace SpeexWebRTCTest {
namespace _detail {
template <typename... Ts>
struct is_one_of;
template <typename T, typename U>
struct is_one_of<T, U>
{
static constexpr bool value = std::is_same<T, U>::value;
};
template <typename T, typename U, typename... Ts>
struct is_one_of<T, U, Ts...>
{
static constexpr bool value = std::is_same<T, U>::value || is_one_of<T, Ts...>::value;
};
template <class Duration,
typename = typename std::enable_if<is_one_of<Duration,
std::chrono::hours,
std::chrono::minutes,
std::chrono::seconds,
std::chrono::milliseconds,
std::chrono::microseconds,
std::chrono::nanoseconds>::value>::type>
constexpr const char* duration_suffix()
{
if (std::is_same<Duration, std::chrono::hours>::value)
{
return "h";
}
else if (std::is_same<Duration, std::chrono::minutes>::value)
{
return "min";
}
else if (std::is_same<Duration, std::chrono::seconds>::value)
{
return "s";
}
else if (std::is_same<Duration, std::chrono::milliseconds>::value)
{
return "ms";
}
else if (std::is_same<Duration, std::chrono::microseconds>::value)
{
return "us";
}
else if (std::is_same<Duration, std::chrono::nanoseconds>::value)
{
return "ns";
}
return "";
}
template <int, class DurationIn>
std::string format_impl(DurationIn)
{
return {};
}
template <int count, class DurationIn, class FirstDuration, class... RestDurations>
std::string format_impl(DurationIn d)
{
std::string out;
auto val = std::chrono::duration_cast<FirstDuration>(d);
if (val.count() != 0)
{
out = std::to_string(val.count()) + duration_suffix<FirstDuration>();
if (sizeof...(RestDurations) > 0)
{
out += " " + format_impl<count + 1, DurationIn, RestDurations...>(d - val);
}
}
else
{
if (sizeof...(RestDurations) > 0)
{
out = format_impl<count, DurationIn, RestDurations...>(d);
}
else if (count == 0)
{
out = std::to_string(0) + duration_suffix<FirstDuration>();
}
}
return out;
}
} // namespace _detail
template <class... RestDurations, class DurationIn>
std::string format(DurationIn d)
{
return _detail::format_impl<0, DurationIn, RestDurations...>(d);
}
template <typename... Durations>
class Timer final
{
public:
explicit Timer(const QDebug& debug, QString name = "") : debug_(debug), name_(std::move(name))
{
start_ = std::chrono::high_resolution_clock::now();
}
~Timer()
{
TimePoint stop = std::chrono::high_resolution_clock::now();
if (name_.isEmpty())
debug_.nospace().noquote() << "Timer: " << format<Durations...>(stop - start_).c_str();
else
debug_.nospace().noquote()
<< "Timer [" << name_ << "]: " << format<Durations...>(stop - start_).c_str();
}
private:
using TimePoint = std::chrono::time_point<std::chrono::high_resolution_clock,
std::chrono::high_resolution_clock::duration>;
QDebug debug_;
QString name_;
TimePoint start_;
};
using MsTimer = Timer<std::chrono::seconds, std::chrono::milliseconds, std::chrono::microseconds>;
} // namespace SpeexWebRTCTest
#define TIMER(debug) MsTimer timer(debug, __func__);
#endif //_SPEEX_WEBRTC_TEST_TIMER_H_

View file

@ -1,5 +1,7 @@
#include "WebRTCDSP.h"
#include "Timer.h"
#include <webrtc/api/audio/audio_frame.h>
#include <webrtc/modules/audio_processing/include/audio_processing.h>
@ -110,6 +112,8 @@ QString errorDescription(int error)
void WebRTCDSP::processFrame(QAudioBuffer& mainBuffer, const QAudioBuffer& auxBuffer)
{
TIMER(qDebug(WebRTC))
qDebug(WebRTC).noquote() << QString(
"got %1 near-end samples at %2 Hz (%3ms) and %4 far-end "
"samples at %5 (%6ms)")