mirror of
https://github.com/saitohirga/WSJT-X.git
synced 2024-11-16 00:51:56 -05:00
9b942910e0
Audio will not start until at least one buffer full is achieved and as we use a large target latency of 2s to minimize CPU usage and glitches we must pad with silence when the QAudioOutput pulls buffers from the Modulator i/o device. This is all necessary with pulseaudio using the underlying o/s ALSA device, i.e. on Linux.
205 lines
4.9 KiB
C++
205 lines
4.9 KiB
C++
#include "soundout.h"
|
|
|
|
#include <QDateTime>
|
|
#include <QAudioDeviceInfo>
|
|
#include <QAudioOutput>
|
|
#include <QSysInfo>
|
|
#include <qmath.h>
|
|
#include <QDebug>
|
|
|
|
#include "moc_soundout.cpp"
|
|
|
|
#if defined (WIN32)
|
|
# define MS_BUFFERED 1000u
|
|
#else
|
|
# define MS_BUFFERED 2000u
|
|
#endif
|
|
|
|
bool SoundOutput::audioError () const
|
|
{
|
|
bool result (true);
|
|
|
|
Q_ASSERT_X (m_stream, "SoundOutput", "programming error");
|
|
if (m_stream) {
|
|
switch (m_stream->error ())
|
|
{
|
|
case QAudio::OpenError:
|
|
Q_EMIT error (tr ("An error opening the audio output device has occurred."));
|
|
break;
|
|
|
|
case QAudio::IOError:
|
|
Q_EMIT error (tr ("An error occurred during write to the audio output device."));
|
|
break;
|
|
|
|
case QAudio::UnderrunError:
|
|
Q_EMIT error (tr ("Audio data not being fed to the audio output device fast enough."));
|
|
break;
|
|
|
|
case QAudio::FatalError:
|
|
Q_EMIT error (tr ("Non-recoverable error, audio output device not usable at this time."));
|
|
break;
|
|
|
|
case QAudio::NoError:
|
|
result = false;
|
|
break;
|
|
}
|
|
}
|
|
return result;
|
|
}
|
|
|
|
void SoundOutput::setFormat (QAudioDeviceInfo const& device, unsigned channels, unsigned msBuffered)
|
|
{
|
|
Q_ASSERT (0 < channels && channels < 3);
|
|
|
|
m_msBuffered = msBuffered;
|
|
|
|
QAudioFormat format (device.preferredFormat ());
|
|
// qDebug () << "Preferred audio output format:" << format;
|
|
format.setChannelCount (channels);
|
|
format.setCodec ("audio/pcm");
|
|
format.setSampleRate (48000);
|
|
format.setSampleType (QAudioFormat::SignedInt);
|
|
format.setSampleSize (16);
|
|
format.setByteOrder (QAudioFormat::Endian (QSysInfo::ByteOrder));
|
|
if (!format.isValid ())
|
|
{
|
|
Q_EMIT error (tr ("Requested output audio format is not valid."));
|
|
}
|
|
if (!device.isFormatSupported (format))
|
|
{
|
|
Q_EMIT error (tr ("Requested output audio format is not supported on device."));
|
|
}
|
|
// qDebug () << "Selected audio output format:" << format;
|
|
|
|
m_stream.reset (new QAudioOutput (device, format));
|
|
audioError ();
|
|
m_stream->setVolume (m_volume);
|
|
m_stream->setNotifyInterval(100);
|
|
|
|
connect (m_stream.data(), &QAudioOutput::stateChanged, this, &SoundOutput::handleStateChanged);
|
|
|
|
// qDebug() << "A" << m_volume << m_stream->notifyInterval();
|
|
}
|
|
|
|
void SoundOutput::restart (QIODevice * source)
|
|
{
|
|
Q_ASSERT (m_stream);
|
|
|
|
//
|
|
// This buffer size is critical since for proper sound streaming. If
|
|
// it is too short; high activity levels on the machine can starve
|
|
// the audio buffer. On the other hand the Windows implementation
|
|
// seems to take the length of the buffer in time to stop the audio
|
|
// stream even if reset() is used.
|
|
//
|
|
// 2 seconds seems a reasonable compromise except for Windows
|
|
// where things are probably broken.
|
|
//
|
|
// we have to set this before every start on the stream because the
|
|
// Windows implementation seems to forget the buffer size after a
|
|
// stop.
|
|
m_stream->setBufferSize (m_stream->format().bytesForDuration((m_msBuffered ? m_msBuffered : MS_BUFFERED) * 1000));
|
|
// qDebug() << "B" << m_stream->bufferSize() <<
|
|
// m_stream->periodSize() << m_stream->notifyInterval();
|
|
m_stream->setCategory ("production");
|
|
m_stream->start (source);
|
|
}
|
|
|
|
void SoundOutput::suspend ()
|
|
{
|
|
if (m_stream && QAudio::ActiveState == m_stream->state ())
|
|
{
|
|
m_stream->suspend ();
|
|
audioError ();
|
|
}
|
|
}
|
|
|
|
void SoundOutput::resume ()
|
|
{
|
|
if (m_stream && QAudio::SuspendedState == m_stream->state ())
|
|
{
|
|
m_stream->resume ();
|
|
audioError ();
|
|
}
|
|
}
|
|
|
|
void SoundOutput::reset ()
|
|
{
|
|
if (m_stream)
|
|
{
|
|
m_stream->reset ();
|
|
audioError ();
|
|
}
|
|
}
|
|
|
|
void SoundOutput::stop ()
|
|
{
|
|
if (m_stream)
|
|
{
|
|
m_stream->stop ();
|
|
audioError ();
|
|
}
|
|
}
|
|
|
|
qreal SoundOutput::attenuation () const
|
|
{
|
|
return -(20. * qLn (m_volume) / qLn (10.));
|
|
}
|
|
|
|
void SoundOutput::setAttenuation (qreal a)
|
|
{
|
|
Q_ASSERT (0. <= a && a <= 999.);
|
|
m_volume = qPow(10.0, -a/20.0);
|
|
// qDebug () << "SoundOut: attn = " << a << ", vol = " << m_volume;
|
|
if (m_stream)
|
|
{
|
|
m_stream->setVolume (m_volume);
|
|
}
|
|
}
|
|
|
|
void SoundOutput::resetAttenuation ()
|
|
{
|
|
m_volume = 1.;
|
|
if (m_stream)
|
|
{
|
|
m_stream->setVolume (m_volume);
|
|
}
|
|
}
|
|
|
|
void SoundOutput::handleStateChanged (QAudio::State newState)
|
|
{
|
|
// qDebug () << "SoundOutput::handleStateChanged: newState:" << newState;
|
|
|
|
switch (newState)
|
|
{
|
|
case QAudio::IdleState:
|
|
Q_EMIT status (tr ("Idle"));
|
|
break;
|
|
|
|
case QAudio::ActiveState:
|
|
Q_EMIT status (tr ("Sending"));
|
|
break;
|
|
|
|
case QAudio::SuspendedState:
|
|
Q_EMIT status (tr ("Suspended"));
|
|
break;
|
|
|
|
#if QT_VERSION >= QT_VERSION_CHECK (5, 10, 0)
|
|
case QAudio::InterruptedState:
|
|
Q_EMIT status (tr ("Interrupted"));
|
|
break;
|
|
#endif
|
|
|
|
case QAudio::StoppedState:
|
|
if (audioError ())
|
|
{
|
|
Q_EMIT status (tr ("Error"));
|
|
}
|
|
else
|
|
{
|
|
Q_EMIT status (tr ("Stopped"));
|
|
}
|
|
break;
|
|
}
|
|
}
|