def __init__(self, dialog, inference_thread):
super(AudioRecorder, self).__init__()
self.dialog = dialog
self.inference_thread = inference_thread
self.inference_thread.finished.connect(self.on_transcription_finished)
self.format = QAudioFormat()
self.format.setSampleRate(16000)
self.format.setChannelCount(1)
self.format.setSampleSize(16)
self.format.setCodec("audio/pcm")
self.format.setByteOrder(QAudioFormat.LittleEndian)
self.format.setSampleType(QAudioFormat.SignedInt)
self.recorder = QAudioInput(self.format, self)
self.is_recording = False
def initialize_audio(self):
"""Set up parameters for audio recording."""
self._format = QAudioFormat()
self._format.setSampleRate(44100)
self._format.setChannelCount(1)
self._format.setSampleSize(8)
self._format.setSampleType(QAudioFormat.UnSignedInt)
self._format.setByteOrder(QAudioFormat.LittleEndian)
self._format.setCodec("audio/pcm")
device_info = QAudioDeviceInfo(self._device)
if not device_info.isFormatSupported(self._format):
print("Default format not supported - trying to use nearest.")
self._format = device_info.nearestFormat(self._format)
self._audio_data_handler = AudioDataHandler(self._format)
self._audio_input = QAudioInput(self._device, self._format)
self._audio_data_handler.data_ready.connect(self.data_ready)
def initializeAudio(self):
self.m_pullTimer = QTimer(self)
self.m_pullTimer.timeout.connect(self.pullTimerExpired)
self.m_pullMode = True
self.m_format = QAudioFormat()
self.m_format.setSampleRate(self.DataSampleRateHz)
self.m_format.setChannelCount(1)
self.m_format.setSampleSize(16)
self.m_format.setCodec('audio/pcm')
self.m_format.setByteOrder(QAudioFormat.LittleEndian)
self.m_format.setSampleType(QAudioFormat.SignedInt)
info = QAudioDeviceInfo(QAudioDeviceInfo.defaultOutputDevice())
if not info.isFormatSupported(self.m_format):
qWarning("Default format not supported - trying to use nearest")
self.m_format = info.nearestFormat(self.m_format)
self.m_generator = Generator(self.m_format,
self.DurationSeconds * 1000000, self.ToneSampleRateHz, self)
self.createAudioOutput()
def __init__(self, inputDevice, title, xlim, ylim):
super(ZeStreamWidget, self).__init__()
self.inputDevice = inputDevice
# Creating QChart
self.series = QtCharts.QLineSeries()
self.chart = QtCharts.QChart()
self.chart.addSeries(self.series)
self.axisX = QtCharts.QValueAxis()
self.axisX.setRange(0, sampleCount)
self.axisX.setLabelFormat("%g")
self.axisX.setTitleText("Samples")
self.chart.setAxisX(self.axisX, self.series)
self.axisY = QtCharts.QValueAxis()
self.axisY.setRange(-1, 1)
self.chart.setAxisY(self.axisY, self.series)
self.chart.legend().hide()
self.chart.setTitle(title)
formatAudio = QAudioFormat()
formatAudio.setSampleRate(8000)
formatAudio.setChannelCount(1)
formatAudio.setSampleSize(8)
formatAudio.setCodec("audio/pcm")
formatAudio.setByteOrder(QAudioFormat.LittleEndian)
formatAudio.setSampleType(QAudioFormat.UnSignedInt)
self.audioInput = QAudioInput(self.inputDevice, formatAudio, self)
self.ioDevice = self.audioInput.start()
self.ioDevice.readyRead.connect(self._readyRead)
self.buffer = [QPointF(x, 0) for x in range(sampleCount)]
self.series.append(self.buffer)
class AudioRecorder(QObject):
def __init__(self, dialog, inference_thread):
super(AudioRecorder, self).__init__()
self.dialog = dialog
self.inference_thread = inference_thread
self.inference_thread.finished.connect(self.on_transcription_finished)
self.format = QAudioFormat()
self.format.setSampleRate(16000)
self.format.setChannelCount(1)
self.format.setSampleSize(16)
self.format.setCodec("audio/pcm")
self.format.setByteOrder(QAudioFormat.LittleEndian)
self.format.setSampleType(QAudioFormat.SignedInt)
self.recorder = QAudioInput(self.format, self)
self.is_recording = False
@Slot()
def toggle_record(self):
if self.is_recording == False:
logging.info("Capturing sound")
self.is_recording = True
self.inference_thread.send_cmd(("start",))
self.recorded_message = self.recorder.start()
self.recorded_message.readyRead.connect(self.read_from_IO_devide)
else:
logging.info("Finished sound capturing")
self.is_recording = False
self.recorder.stop()
self.inference_thread.send_cmd(("finish",))
@Slot()
def read_from_IO_devide(self):
''' Forward available audio data to the inference thread. '''
# self.sender() is the IO device returned by QAudioInput.start()
self.inference_thread.send_cmd(("data", self.sender().readAll()))
@Slot(str)
def on_transcription_finished(self, result):
logging.info("Transcription: {result}")
self.dialog.set_user_message(result)
self.dialog.process_user_message()
self.dialog.process_machine_message()
class AudioTest(QMainWindow):
PUSH_MODE_LABEL = "Enable push mode"
PULL_MODE_LABEL = "Enable pull mode"
SUSPEND_LABEL = "Suspend playback"
RESUME_LABEL = "Resume playback"
DurationSeconds = 1
ToneSampleRateHz = 600
DataSampleRateHz = 44100
def __init__(self):
super(AudioTest, self).__init__()
self.m_device = QAudioDeviceInfo.defaultOutputDevice()
self.m_output = None
self.initializeWindow()
self.initializeAudio()
def initializeWindow(self):
layout = QVBoxLayout()
self.m_deviceBox = QComboBox()
self.m_deviceBox.activated[int].connect(self.deviceChanged)
for deviceInfo in QAudioDeviceInfo.availableDevices(
QAudio.AudioOutput):
self.m_deviceBox.addItem(deviceInfo.deviceName(), deviceInfo)
layout.addWidget(self.m_deviceBox)
self.m_modeButton = QPushButton()
self.m_modeButton.clicked.connect(self.toggleMode)
self.m_modeButton.setText(self.PUSH_MODE_LABEL)
layout.addWidget(self.m_modeButton)
self.m_suspendResumeButton = QPushButton(
clicked=self.toggleSuspendResume)
self.m_suspendResumeButton.setText(self.SUSPEND_LABEL)
layout.addWidget(self.m_suspendResumeButton)
volumeBox = QHBoxLayout()
volumeLabel = QLabel("Volume:")
self.m_volumeSlider = QSlider(Qt.Horizontal,
minimum=0,
maximum=100,
singleStep=10)
self.m_volumeSlider.valueChanged.connect(self.volumeChanged)
volumeBox.addWidget(volumeLabel)
volumeBox.addWidget(self.m_volumeSlider)
layout.addLayout(volumeBox)
window = QWidget()
window.setLayout(layout)
self.setCentralWidget(window)
def initializeAudio(self):
self.m_pullTimer = QTimer(self)
self.m_pullTimer.timeout.connect(self.pullTimerExpired)
self.m_pullMode = True
self.m_format = QAudioFormat()
self.m_format.setSampleRate(self.DataSampleRateHz)
self.m_format.setChannelCount(1)
self.m_format.setSampleSize(16)
self.m_format.setCodec('audio/pcm')
self.m_format.setByteOrder(QAudioFormat.LittleEndian)
self.m_format.setSampleType(QAudioFormat.SignedInt)
info = QAudioDeviceInfo(QAudioDeviceInfo.defaultOutputDevice())
if not info.isFormatSupported(self.m_format):
qWarning("Default format not supported - trying to use nearest")
self.m_format = info.nearestFormat(self.m_format)
self.m_generator = Generator(self.m_format,
self.DurationSeconds * 1000000,
self.ToneSampleRateHz, self)
self.createAudioOutput()
def createAudioOutput(self):
self.m_audioOutput = QAudioOutput(self.m_device, self.m_format)
self.m_audioOutput.notify.connect(self.notified)
self.m_audioOutput.stateChanged.connect(self.handleStateChanged)
self.m_generator.start()
self.m_audioOutput.start(self.m_generator)
self.m_volumeSlider.setValue(self.m_audioOutput.volume() * 100)
def deviceChanged(self, index):
self.m_pullTimer.stop()
self.m_generator.stop()
self.m_audioOutput.stop()
self.m_device = self.m_deviceBox.itemData(index)
self.createAudioOutput()
def volumeChanged(self, value):
if self.m_audioOutput is not None:
self.m_audioOutput.setVolume(value / 100.0)
def notified(self):
qWarning(
"bytesFree = %d, elapsedUSecs = %d, processedUSecs = %d" %
(self.m_audioOutput.bytesFree(), self.m_audioOutput.elapsedUSecs(),
self.m_audioOutput.processedUSecs()))
def pullTimerExpired(self):
if self.m_audioOutput is not None and self.m_audioOutput.state(
) != QAudio.StoppedState:
chunks = self.m_audioOutput.bytesFree(
) // self.m_audioOutput.periodSize()
for _ in range(chunks):
data = self.m_generator.read(self.m_audioOutput.periodSize())
if data is None or len(
data) != self.m_audioOutput.periodSize():
break
self.m_output.write(data)
def toggleMode(self):
self.m_pullTimer.stop()
self.m_audioOutput.stop()
if self.m_pullMode:
self.m_modeButton.setText(self.PULL_MODE_LABEL)
self.m_output = self.m_audioOutput.start()
self.m_pullMode = False
self.m_pullTimer.start(20)
else:
self.m_modeButton.setText(self.PUSH_MODE_LABEL)
self.m_pullMode = True
self.m_audioOutput.start(self.m_generator)
self.m_suspendResumeButton.setText(self.SUSPEND_LABEL)
def toggleSuspendResume(self):
if self.m_audioOutput.state() == QAudio.SuspendedState:
qWarning("status: Suspended, resume()")
self.m_audioOutput.resume()
self.m_suspendResumeButton.setText(self.SUSPEND_LABEL)
elif self.m_audioOutput.state() == QAudio.ActiveState:
qWarning("status: Active, suspend()")
self.m_audioOutput.suspend()
self.m_suspendResumeButton.setText(self.RESUME_LABEL)
elif self.m_audioOutput.state() == QAudio.StoppedState:
qWarning("status: Stopped, resume()")
self.m_audioOutput.resume()
self.m_suspendResumeButton.setText(self.SUSPEND_LABEL)
elif self.m_audioOutput.state() == QAudio.IdleState:
qWarning("status: IdleState")
stateMap = {
QAudio.ActiveState: "ActiveState",
QAudio.SuspendedState: "SuspendedState",
QAudio.StoppedState: "StoppedState",
QAudio.IdleState: "IdleState"
}
def handleStateChanged(self, state):
qWarning("state = " + self.stateMap.get(state, "Unknown"))
class AudioDevice(QObject):
"""Class for storing computer's audio system information."""
data_ready = Signal(np.ndarray)
audio_inputs = Signal(object)
def __init__(self, default_device_name):
super().__init__()
self.default_device_name = default_device_name
self._pull_timer = QTimer()
self._pull_timer.setInterval(3000)
self._pull_timer.timeout.connect(self.write_to_buffer)
self._audio_input = None
self._input = None
self._audio_data_handler = None
devices = QAudioDeviceInfo.availableDevices(QAudio.AudioInput)
self._device = None
self.monitors = []
for item in devices:
dev_name = item.deviceName()
if dev_name.endswith(".monitor"):
self.monitors.append(item)
if self.default_device_name and self.default_device_name == item.deviceName(
):
self._device = item
if not self._device:
try:
self._device = self.monitors[0]
except IndexError:
self._device = QAudioDeviceInfo.defaultInputDevice()
self.initialize_audio()
self.start_audio()
def initialize_audio(self):
"""Set up parameters for audio recording."""
self._format = QAudioFormat()
self._format.setSampleRate(44100)
self._format.setChannelCount(1)
self._format.setSampleSize(8)
self._format.setSampleType(QAudioFormat.UnSignedInt)
self._format.setByteOrder(QAudioFormat.LittleEndian)
self._format.setCodec("audio/pcm")
device_info = QAudioDeviceInfo(self._device)
if not device_info.isFormatSupported(self._format):
print("Default format not supported - trying to use nearest.")
self._format = device_info.nearestFormat(self._format)
self._audio_data_handler = AudioDataHandler(self._format)
self._audio_input = QAudioInput(self._device, self._format)
self._audio_data_handler.data_ready.connect(self.data_ready)
def start_audio(self):
"""Start running all audio objects."""
self._audio_data_handler.start()
self._audio_input.start(self._audio_data_handler)
self._input = self._audio_input.start()
self._pull_timer.start()
def get_input_devices(self):
devices = []
if self._device:
devices.append(self._device)
for item in self.monitors:
if item.deviceName() not in [x.deviceName() for x in devices]:
devices.append(item)
system_default = QAudioDeviceInfo.defaultInputDevice()
if system_default not in [x.deviceName() for x in devices]:
devices.append(system_default)
return devices
def get_input_device_names(self):
devices = self.get_input_devices()
names = [x.deviceName() for x in devices]
return names
@Slot(str)
def change_audio_input(self, input_name):
input_devices = self.monitors
input_devices.append(QAudioDeviceInfo.defaultInputDevice())
for i in range(len(input_devices)):
if input_devices[i].deviceName() == input_name:
self._device = input_devices[i]
self.initialize_audio()
self.start_audio()
break
@Slot()
def write_to_buffer(self):
"""Write data to buffer for later analysis."""
len_ = self._audio_input.bytesReady()
if len_ > 0:
self._audio_data_handler.writeData(self._input.readAll(), len_)
def __init__(self, device):
super(MainWindow, self).__init__()
self.series = QtCharts.QLineSeries()
self.chart = QtCharts.QChart()
self.chart.addSeries(self.series)
self.axisX = QtCharts.QValueAxis()
self.axisX.setRange(0, sampleCount)
self.axisX.setLabelFormat("%g")
self.axisX.setTitleText("Samples")
self.axisY = QtCharts.QValueAxis()
self.axisY.setRange(-1, 1)
self.axisY.setTitleText("Audio level")
self.chart.setAxisX(self.axisX, self.series)
self.chart.setAxisY(self.axisY, self.series)
self.chart.legend().hide()
self.chart.setTitle("Data from the microphone ({})".format(
device.deviceName()))
formatAudio = QAudioFormat()
formatAudio.setSampleRate(8000)
formatAudio.setChannelCount(1)
formatAudio.setSampleSize(8)
formatAudio.setCodec("audio/pcm")
formatAudio.setByteOrder(QAudioFormat.LittleEndian)
formatAudio.setSampleType(QAudioFormat.UnSignedInt)
self.audioInput = QAudioInput(device, formatAudio, self)
self.ioDevice = self.audioInput.start()
self.ioDevice.readyRead.connect(self._readyRead)
self.chartView = QtCharts.QChartView(self.chart)
self.setCentralWidget(self.chartView)
self.buffer = [QPointF(x, 0) for x in range(sampleCount)]
self.series.append(self.buffer)
def __init__(self, device):
super(MainWindow, self).__init__()
self.series = QtCharts.QLineSeries()
self.chart = QtCharts.QChart()
self.chart.addSeries(self.series)
self.axisX = QtCharts.QValueAxis()
self.axisX.setRange(0, sampleCount)
self.axisX.setLabelFormat("%g")
self.axisX.setTitleText("Samples")
self.axisY = QtCharts.QValueAxis()
self.axisY.setRange(-1, 1)
self.axisY.setTitleText("Audio level")
self.chart.setAxisX(self.axisX, self.series)
self.chart.setAxisY(self.axisY, self.series)
self.chart.legend().hide()
self.chart.setTitle("Data from the microphone ({})".format(device.deviceName()))
formatAudio = QAudioFormat()
formatAudio.setSampleRate(8000)
formatAudio.setChannelCount(1)
formatAudio.setSampleSize(8)
formatAudio.setCodec("audio/pcm")
formatAudio.setByteOrder(QAudioFormat.LittleEndian)
formatAudio.setSampleType(QAudioFormat.UnSignedInt)
self.audioInput = QAudioInput(device, formatAudio, self)
self.ioDevice = self.audioInput.start()
self.ioDevice.readyRead.connect(self._readyRead)
self.chartView = QtCharts.QChartView(self.chart)
self.setCentralWidget(self.chartView)
self.buffer = [QPointF(x, 0) for x in range(sampleCount)]
self.series.append(self.buffer)
class AudioTest(QMainWindow):
PUSH_MODE_LABEL = "Enable push mode"
PULL_MODE_LABEL = "Enable pull mode"
SUSPEND_LABEL = "Suspend playback"
RESUME_LABEL = "Resume playback"
DurationSeconds = 1
ToneSampleRateHz = 600
DataSampleRateHz = 44100
def __init__(self):
super(AudioTest, self).__init__()
self.m_device = QAudioDeviceInfo.defaultOutputDevice()
self.m_output = None
self.initializeWindow()
self.initializeAudio()
def initializeWindow(self):
layout = QVBoxLayout()
self.m_deviceBox = QComboBox()
self.m_deviceBox.activated[int].connect(self.deviceChanged)
for deviceInfo in QAudioDeviceInfo.availableDevices(QAudio.AudioOutput):
self.m_deviceBox.addItem(deviceInfo.deviceName(), deviceInfo)
layout.addWidget(self.m_deviceBox)
self.m_modeButton = QPushButton()
self.m_modeButton.clicked.connect(self.toggleMode)
self.m_modeButton.setText(self.PUSH_MODE_LABEL)
layout.addWidget(self.m_modeButton)
self.m_suspendResumeButton = QPushButton(
clicked=self.toggleSuspendResume)
self.m_suspendResumeButton.setText(self.SUSPEND_LABEL)
layout.addWidget(self.m_suspendResumeButton)
volumeBox = QHBoxLayout()
volumeLabel = QLabel("Volume:")
self.m_volumeSlider = QSlider(Qt.Horizontal, minimum=0, maximum=100,
singleStep=10)
self.m_volumeSlider.valueChanged.connect(self.volumeChanged)
volumeBox.addWidget(volumeLabel)
volumeBox.addWidget(self.m_volumeSlider)
layout.addLayout(volumeBox)
window = QWidget()
window.setLayout(layout)
self.setCentralWidget(window)
def initializeAudio(self):
self.m_pullTimer = QTimer(self)
self.m_pullTimer.timeout.connect(self.pullTimerExpired)
self.m_pullMode = True
self.m_format = QAudioFormat()
self.m_format.setSampleRate(self.DataSampleRateHz)
self.m_format.setChannelCount(1)
self.m_format.setSampleSize(16)
self.m_format.setCodec('audio/pcm')
self.m_format.setByteOrder(QAudioFormat.LittleEndian)
self.m_format.setSampleType(QAudioFormat.SignedInt)
info = QAudioDeviceInfo(QAudioDeviceInfo.defaultOutputDevice())
if not info.isFormatSupported(self.m_format):
qWarning("Default format not supported - trying to use nearest")
self.m_format = info.nearestFormat(self.m_format)
self.m_generator = Generator(self.m_format,
self.DurationSeconds * 1000000, self.ToneSampleRateHz, self)
self.createAudioOutput()
def createAudioOutput(self):
self.m_audioOutput = QAudioOutput(self.m_device, self.m_format)
self.m_audioOutput.notify.connect(self.notified)
self.m_audioOutput.stateChanged.connect(self.handleStateChanged)
self.m_generator.start()
self.m_audioOutput.start(self.m_generator)
self.m_volumeSlider.setValue(self.m_audioOutput.volume() * 100)
def deviceChanged(self, index):
self.m_pullTimer.stop()
self.m_generator.stop()
self.m_audioOutput.stop()
self.m_device = self.m_deviceBox.itemData(index)
self.createAudioOutput()
def volumeChanged(self, value):
if self.m_audioOutput is not None:
self.m_audioOutput.setVolume(value / 100.0)
def notified(self):
qWarning("bytesFree = %d, elapsedUSecs = %d, processedUSecs = %d" % (
self.m_audioOutput.bytesFree(),
self.m_audioOutput.elapsedUSecs(),
self.m_audioOutput.processedUSecs()))
def pullTimerExpired(self):
if self.m_audioOutput is not None and self.m_audioOutput.state() != QAudio.StoppedState:
chunks = self.m_audioOutput.bytesFree() // self.m_audioOutput.periodSize()
for _ in range(chunks):
data = self.m_generator.read(self.m_audioOutput.periodSize())
if data is None or len(data) != self.m_audioOutput.periodSize():
break
self.m_output.write(data)
def toggleMode(self):
self.m_pullTimer.stop()
self.m_audioOutput.stop()
if self.m_pullMode:
self.m_modeButton.setText(self.PULL_MODE_LABEL)
self.m_output = self.m_audioOutput.start()
self.m_pullMode = False
self.m_pullTimer.start(20)
else:
self.m_modeButton.setText(self.PUSH_MODE_LABEL)
self.m_pullMode = True
self.m_audioOutput.start(self.m_generator)
self.m_suspendResumeButton.setText(self.SUSPEND_LABEL)
def toggleSuspendResume(self):
if self.m_audioOutput.state() == QAudio.SuspendedState:
qWarning("status: Suspended, resume()")
self.m_audioOutput.resume()
self.m_suspendResumeButton.setText(self.SUSPEND_LABEL)
elif self.m_audioOutput.state() == QAudio.ActiveState:
qWarning("status: Active, suspend()")
self.m_audioOutput.suspend()
self.m_suspendResumeButton.setText(self.RESUME_LABEL)
elif self.m_audioOutput.state() == QAudio.StoppedState:
qWarning("status: Stopped, resume()")
self.m_audioOutput.resume()
self.m_suspendResumeButton.setText(self.SUSPEND_LABEL)
elif self.m_audioOutput.state() == QAudio.IdleState:
qWarning("status: IdleState")
stateMap = {
QAudio.ActiveState: "ActiveState",
QAudio.SuspendedState: "SuspendedState",
QAudio.StoppedState: "StoppedState",
QAudio.IdleState: "IdleState"}
def handleStateChanged(self, state):
qWarning("state = " + self.stateMap.get(state, "Unknown"))
def __init__(self, inference_thread):
super(Dialog, self).__init__()
self._btn = QPushButton('Record', self)
self._label = QLabel('', self)
layout = QVBoxLayout()
layout.addWidget(self._btn)
layout.addWidget(self._label)
self.setLayout(layout)
self._btn.clicked.connect(self._btn_clicked)
self._is_recording = False
self._inference_thread = inference_thread
self._inference_thread.finished.connect(
self._on_transcription_finished)
audio_format = QAudioFormat()
audio_format.setCodec('audio/pcm')
audio_format.setChannelCount(1)
audio_format.setSampleSize(16)
audio_format.setSampleRate(16000)
audio_format.setByteOrder(QAudioFormat.LittleEndian)
audio_format.setSampleType(QAudioFormat.SignedInt)
input_device_info = QAudioDeviceInfo.defaultInputDevice()
if not input_device_info.isFormatSupported(audio_format):
print('Can\'t record audio in 16kHz 16-bit signed PCM format.')
exit(1)
self._audio_input = QAudioInput(audio_format)
