def run(self):
wait_for_user_trigger = False
self._conversation_stream.start_recording()
logging.info('Recording audio request.')
i = -1
# This generator yields ConverseResponse proto messages
# received from the gRPC Google Assistant API.
for resp in self._assistant.Converse(self.iter_converse_requests(),
grpc_deadline):
assistant_helpers.log_converse_response_without_audio(resp)
if resp.error.code != code_pb2.OK:
logging.error('server error: %s', resp.error.message)
break
if resp.event_type == END_OF_UTTERANCE:
logging.info('End of audio request detected')
self._conversation_stream.stop_recording()
if resp.result.spoken_request_text:
i = 0
self._stop_playback = False
logging.info('Transcript of user request: "%s".',
resp.result.spoken_request_text)
logging.info('Playing assistant response.')
#Check if the voice spoken is part of a USER COMMAND, otherwise use Google Assistant SDK
if self.search_command(resp.result.spoken_request_text):
self._found_command = True
else:
i = -1
self._stop_playback = True
if len(resp.audio_out.audio_data) > 0:
self._conversation_stream.write(resp.audio_out.audio_data)
if resp.result.spoken_response_text:
logging.info(
'Transcript of TTS response '
'(only populated from IFTTT): "%s".',
resp.result.spoken_response_text)
if resp.result.conversation_state:
conversation_state_bytes = resp.result.conversation_state
if resp.result.volume_percentage != 0:
volume_percentage = resp.result.volume_percentage
logging.info('Volume should be set to %s%%', volume_percentage)
if resp.result.microphone_mode == DIALOG_FOLLOW_ON:
wait_for_user_trigger = False
logging.info('Expecting follow-on query from user.')
print "expetinggg"
elif resp.result.microphone_mode == CLOSE_MICROPHONE:
wait_for_user_trigger = True
#Little hack to break the loop after 2 times, this has been found out that can block play back audio if the command is custom
if i >= 0:
i = i + 1
if i > 2:
break
logging.info('Finished playing assistant response.')
self._conversation_stream.stop_playback()
return wait_for_user_trigger
def converse(self):
"""Send a voice request to the Assistant and playback the response.
Returns: True if conversation should continue.
"""
continue_conversation = False
self.conversation_stream.start_recording()
logging.info('Recording audio request.')
def iter_converse_requests():
for c in self.gen_converse_requests():
assistant_helpers.log_converse_request_without_audio(c)
yield c
self.conversation_stream.start_playback()
# This generator yields ConverseResponse proto messages
# received from the gRPC Google Assistant API.
for resp in self.assistant.Converse(iter_converse_requests(),
self.deadline):
assistant_helpers.log_converse_response_without_audio(resp)
if resp.error.code != code_pb2.OK:
logging.error('server error: %s', resp.error.message)
break
if resp.event_type == END_OF_UTTERANCE:
logging.info('End of audio request detected')
self.conversation_stream.stop_recording()
if resp.result.spoken_request_text:
logging.info('Transcript of user request: "%s".',
resp.result.spoken_request_text)
logging.info('Playing assistant response.')
if len(resp.audio_out.audio_data) > 0:
self.conversation_stream.write(resp.audio_out.audio_data)
if resp.result.spoken_response_text:
logging.info(
'Transcript of TTS response '
'(only populated from IFTTT): "%s".',
resp.result.spoken_response_text)
if resp.result.conversation_state:
self.conversation_state = resp.result.conversation_state
if resp.result.volume_percentage != 0:
self.conversation_stream.volume_percentage = (
resp.result.volume_percentage
)
if resp.result.microphone_mode == DIALOG_FOLLOW_ON:
continue_conversation = True
logging.info('Expecting follow-on query from user.')
elif resp.result.microphone_mode == CLOSE_MICROPHONE:
continue_conversation = False
logging.info('Finished playing assistant response.')
self.conversation_stream.stop_playback()
return continue_conversation
def converse(self):
"""Send a voice request to the Assistant and playback the response.
Returns: True if conversation should continue.
"""
continue_conversation = False
self.conversation_stream.start_recording()
logging.info('Recording audio request.')
def iter_converse_requests():
for c in self.gen_converse_requests():
assistant_helpers.log_converse_request_without_audio(c)
yield c
self.conversation_stream.start_playback()
# This generator yields ConverseResponse proto messages
# received from the gRPC Google Assistant API.
for resp in self.assistant.Converse(iter_converse_requests(),
self.deadline):
assistant_helpers.log_converse_response_without_audio(resp)
if resp.error.code != code_pb2.OK:
logging.error('server error: %s', resp.error.message)
break
if resp.event_type == END_OF_UTTERANCE:
logging.info('End of audio request detected')
self.conversation_stream.stop_recording()
if resp.result.spoken_request_text:
logging.info('Transcript of user request: "%s".',
resp.result.spoken_request_text)
logging.info('Playing assistant response.')
if len(resp.audio_out.audio_data) > 0:
self.conversation_stream.write(resp.audio_out.audio_data)
if resp.result.spoken_response_text:
logging.info(
'Transcript of TTS response '
'(only populated from IFTTT): "%s".',
resp.result.spoken_response_text)
if resp.result.conversation_state:
self.conversation_state = resp.result.conversation_state
if resp.result.volume_percentage != 0:
self.conversation_stream.volume_percentage = (
resp.result.volume_percentage
)
if resp.result.microphone_mode == DIALOG_FOLLOW_ON:
continue_conversation = True
logging.info('Expecting follow-on query from user.')
elif resp.result.microphone_mode == CLOSE_MICROPHONE:
continue_conversation = False
logging.info('Finished playing assistant response.')
self.conversation_stream.stop_playback()
return continue_conversation
def main(api_endpoint, credentials, verbose, input_audio_file,
output_audio_file, audio_sample_rate, audio_sample_width,
audio_iter_size, audio_block_size, audio_flush_size, grpc_deadline,
*args, **kwargs):
"""Samples for the Google Assistant API.
Examples:
Run the sample with microphone input and speaker output:
$ python -m googlesamples.assistant
Run the sample with file input and speaker output:
$ python -m googlesamples.assistant -i <input file>
Run the sample with file input and output:
$ python -m googlesamples.assistant -i <input file> -o <output file>
"""
# Setup logging.
logging.basicConfig(level=logging.DEBUG if verbose else logging.INFO)
# Load credentials.
try:
creds = auth_helpers.load_credentials(
credentials, scopes=[common_settings.ASSISTANT_OAUTH_SCOPE])
except Exception as e:
logging.error('Error loading credentials: %s', e)
logging.error('Run auth_helpers to initialize new OAuth2 credentials.')
return
# Create gRPC channel
grpc_channel = auth_helpers.create_grpc_channel(
api_endpoint,
creds,
ssl_credentials_file=kwargs.get('ssl_credentials_for_testing'),
grpc_channel_options=kwargs.get('grpc_channel_option'))
logging.info('Connecting to %s', api_endpoint)
# Create Google Assistant API gRPC client.
assistant = embedded_assistant_pb2.EmbeddedAssistantStub(grpc_channel)
# Configure audio source and sink.
audio_device = None
if input_audio_file:
audio_source = audio_helpers.WaveSource(
open(input_audio_file, 'rb'),
sample_rate=audio_sample_rate,
sample_width=audio_sample_width)
else:
audio_source = audio_device = (audio_device
or audio_helpers.SoundDeviceStream(
sample_rate=audio_sample_rate,
sample_width=audio_sample_width,
block_size=audio_block_size,
flush_size=audio_flush_size))
if output_audio_file:
audio_sink = audio_helpers.WaveSink(open(output_audio_file, 'wb'),
sample_rate=audio_sample_rate,
sample_width=audio_sample_width)
else:
audio_sink = audio_device = (audio_device
or audio_helpers.SoundDeviceStream(
sample_rate=audio_sample_rate,
sample_width=audio_sample_width,
block_size=audio_block_size,
flush_size=audio_flush_size))
# Create conversation stream with the given audio source and sink.
conversation_stream = audio_helpers.ConversationStream(
source=audio_source,
sink=audio_sink,
iter_size=audio_iter_size,
)
# Interactive by default.
wait_for_user_trigger = True
# If file arguments are supplied, don't wait for user trigger.
if input_audio_file or output_audio_file:
wait_for_user_trigger = False
# Stores an opaque blob provided in ConverseResponse that,
# when provided in a follow-up ConverseRequest,
# gives the Assistant a context marker within the current state
# of the multi-Converse()-RPC "conversation".
# This value, along with MicrophoneMode, supports a more natural
# "conversation" with the Assistant.
conversation_state_bytes = None
# Stores the current volument percentage.
# Note: No volume change is currently implemented in this sample
volume_percentage = 50
while True:
conversation_stream.start_recording()
logging.info('Recording audio request.')
# This generator yields ConverseRequest to send to the gRPC
# Google Assistant API.
def gen_converse_requests():
converse_state = None
if conversation_state_bytes:
logging.debug('Sending converse_state: %s',
conversation_state_bytes)
converse_state = embedded_assistant_pb2.ConverseState(
conversation_state=conversation_state_bytes, )
config = embedded_assistant_pb2.ConverseConfig(
audio_in_config=embedded_assistant_pb2.AudioInConfig(
encoding='LINEAR16',
sample_rate_hertz=int(audio_sample_rate),
),
audio_out_config=embedded_assistant_pb2.AudioOutConfig(
encoding='LINEAR16',
sample_rate_hertz=int(audio_sample_rate),
volume_percentage=volume_percentage,
),
converse_state=converse_state)
# The first ConverseRequest must contain the ConverseConfig
# and no audio data.
yield embedded_assistant_pb2.ConverseRequest(config=config)
for data in conversation_stream:
# Subsequent requests need audio data, but not config.
yield embedded_assistant_pb2.ConverseRequest(audio_in=data)
def iter_converse_requests():
for c in gen_converse_requests():
assistant_helpers.log_converse_request_without_audio(c)
yield c
conversation_stream.start_playback()
# This generator yields ConverseResponse proto messages
# received from the gRPC Google Assistant API.
for resp in assistant.Converse(iter_converse_requests(),
grpc_deadline):
assistant_helpers.log_converse_response_without_audio(resp)
if resp.error.code != code_pb2.OK:
logging.error('server error: %s', resp.error.message)
break
if resp.event_type == END_OF_UTTERANCE:
logging.info('End of audio request detected')
conversation_stream.stop_recording()
if resp.result.spoken_request_text:
logging.info('Transcript of user request: "%s".',
resp.result.spoken_request_text)
logging.info('Playing assistant response.')
if len(resp.audio_out.audio_data) > 0:
conversation_stream.write(resp.audio_out.audio_data)
if resp.result.spoken_response_text:
logging.info(
'Transcript of TTS response '
'(only populated from IFTTT): "%s".',
resp.result.spoken_response_text)
if resp.result.conversation_state:
conversation_state_bytes = resp.result.conversation_state
if resp.result.volume_percentage != 0:
volume_percentage = resp.result.volume_percentage
logging.info('Volume should be set to %s%%', volume_percentage)
if resp.result.microphone_mode == DIALOG_FOLLOW_ON:
wait_for_user_trigger = False
logging.info('Expecting follow-on query from user.')
elif resp.result.microphone_mode == CLOSE_MICROPHONE:
wait_for_user_trigger = True
logging.info('Finished playing assistant response.')
conversation_stream.stop_playback()
# If file arguments are supplied, end the conversation.
if input_audio_file or output_audio_file:
break
if wait_for_user_trigger:
break
conversation_stream.close()
def assist(self):
# Configure audio source and sink.
self.audio_device = None
self.audio_source = self.audio_device = (
self.audio_device or audio_helpers.SoundDeviceStream(
sample_rate=self.audio_sample_rate,
sample_width=self.audio_sample_width,
block_size=self.audio_block_size,
flush_size=self.audio_flush_size))
self.audio_sink = self.audio_device = (
self.audio_device or audio_helpers.SoundDeviceStream(
sample_rate=self.audio_sample_rate,
sample_width=self.audio_sample_width,
block_size=self.audio_block_size,
flush_size=self.audio_flush_size))
# Create conversation stream with the given audio source and sink.
self.conversation_stream = audio_helpers.ConversationStream(
source=self.audio_source,
sink=self.audio_sink,
iter_size=self.audio_iter_size,
sample_width=self.audio_sample_width)
restart = False
continue_dialog = True
try:
while continue_dialog:
continue_dialog = False
# snowboydecoder.play_audio_file(snowboydecoder.DETECT_DING)
self.conversation_stream.start_recording()
self.logger.info('Recording audio request.')
# This generator yields ConverseResponse proto messages
# received from the gRPC Google Assistant API.
for resp in self.assistant.Converse(
self._iter_converse_requests(), self.grpc_deadline):
assistant_helpers.log_converse_response_without_audio(resp)
if resp.error.code != code_pb2.OK:
self.logger.error('server error: %s',
resp.error.message)
break
if resp.event_type == Assistant.END_OF_UTTERANCE:
self.logger.info('End of audio request detected')
self.conversation_stream.stop_recording()
if resp.result.spoken_request_text:
self.logger.info('Transcript of user request: "%s".',
resp.result.spoken_request_text)
self.logger.info('Playing assistant response.')
if len(resp.audio_out.audio_data) > 0:
self.conversation_stream.write(
resp.audio_out.audio_data)
if resp.result.spoken_response_text:
self.logger.info(
'Transcript of TTS response '
'(only populated from IFTTT): "%s".',
resp.result.spoken_response_text)
if resp.result.conversation_state:
self.conversation_state_bytes = resp.result.conversation_state
if resp.result.volume_percentage != 0:
volume_percentage = resp.result.volume_percentage
self.logger.info('Volume should be set to %s%%',
volume_percentage)
if resp.result.microphone_mode == self.DIALOG_FOLLOW_ON:
continue_dialog = True
self.logger.info(
'Expecting follow-on query from user.')
self.logger.info('Finished playing assistant response.')
self.conversation_stream.stop_playback()
except Exception as e:
self._create_assistant()
self.logger.exception('Skipping because of connection reset')
restart = True
try:
self.conversation_stream.close()
if restart:
self.assist()
except Exception:
self.logger.error('Failed to close conversation_stream.')