def gen_assist_requests(input_stream):
dialog_state_in = embedded_assistant_pb2.DialogStateIn(
language_code=lang,
conversation_state=b''
)
config = embedded_assistant_pb2.AssistConfig(
audio_in_config=embedded_assistant_pb2.AudioInConfig(
encoding='LINEAR16',
sample_rate_hertz=16000,
),
audio_out_config=embedded_assistant_pb2.AudioOutConfig(
encoding='LINEAR16',
sample_rate_hertz=16000,
volume_percentage=100,
),
dialog_state_in=dialog_state_in,
device_config=embedded_assistant_pb2.DeviceConfig(
device_id=device_id,
device_model_id=device_model_id,
)
)
# Send first AssistRequest message with configuration.
yield embedded_assistant_pb2.AssistRequest(config=config)
while True:
# Read user request from file.
data = input_stream.read(block_size)
if not data:
break
# Send following AssitRequest message with audio chunks.
yield embedded_assistant_pb2.AssistRequest(audio_in=data)
def gen_assist_requests(self):
"""Yields: AssistRequest messages to send to the API."""
dialog_state_in = embedded_assistant_pb2.DialogStateIn(
language_code=self.language_code,
conversation_state=b''
)
if self.conversation_state:
logging.debug('Sending conversation state.')
dialog_state_in.conversation_state = self.conversation_state
config = embedded_assistant_pb2.AssistConfig(
audio_in_config=embedded_assistant_pb2.AudioInConfig(
encoding='LINEAR16',
sample_rate_hertz=self.conversation_stream.sample_rate,
),
audio_out_config=embedded_assistant_pb2.AudioOutConfig(
encoding='LINEAR16',
sample_rate_hertz=self.conversation_stream.sample_rate,
volume_percentage=self.conversation_stream.volume_percentage,
),
dialog_state_in=dialog_state_in,
device_config=embedded_assistant_pb2.DeviceConfig(
device_id=self.device_id,
device_model_id=self.device_model_id,
)
)
# The first AssistRequest must contain the AssistConfig
# and no audio data.
yield embedded_assistant_pb2.AssistRequest(config=config)
for data in self.conversation_stream:
# Subsequent requests need audio data, but not config.
yield embedded_assistant_pb2.AssistRequest(audio_in=data)
def _requests(self, recorder):
audio_in_config = embedded_assistant_pb2.AudioInConfig(
encoding='LINEAR16',
sample_rate_hertz=AUDIO_SAMPLE_RATE_HZ)
audio_out_config = embedded_assistant_pb2.AudioOutConfig(
encoding='LINEAR16',
sample_rate_hertz=AUDIO_SAMPLE_RATE_HZ,
volume_percentage=self._volume_percentage)
dialog_state_in = embedded_assistant_pb2.DialogStateIn(
conversation_state=self._conversation_state,
language_code=self._language_code)
config = embedded_assistant_pb2.AssistConfig(
audio_in_config=audio_in_config,
audio_out_config=audio_out_config,
device_config=self._device_config,
dialog_state_in=dialog_state_in)
yield embedded_assistant_pb2.AssistRequest(config=config)
for chunk in recorder.record(AUDIO_FORMAT,
chunk_duration_sec=0.1,
on_start=self._recording_started,
on_stop=self._recording_stopped):
yield embedded_assistant_pb2.AssistRequest(audio_in=chunk)
def gen_assist_requests(self):
"""Yields: AssistRequest messages to send to the API."""
config = embedded_assistant_pb2.AssistConfig(
audio_in_config=embedded_assistant_pb2.AudioInConfig(
encoding='LINEAR16',
sample_rate_hertz=self.conversation_stream.sample_rate,
),
audio_out_config=embedded_assistant_pb2.AudioOutConfig(
encoding='LINEAR16',
sample_rate_hertz=self.conversation_stream.sample_rate,
volume_percentage=self.conversation_stream.volume_percentage,
),
dialog_state_in=embedded_assistant_pb2.DialogStateIn(
language_code=self.language_code,
conversation_state=self.conversation_state,
is_new_conversation=self.is_new_conversation,
),
device_config=embedded_assistant_pb2.DeviceConfig(
device_id=self.device_id,
device_model_id=self.device_model_id,
))
if self.display:
config.screen_out_config.screen_mode = PLAYING
# Continue current conversation with later requests.
self.is_new_conversation = False
# The first AssistRequest must contain the AssistConfig
# and no audio data.
yield embedded_assistant_pb2.AssistRequest(config=config)
for data in self.conversation_stream:
# Subsequent requests need audio data, but not config.
yield embedded_assistant_pb2.AssistRequest(audio_in=data)
def gen_assist_requests(self):
"""Yields: AssistRequest messages to send to the API."""
config = embedded_assistant_pb2.AssistConfig(
audio_in_config=embedded_assistant_pb2.AudioInConfig(
encoding='LINEAR16',
sample_rate_hertz=self.conversation_stream.sample_rate,
),
audio_out_config=embedded_assistant_pb2.AudioOutConfig(
encoding='LINEAR16',
sample_rate_hertz=self.conversation_stream.sample_rate,
volume_percentage=self.conversation_stream.volume_percentage,
),
dialog_state_in=embedded_assistant_pb2.DialogStateIn(
language_code=self.language_code,
conversation_state=self.conversation_state,
is_new_conversation=self.is_new_conversation,
),
device_config=embedded_assistant_pb2.DeviceConfig(
device_id=self.device_id,
device_model_id=self.device_model_id,
))
if self.display:
config.screen_out_config.screen_mode = embedded_assistant_pb2.ScreenOutConfig.PLAYING
self.is_new_conversation = False
yield embedded_assistant_pb2.AssistRequest(config=config)
for data in self.conversation_stream:
yield embedded_assistant_pb2.AssistRequest(audio_in=data)
def gen_assist_requests(self):
"""Yields: AssistRequest messages to send to the API."""
config = embedded_assistant_pb2.AssistConfig(
audio_in_config=embedded_assistant_pb2.AudioInConfig(
encoding='LINEAR16',
sample_rate_hertz=self.sample_rate,
),
audio_out_config=embedded_assistant_pb2.AudioOutConfig(
encoding='LINEAR16',
sample_rate_hertz=self.sample_rate,
volume_percentage=self.volume_percentage,
),
dialog_state_in=embedded_assistant_pb2.DialogStateIn(
language_code=self.language_code,
conversation_state=self.conversation_state,
is_new_conversation=self.is_new_conversation,
),
device_config=embedded_assistant_pb2.DeviceConfig(
device_id=self.device_id,
device_model_id=self.device_model_id,
))
# Continue current conversation with later requests.
self.is_new_conversation = False
# The first AssistRequest must contain the AssistConfig
# and no audio data.
yield embedded_assistant_pb2.AssistRequest(config=config)
while self.listening:
try:
data = self.audio_queue.get(timeout=1)
except queue.Empty:
print('no data available')
break
# Subsequent requests need audio data, but not config.
yield embedded_assistant_pb2.AssistRequest(audio_in=data)
def gen_assist_requests(self, chunks):
"""Yields: AssistRequest messages to send to the API."""
config = embedded_assistant_pb2.AssistConfig(
audio_in_config=embedded_assistant_pb2.AudioInConfig(
encoding='LINEAR16',
sample_rate_hertz=16000,
),
audio_out_config=embedded_assistant_pb2.AudioOutConfig(
encoding='LINEAR16',
sample_rate_hertz=16000,
volume_percentage=1,
),
dialog_state_in=embedded_assistant_pb2.DialogStateIn(
# language_code=self.language_code,
conversation_state=self.conversation_state,
is_new_conversation=True,
),
device_config=embedded_assistant_pb2.DeviceConfig(
device_id=self.device_id,
device_model_id=self.device_model_id,
))
# The first AssistRequest must contain the AssistConfig
# and no audio data.
yield embedded_assistant_pb2.AssistRequest(config=config)
for data in chunks():
# Subsequent requests need audio data, but not config.
if data:
yield embedded_assistant_pb2.AssistRequest(audio_in=data)
def converseRequestGenerator(self):
assistantConfig = embedded_assistant_pb2.AssistConfig(
audio_in_config=embedded_assistant_pb2.AudioInConfig(
encoding='LINEAR16',
sample_rate_hertz=self.conversationStream.sample_rate,
),
audio_out_config=embedded_assistant_pb2.AudioOutConfig(
encoding='LINEAR16',
sample_rate_hertz=self.conversationStream.sample_rate,
volume_percentage=self.conversationStream.volume_percentage,
),
device_config=embedded_assistant_pb2.DeviceConfig(
device_id=self.deviceId,
device_model_id=self.modelId,
),
dialog_state_in=embedded_assistant_pb2.DialogStateIn(
language_code='en-US',
conversation_state=self.conversationStateBytes,
is_new_conversation=self.isNewConversation,
),
)
# The first request to send the the metadata about the voice request
yield embedded_assistant_pb2.AssistRequest(config=assistantConfig)
# Send the rest of the audio data
for audioData in self.conversationStream:
yield embedded_assistant_pb2.AssistRequest(audio_in=audioData)
def gen_text_assist_requests(self, text_query):
"""Yields: AssistRequest messages to send to the API in text form"""
config = embedded_assistant_pb2.AssistConfig(
audio_out_config=embedded_assistant_pb2.AudioOutConfig(
encoding='LINEAR16',
sample_rate_hertz=self.conversation_stream.sample_rate,
volume_percentage=self.conversation_stream.volume_percentage,
),
dialog_state_in=embedded_assistant_pb2.DialogStateIn(
language_code=self.language_code,
conversation_state=self.conversation_state,
is_new_conversation=self.is_new_conversation,
),
device_config=embedded_assistant_pb2.DeviceConfig(
device_id=self.device_id,
device_model_id=self.device_model_id,
),
text_query=text_query,
)
# Continue current conversation with later requests.
self.is_new_conversation = False
req = embedded_assistant_pb2.AssistRequest(config=config)
assistant_helpers.log_assist_request_without_audio(req)
yield req
def iter_log_assist_requests():
if (text_query):
config = embedded_assistant_pb2.AssistConfig(
audio_out_config=embedded_assistant_pb2.AudioOutConfig(
encoding='LINEAR16',
sample_rate_hertz=16000,
volume_percentage=0,
),
dialog_state_in=embedded_assistant_pb2.DialogStateIn(
language_code=self.language_code,
conversation_state=self.conversation_state,
is_new_conversation=self.is_new_conversation,
),
device_config=embedded_assistant_pb2.DeviceConfig(
device_id=self.device_id,
device_model_id=self.device_model_id,
),
text_query=text_query,
)
self.is_new_conversation = False
req = embedded_assistant_pb2.AssistRequest(config=config)
assistant_helpers.log_assist_request_without_audio(req)
yield req
else:
for c in self.gen_assist_requests():
assistant_helpers.log_assist_request_without_audio(c)
yield c
logging.debug('Reached end of AssistRequest iteration.')
def iter_assist_requests():
config = embedded_assistant_pb2.AssistConfig(
audio_out_config=embedded_assistant_pb2.AudioOutConfig(
encoding='LINEAR16',
sample_rate_hertz=16000,
volume_percentage=0,
),
dialog_state_in=embedded_assistant_pb2.DialogStateIn(
language_code=self.language_code,
conversation_state=self.conversation_state,
is_new_conversation=self.is_new_conversation,
),
device_config=embedded_assistant_pb2.DeviceConfig(
device_id=self.device_id,
device_model_id=self.device_model_id,
),
text_query=text_query,
)
# Continue current conversation with later requests.
self.is_new_conversation = False
if self.display:
config.screen_out_config.screen_mode = PLAYING
req = embedded_assistant_pb2.AssistRequest(config=config)
assistant_helpers.log_assist_request_without_audio(req)
yield req
def _create_config_request(self):
audio_in_config = embedded_assistant_pb2.AudioInConfig(
encoding='LINEAR16',
sample_rate_hertz=AUDIO_SAMPLE_RATE_HZ,
)
audio_out_config = embedded_assistant_pb2.AudioOutConfig(
encoding='LINEAR16',
sample_rate_hertz=AUDIO_SAMPLE_RATE_HZ,
volume_percentage=50,
)
device_config = embedded_assistant_pb2.DeviceConfig(
device_id=self.device_id,
device_model_id=self.model_id,
)
dialog_state_in = embedded_assistant_pb2.DialogStateIn(
conversation_state=self._conversation_state,
language_code=aiy.i18n.get_language_code(),
)
assist_config = embedded_assistant_pb2.AssistConfig(
audio_in_config=audio_in_config,
audio_out_config=audio_out_config,
device_config=device_config,
dialog_state_in=dialog_state_in,
)
return embedded_assistant_pb2.AssistRequest(config=assist_config)
def gen_assist_requests2(self, str):
if self.toLang == '':
self.toLang = '英語'
config = embedded_assistant_pb2.AssistConfig(
audio_out_config=embedded_assistant_pb2.AudioOutConfig(
encoding='LINEAR16',
sample_rate_hertz=self.conversation_stream.sample_rate,
volume_percentage=self.conversation_stream.volume_percentage,
),
dialog_state_in=embedded_assistant_pb2.DialogStateIn(
language_code=self.language_code,
conversation_state=self.conversation_state,
is_new_conversation=self.is_new_conversation,
),
device_config=embedded_assistant_pb2.DeviceConfig(
device_id=self.device_id,
device_model_id=self.device_model_id,
),
text_query='"' + str + '" を' + self.toLang + 'に翻訳して')
if self.display:
config.screen_out_config.screen_mode = PLAYING
# Continue current conversation with later requests.
self.is_new_conversation = False
# The first AssistRequest must contain the AssistConfig
# and no audio data.
yield embedded_assistant_pb2.AssistRequest(config=config)
def _iter_assist_requests(handler_input: HandlerInput,
text_query: str) -> AssistRequest:
"""Yields: AssistRequest messages to send to the API."""
model_id = data.GOOGLE_ASSISTANT_API['model_id']
device_id = skill_helpers.get_device_id(handler_input)
locale = getattr(handler_input.request_envelope.request, 'locale', 'en-US')
conversation_state = skill_helpers.get_session_attribute(
handler_input, 'conversation_state') # type: list
is_new_conversation = conversation_state is None
blob = bytes(conversation_state) if not is_new_conversation else None
config = embedded_assistant_pb2.AssistConfig(
audio_out_config=embedded_assistant_pb2.AudioOutConfig(
encoding='LINEAR16',
sample_rate_hertz=data.DEFAULT_AUDIO_SAMPLE_RATE,
volume_percentage=100,
),
dialog_state_in=embedded_assistant_pb2.DialogStateIn(
language_code=locale,
conversation_state=blob,
is_new_conversation=is_new_conversation,
),
device_config=embedded_assistant_pb2.DeviceConfig(
device_id=device_id,
device_model_id=model_id,
),
text_query=text_query)
# Continue current conversation with later requests.
req = embedded_assistant_pb2.AssistRequest(config=config)
yield req
def gen_final_request(self, text_query, valence, action):
# The "repeat after me [string] command makes the assistant say "You said [string]"
# I'm only asking it to play music so I have it always saying the following hard code custom command
# With some extra time, I could figure out the proper way to make it say a custom response
if action == 'pause':
text_query = 'repeat after me to pause the music, so I\'ll do that now'
elif action == 'play':
text_query = "repeat after me play music with a " + valence + " tone, so I will play " + valence + " music for you"
config = embedded_assistant_pb2.AssistConfig(
text_query=text_query,
audio_out_config=embedded_assistant_pb2.AudioOutConfig(
encoding='LINEAR16',
sample_rate_hertz=self.conversation_stream.sample_rate,
volume_percentage=self.conversation_stream.volume_percentage,
),
dialog_state_in=embedded_assistant_pb2.DialogStateIn(
language_code=self.language_code,
conversation_state=self.conversation_state,
is_new_conversation=True,
),
device_config=embedded_assistant_pb2.DeviceConfig(
device_id=self.device_id,
device_model_id=self.device_model_id,
),
)
yield embedded_assistant_pb2.AssistRequest(config=config)
def log_assist_request_without_audio(assist_request):
"""Log AssistRequest fields without audio data."""
if logging.getLogger().isEnabledFor(logging.DEBUG):
resp_copy = embedded_assistant_pb2.AssistRequest()
resp_copy.CopyFrom(assist_request)
if len(resp_copy.audio_in) > 0:
size = len(resp_copy.audio_in)
resp_copy.ClearField('audio_in')
logging.debug('AssistRequest: audio_in (%d bytes)', size)
return
logging.debug('AssistRequest: %s', resp_copy)
def gen_assist_requests(self):
dialog_state_in = embedded_assistant_pb2.DialogStateIn(
language_code=self.language_code, conversation_state=b'')
if self.conversation_state:
logging.debug('Sending conversation state.')
dialog_state_in.conversation_state = self.conversation_state
config = embedded_assistant_pb2.AssistConfig(
audio_in_config=embedded_assistant_pb2.AudioInConfig(
encoding='LINEAR16',
sample_rate_hertz=self.conversation_stream.sample_rate,
),
audio_out_config=embedded_assistant_pb2.AudioOutConfig(
encoding='LINEAR16',
sample_rate_hertz=self.conversation_stream.sample_rate,
volume_percentage=self.conversation_stream.volume_percentage,
),
dialog_state_in=dialog_state_in,
device_config=embedded_assistant_pb2.DeviceConfig(
device_id=self.device_id,
device_model_id=self.device_model_id,
))
yield embedded_assistant_pb2.AssistRequest(config=config)
for data in self.conversation_stream:
yield embedded_assistant_pb2.AssistRequest(audio_in=data)
def iter_requests():
dialog_state = embedded_assistant_pb2.DialogStateIn(
language_code=self.language,
conversation_state=self.conversation_state or b"")
audio_config = embedded_assistant_pb2.AudioOutConfig(
encoding="LINEAR16",
sample_rate_hertz=16000,
volume_percentage=0)
device_config = embedded_assistant_pb2.DeviceConfig(
device_id=self.deviceID, device_model_id=self.modelID)
config = embedded_assistant_pb2.AssistConfig(
audio_out_config=audio_config,
dialog_state_in=dialog_state,
device_config=device_config,
text_query=query)
req = embedded_assistant_pb2.AssistRequest(config=config)
yield req
def iter_assist_requests():
config = embedded_assistant_pb2.AssistConfig(
audio_out_config=embedded_assistant_pb2.AudioOutConfig(
encoding='LINEAR16',
sample_rate_hertz=16000,
volume_percentage=0,
),
dialog_state_in=embedded_assistant_pb2.DialogStateIn(
language_code='en-US',
conversation_state=None,
is_new_conversation=True,
),
device_config=embedded_assistant_pb2.DeviceConfig(
device_id='5a1b2c3d4',
device_model_id='assistant',
),
text_query=text_query)
req = embedded_assistant_pb2.AssistRequest(config=config)
yield req
def iter_assist_requests():
dialog_state_in = embedded_assistant_pb2.DialogStateIn(
language_code=self.language_code, conversation_state=b'')
if self.conversation_state:
dialog_state_in.conversation_state = self.conversation_state
config = embedded_assistant_pb2.AssistConfig(
audio_out_config=embedded_assistant_pb2.AudioOutConfig(
encoding='LINEAR16',
sample_rate_hertz=16000,
volume_percentage=0,
),
dialog_state_in=dialog_state_in,
device_config=embedded_assistant_pb2.DeviceConfig(
device_id=self.device_id,
device_model_id=self.device_model_id,
),
text_query=text_query,
)
req = embedded_assistant_pb2.AssistRequest(config=config)
yield req
def iter_assist_requests(handler_input: HandlerInput,
text_query: str) -> AssistRequest:
"""Yields: AssistRequest messages to send to the API."""
model_id = data.GOOGLE_ASSISTANT_API['model_id']
device_id = util.get_device_id(handler_input)
# TODO: hardcoded locale?
language_code = 'it-IT'
# TODO: hardcoded default volume?
volume = util.get_persistent_attribute(handler_input, 'volume', default=50)
conversation_state = util.get_session_attribute(
handler_input, 'conversation_state') # type: list
conversation_state = bytes(
conversation_state) if conversation_state is not None else None
is_new_conversation = conversation_state is None
config = embedded_assistant_pb2.AssistConfig(
audio_out_config=embedded_assistant_pb2.AudioOutConfig(
encoding='LINEAR16',
sample_rate_hertz=data.DEFAULT_AUDIO_SAMPLE_RATE,
volume_percentage=volume,
),
dialog_state_in=embedded_assistant_pb2.DialogStateIn(
language_code=language_code,
conversation_state=conversation_state,
is_new_conversation=is_new_conversation,
),
device_config=embedded_assistant_pb2.DeviceConfig(
device_id=device_id,
device_model_id=model_id,
),
text_query=text_query)
# Continue current conversation with later requests.
req = embedded_assistant_pb2.AssistRequest(config=config)
yield req
def iter_assist_requests():
config = embedded_assistant_pb2.AssistConfig(
audio_out_config=embedded_assistant_pb2.AudioOutConfig(
encoding='MP3',
sample_rate_hertz=16000,
volume_percentage=self.volume,
),
dialog_state_in=embedded_assistant_pb2.DialogStateIn(
# https://github.com/googlesamples/assistant-sdk-python/issues/284
# language_code=self.language_code,
conversation_state=self.conversation_state,
is_new_conversation=self.is_new_conversation,
),
device_config=embedded_assistant_pb2.DeviceConfig(
device_id=self.device_id,
device_model_id=self.device_model_id,
),
text_query=text_query,
)
# Continue current conversation with later requests.
self.is_new_conversation = False
req = embedded_assistant_pb2.AssistRequest(config=config)
yield req
def iter_assist_requests():
dialog_state_in = embedded_assistant_pb2.DialogStateIn(
language_code=self.language_code, conversation_state=b'')
if self.conversation_state:
dialog_state_in.conversation_state = self.conversation_state
gConfig = embedded_assistant_pb2.AssistConfig(
audio_out_config=embedded_assistant_pb2.AudioOutConfig(
encoding=config.writtenAssist['audio_out_config']
['encoding'],
sample_rate_hertz=config.writtenAssist['audio_out_config']
['sample_rate_hertz'],
volume_percentage=config.writtenAssist['audio_out_config']
['volume_percentage'],
),
dialog_state_in=dialog_state_in,
device_config=embedded_assistant_pb2.DeviceConfig(
device_id=self.device_id,
device_model_id=self.device_model_id,
),
text_query=text_query,
)
req = embedded_assistant_pb2.AssistRequest(config=gConfig)
assistant_helpers.log_assist_request_without_audio(req)
yield req
def iter_assist_requests():
config = embedded_assistant_pb2.AssistConfig(
audio_out_config=embedded_assistant_pb2.AudioOutConfig(
encoding='LINEAR16',
sample_rate_hertz=16000,
volume_percentage=100,
),
dialog_state_in=embedded_assistant_pb2.DialogStateIn(
language_code=r.language,
conversation_state=self.conversation_state,
is_new_conversation=True,
),
device_config=embedded_assistant_pb2.DeviceConfig(
device_id=self.cfg.device_id,
device_model_id=self.cfg.device_model_id,
),
text_query=r.request)
screen_mode = r.screen_mode if r.screen_mode else self.cfg.screen_mode
config.screen_out_config.screen_mode = getattr(
embedded_assistant_pb2.ScreenOutConfig, screen_mode)
self.is_new_conversation = True
req = embedded_assistant_pb2.AssistRequest(config=config)
yield req
def _create_audio_request(self, data):
return embedded_assistant_pb2.AssistRequest(audio_in=data)