def request_stream(data_stream, rate, interim_results=True):
"""Yields `StreamingRecognizeRequest`s constructed from a recording audio
stream.
Args:
data_stream: A generator that yields raw audio data to send.
rate: The sampling rate in hertz.
interim_results: Whether to return intermediate results, before the
transcription is finalized.
"""
# The initial request must contain metadata about the stream, so the
# server knows how to interpret it.
recognition_config = cloud_speech_pb2.RecognitionConfig(
# There are a bunch of config options you can specify. See
# https://goo.gl/KPZn97 for the full list.
encoding='LINEAR16', # raw 16-bit signed LE samples
sample_rate=rate, # the rate in hertz
# See http://g.co/cloud/speech/docs/languages
# for a list of supported languages.
language_code='en-US', # a BCP-47 language tag
)
streaming_config = cloud_speech_pb2.StreamingRecognitionConfig(
interim_results=interim_results,
config=recognition_config,
)
yield cloud_speech_pb2.StreamingRecognizeRequest(
streaming_config=streaming_config)
for data in data_stream:
# Subsequent requests can all just have the content
yield cloud_speech_pb2.StreamingRecognizeRequest(audio_content=data)
def _create_config_request(self):
recognition_config = cloud_speech.RecognitionConfig(
# There are a bunch of config options you can specify. See
# https://goo.gl/KPZn97 for the full list.
encoding='LINEAR16', # raw 16-bit signed LE samples
sample_rate=AUDIO_SAMPLE_RATE_HZ,
# For a list of supported languages see:
# https://cloud.google.com/speech/docs/languages.
language_code=self.language_code, # a BCP-47 language tag
speech_context=self._get_speech_context(),
)
streaming_config = cloud_speech.StreamingRecognitionConfig(
config=recognition_config,
single_utterance=True, # TODO(rodrigoq): find a way to handle pauses
)
return cloud_speech.StreamingRecognizeRequest(
streaming_config=streaming_config)
def _create_audio_request(self, data):
return cloud_speech.StreamingRecognizeRequest(audio_content=data)
