def transcribe_with_word_time_offsets(
speech_content: bytes, ) -> Iterable[Tuple[str, float, float]]:
"""Recognize words with time offsets from a speech.
Args:
speech_content: Binary data of the speech.
Yields:
The word with start time and end time that api recognized.
[
('여기요', 0.0, 2.0),
('저기요', 3.6, 5.4),
('저', 5.4, 9.2),
('밖에서', 9.2, 9.6),
('장애인', 9.6, 10.0),
('주차', 10.0, 10.3),
('가능', 10.3, 10.5),
('까만색', 10.5, 11.3),
('소나타', 11.3, 11.7),
('글', 11.7, 11.8),
('찾아요', 11.8, 12.2),
('근데요', 12.2, 13.2)
]
See:
https://cloud.google.com/speech-to-text/docs/sync-recognize
"""
client = SpeechClient()
audio = types.RecognitionAudio(content=speech_content)
config = types.RecognitionConfig(
encoding=enums.RecognitionConfig.AudioEncoding.LINEAR16,
sample_rate_hertz=44100,
language_code="ko-KR",
enable_word_time_offsets=True,
)
response = client.recognize(config, audio)
for result in response.results:
alternative = result.alternatives[0]
for word_info in alternative.words:
word = word_info.word
start_time = word_info.start_time
end_time = word_info.end_time
yield (
word,
start_time.seconds + start_time.nanos * 1e-9,
end_time.seconds + end_time.nanos * 1e-9,
)
def __init__(self, credentials_path, language_code, phrase_hints=[]):
"""
Args:
credentials_path (str): The path to the service account private key json file.
language_code (str): The language of the supplied audio as a BCP-47 language tag. Example: “en-US”.
phrase_hints (str[]): https://cloud.google.com/speech-to-text/docs/basics#phrase-hints
"""
self.language_code = language_code
self.client = SpeechClient().from_service_account_json(
credentials_path)
self.speech_context = [types.SpeechContext(phrases=phrase_hints)]
self._mic: MicrophoneInput = None
def get_conn(self):
"""
Retrieves connection to Cloud Speech.
:return: Google Cloud Speech client object.
:rtype: google.cloud.speech_v1.SpeechClient
"""
if not self._client:
self._client = SpeechClient(credentials=self._get_credentials())
return self._client
def _make_one():
import google.auth.credentials
from google.cloud.speech_v1 import SpeechClient
credentials = mock.Mock(spec=google.auth.credentials.Credentials)
return SpeechClient(credentials=credentials)
def make_speech_client():
credentials = mock.Mock(spec=google.auth.credentials.Credentials)
return SpeechClient(credentials=credentials)
class SpeechToTextClient:
"""
A speech to text client that parses microphone input into text.
"""
def __init__(self, credentials_path, language_code, phrase_hints=[]):
"""
Args:
credentials_path (str): The path to the service account private key json file.
language_code (str): The language of the supplied audio as a BCP-47 language tag. Example: “en-US”.
phrase_hints (str[]): https://cloud.google.com/speech-to-text/docs/basics#phrase-hints
"""
self.language_code = language_code
self.client = SpeechClient().from_service_account_json(
credentials_path)
self.speech_context = [types.SpeechContext(phrases=phrase_hints)]
self._mic: MicrophoneInput = None
def start(self, callback):
"""
Args:
callback (function): Function that is called when text is transcribed from speech
"""
try:
with MicrophoneInput() as mic:
print("Starting SpeechToTextClient")
self._mic = mic
audio_generator = self._mic.generator()
config = types.RecognitionConfig(
encoding=enums.RecognitionConfig.AudioEncoding.LINEAR16,
sample_rate_hertz=self._mic.RATE,
language_code=self.language_code,
use_enhanced=True,
speech_contexts=self.speech_context)
streaming_config = types.StreamingRecognitionConfig(
config=config, interim_results=True)
requests = (types.StreamingRecognizeRequest(
audio_content=content) for content in audio_generator)
responses = self.client.streaming_recognize(
streaming_config, requests)
for response in responses:
if not response.results: # no results
continue
# first result is best result
result = response.results[0]
if not result.alternatives:
continue
transcript = result.alternatives[0].transcript.strip(
).lower()
callback((transcript, result.is_final))
except OutOfRange:
self.restart(callback)
def stop(self):
print("Stopping SpeechToTextClient")
if self._mic is None or self._mic.closed:
return
self._mic.close()
def restart(self, callback):
self.stop()
self.start(callback)
def update_phrase_hints(self, phrase_hints):
self.phrase_hints = [types.SpeechContext(phrases=phrase_hints)]