def __init__(self,
credential: Optional[Union[str, Path]] = None) -> NoReturn:
if credential is None:
self.client = TextToSpeechClient()
else:
self.client = TextToSpeechClient.from_service_account_file(
filename=credential)
def text_to_mp3(client: texttospeech.TextToSpeechClient,
voice: texttospeech.types.VoiceSelectionParams,
audio_config: texttospeech.types.AudioConfig, text: str,
output_file_path: Path) -> None:
"""
Convert a string into voice and save it in an .mp3 file.
:param client: TextToSpeechClient instance.
:param voice: VoiceSelectionParams instance.
:param audio_config: AudioConfig instance.
:param text: String to synthesise.
:param output_file_path: Full path to the output .mp3 file.
:return: None
"""
lines = text.splitlines()
logger.info(f'Synthesising {len(lines)} lines ...')
output_file_log = output_file_path.parent / (output_file_path.stem +
'_log.json')
with open(output_file_path, 'wb') as output_file:
for (i, text_chunk) in enumerate(lines):
# skip empty lines
if len(text_chunk) > 0:
input_text = texttospeech.types.SynthesisInput(text=text_chunk)
try:
logger.info(
f'Synthesising speech for chunk `{i}`, size: `{len(text_chunk)}`'
)
response = client.synthesize_speech(
input_text, voice, audio_config)
except Exception as e:
# If a line could not be synthesised properly, return it along with the error message
# It is possible that textract could not extract the text properly.
logger.error(
f'Speech synthesising failed! Chunk text: `{input_text}`\nError: {e}\n'
)
_error_log = {
'chunk_number': i,
'chunk_length': len(text_chunk),
'chunk_text': str(text_chunk),
'Error message': traceback.format_exc()
}
with open(f'{output_file_log}', 'w') as log_out:
json.dump(_error_log, log_out)
continue
output_file.write(response.audio_content)
logger.info(f'Audio content written to `{output_file_path}`!')
logger.info(f'Output saved to `{output_file_path}`')
logger.info(f'logs at `{output_file_log}`')
class TextToSpeech:
def __init__(self,
credential: Optional[Union[str, Path]] = None) -> NoReturn:
if credential is None:
self.client = TextToSpeechClient()
else:
self.client = TextToSpeechClient.from_service_account_file(
filename=credential)
def synthesize(
self,
text: str,
language: str = 'en-US',
gender: int = 1,
encoding: enums.AudioEncoding = enums.AudioEncoding.MP3
) -> types.SynthesizeSpeechResponse:
"""
Args:
text:
language:
gender:
encoding:
Returns:
"""
if gender == 0:
ssml_gender = enums.SsmlVoiceGender.FEMALE
elif gender == 1:
ssml_gender = enums.SsmlVoiceGender.MALE
else:
ssml_gender = enums.SsmlVoiceGender.NEUTRAL
synthesis_data = types.SynthesisInput(text=text)
voice = types.VoiceSelectionParams(language_code=language,
ssml_gender=ssml_gender)
audio_config = types.AudioConfig(audio_encoding=encoding)
return self.client.synthesize_speech(input_=synthesis_data,
voice=voice,
audio_config=audio_config)
@staticmethod
def save(response: types.SynthesizeSpeechResponse,
filename: Union[str, Path]) -> NoReturn:
"""
Args:
response:
filename:
Returns:
"""
with open(filename, 'wb') as audio_file:
audio_file.write(response.audio_content)
def main(text, locale='en-US'):
client = TextToSpeechClient.from_service_account_file(str(SERVICE_FILE))
input_text = types.SynthesisInput(text=text)
# Note: the voice can also be specified by name.
# Names of voices can be retrieved with client.list_voices().
voice = types.VoiceSelectionParams(
language_code=locale,
# ssml_gender=enums.SsmlVoiceGender.FEMALE,
name="en-US-Wavenet-A")
audio_config = types.AudioConfig(audio_encoding=enums.AudioEncoding.MP3)
response = client.synthesize_speech(input_text, voice, audio_config)
# The response's audio_content is binary.
with open('output.mp3', 'wb') as out:
out.write(response.audio_content)
print('Audio content written to file "output.mp3"')