Azure Text-to-Speech Service

 

Picture from Alex Knight @ https://www.pexels.com/

After blogging about Azure Speech-to-Text Service, here we have the Text-to-Speech blog. The setup is identical to the Speech-to-Text Service blog. and here is the Python code.

import azure.cognitiveservices.speech as speech_sdk
import os

voices = [
    "en-GB-ThomasNeural",  # English (United Kingdom)
    "en-SG-WayneNeural",  # English (Singapore)
    "en-US-MonicaNeural",  # English (USA)
    "en-IN-NeerjaNeural",  # Engish (India)
]

text = "To empower every person and every organization on the planet to achieve more."

if __name__ == "__main__":
    speech_key, service_region = (
        os.environ["AZURE_SPEECH_KEY"],
        os.environ["AZURE_SPEECH_LOC"],
    )
    speech_config = speech_sdk.SpeechConfig(
        subscription=speech_key, region=service_region)
    audio_config = speech_sdk.audio.AudioOutputConfig(
        use_default_speaker=True)

    for voice in voices:
        # https://learn.microsoft.com/en-us/azure/cognitive-services/speech-service/language-support?tabs=stt-tts
        speech_config.speech_synthesis_voice_name = voice
        speech_synthesizer = speech_sdk.SpeechSynthesizer(
            speech_config=speech_config, audio_config=audio_config
        )

        speech_synthesis_result = speech_synthesizer.speak_text_async(text).get()

        if speech_synthesis_result.reason == speech_sdk.ResultReason.Canceled:
            cancellation_details = speech_synthesis_result.cancellation_details
            raise Error("Speech synthesis canceled: {}".format(cancellation_details.reason))

In the code, we can see that we have 4 voices (UK, Singapore, USA, India) for Microsoft's mission statement.