TranslationWithSpeech

Translation with Speech sample

This sample shows the translation capabilities of the speech to text service, converting speech to text in multiple languages. It also shows the voice capabilities, speaking the Chinese translation.

Setup

Before you can run this sample, you will need to ensure you have your Python development environment set up, and do some setup on this project.

Follow the Project Setup Instructions to get this project set up, including the environment variables, virtual environment and pip packages.

Run the code

1. From the Visual Studio Code terminal, run the following command:

   python app.py

The app will run in the terminal and listen for your voice. Speak into your default microphone in English and you will see the words you have said output to the terminal in Chinese, English, French and German. You will also hear the Chinese translation through your default audio device. Say stop to end the program.

(.venv) ➜  TranslationWithSpeech git:(master) ✗ python app.py
Say something! Say stop when you are done.
Chinese   : 世界您好。
English   : Hello world.
French    : Salut tout le monde.
German    : Hallo Welt.

Chinese   : 停止。
English   : Stop.
French    : Arrêter.
German    : Stoppen.

(.venv) ➜  TranslationWithSpeech git:(master) ✗

Change the detected language

This app is configured to listen for words spoken in English, and output the English text of the spoken words, as well as translations into Chinese, English, French and German. This configuration is set on line 27:

translation_config = speechsdk.translation.SpeechTranslationConfig(subscription=key,
                                                                   region=region,
                                                                   speech_recognition_language='en-GB',
                                                                   target_languages=('zh-Hans', 'en', 'fr', 'de'))

To change the recognized language, change the value of 'en-GB' to a different language identifier, such as 'en-US' for American English, or 'zh-CN' for Chinese. You can find the list of supported languages in the Language and voice support for the Speech service docs on Microsoft Docs.

To change the language that the speech is translated into, change the values of target_languages. When the text is translated, the recognized function is called with a list of translations in a dictionary. You will also need to change the code here to access the languages you have specified.

def recognized(args):
    global stop
    if args.result.reason == speechsdk.ResultReason.TranslatedSpeech:
        print("Chinese   :", args.result.translations['zh-Hans'])
        print("English   :", args.result.translations['en'])
        print("French    :", args.result.translations['fr'])
        print("German    :", args.result.translations['de'])
        print()

        if args.result.translations['en'] == "Stop.":
            stop = True

The language of the spoken audio is set on line 19:

language = "zh-CN"
speech_config.speech_synthesis_language = language

Update the value of language to hear the audio in a different language.