imclab / translator

Translator.js is a JavaScript library built top on Google Speech-Recognition & Translation API to transcript and translate voice and text. It supports many locales and brings globalization in WebRTC! https://www.webrtc-experiment.com/Translator/

Translator.js / Demo

WebRTC Voice & Text Translator

Translator.js is a JavaScript library built top on Google Speech-Recognition & Translation API to transcript and translate voice and text. It supports many locales and brings globalization in WebRTC!

How to use?

<script src="//cdn.webrtc-experiment.com/Translator.js"> </script>

Facing issues using in your own domain?

• muaz-khan#1 (comment)

API Reference

var translator = new Translator();

translateLanguage

Pass English language text, and convert into Arabic or into any other supported language.

This method simply converts text from one language into another.

translator.translateLanguage(textToConvert, {
    from: 'language-of-the-text',
    to: 'convert-into',
    api_key: 'AIzaSyCUmCjvKRb-kOYrnoL2xaXb8I-_JJeKpf0', // use your own key
    callback: function (translatedText) {
        console.log('translated text', translatedText);
        
        // here you can use "speakTextUsingRobot"
        // see below sections
    }
});

translateLanguage method takes two arguments:

1. Text to convert
2. Source and Target Languages; also callback method

Second argument is an object.

What you can do is either display translated text in chat-box; or use speakTextUsingRobot or speakTextUsingGoogleSpeaker methods to play voice.

voiceToText

This method allows you convert voice into text; whatever you speak is recognized using Google Speech-Recognition API; and converted into text using same API:

translator.voiceToText(function (text) {
    console.log('Your voice as text!', text);
    
    // here you can use "translateLanguage" method
    // translator.translateLanguage
    // see more info in above section
}, 'en-US');

voiceToText method takes two arguments:

1. Callback; used to return recognized text
2. Language; used to suggest Speech-Recognition API to easily recognize speaker's language

If language is not passed; then en-US will be used as default language.

You can use socket.io, websockets or any other signaling gateway like WebRTC data channels to exchange/share transcripted text with other users.

speakTextUsingRobot

Use a javascript file to speak the text.

This method uses meSpeak.js library to play text using a robot voice. Behind the scene; text is buffered and converted into WAV file; which is played using invisible <audio> element.

translator.speakTextUsingRobot(textToPlay, {
    workerPath: '//cdn.webrtc-experiment.com/Robot-Speaker.js',
    callback: function (WAV_File) {},
    amplitude: 100,
    wordgap: 0,
    pitch: 50,
    speed: 175,
    onSpeakingEnd: function() {},
    onWorkerFileDownloadStart: function() {},
    onWorkerFileDownloadEnd: function() {}
});

speakTextUsingRobot method accepts two arguments; first one is mandatory and last one is optional:

1. Text to Speak i.e. convert text into voice file (WAV)
2. Options like worker-file path etc.

Default worker file's path is https://www.webrtc-experiment.com/Robot-Speaker.js. It is strongly recommended to download and link it from your own domain.

If you want to play WAV file yourself or you want to POST/store WAV file; then you can use callback parameter to override default behaviour:

translator.speakTextUsingRobot(textToPlay, {
    callback: function (WAV_File) {
        HTTP_POST_using_FormData( WAV_File );
    }
});

onSpeakingEnd is useful in text-chat apps; where you can disable text box until text is translated and spoken; then you can enable it again.

speakTextUsingGoogleSpeaker

Use Google Servers (Translation API) to speak the text.

This method uses Google Non-Official Translation API to convert text into mp3 sound. API Key used is taken from someone on the web; and there is no guarantee of its availability; that's why it is strongly suggested to buy your own KEY and pass using api_key parameter:

translator.speakTextUsingGoogleSpeaker({
    textToSpeak: 'text-to-convert',
    targetLanguage: 'your-language',

    // Google Translation service's API Key
    api_key: 'Your-Private-API-Key'
});

Both textToSpeak and targetLanguage are mandatory. Only api_key is optional.

Use Cases

1. You can use it in any WebRTC application to support globalization!
2. You can use WebRTC data channels to share transcripted text among users
3. You can mute original voice; and play translated one. Though, voice and video will NOT be sync

It is really useful in text-chat apps!

Demo

• https://www.webrtc-experiment.com/Translator/

Languages Supported

43+ languages are supported. Please check above demo link for the list.

License

Translator.js is released under MIT licence . Copyright (c) Muaz Khan.