A spoken-word translation tool that leverages Google's speech recognition and translation APIs.

Translator currently depends on apple's built-in say command to vocalize results.

You will need to install the Google Cloud SDK on your system to use the translation and speech parsing APIs - visit https://cloud.google.com/sdk/docs/ for more information.

Once you've installed the SDK, you'll need to create a set of application credentials. Sadly the translation API aint free, but as of this writing, Google offers $300 in free credit on a trial account. So... still pretty free?

Run the following in your terminal:

• gcloud auth application-default login

Next, via the Google Cloud dashboard (https://console.cloud.google.com/home/dashboard), you will need to do the following:

• create a Google Cloud project and obtain a projectID. Enable Billing.
• enable the Google Cloud Translation API and obtain an API key via the "Credentials" menu.
• enable the Google Cloud Speech API.
• create a service account via the "Credentials" menu and generate/download a JSON keyFile for it. Save the keyFile to the project root directory as _keyFile.json. If you do not use a keyFile, the speech streaming will still work, but the API will be unbearably slow to respond.

• brew install sox, and add it to your PATH
• npm install
• copy and rename ENV_template.js to ENV.js, fill in any values you obtained in the Requirements section above.

• Select your input and output langages and expectANovel setting in config.js.
• run npm start, wait for "Translator is listening..." in the terminal, and start sayin stuff!

• For best results, go somewhere quiet.
• If you are in a quiet space, the app should stop listening to you after 2 seconds of silence. If you're somewhere noisey, it will keep listening until 15 seconds have gone by or until you shut it down (ctrl c).
• The API cuts you off after 60 seconds - thats your maximum talk time per run.
• If the mic is too sensetive / not sensetive enough, adjust voiceThreshold in constants.js.

• Pause occasionally while speaking. Your input will be chunked more often and you will receive results more quickly and continuosly.
• Use headphones (or a separate microphone) to prevent the app from attempting to retranslate its own speech from the speaker output.
• If you are in a quiet space, the app should stop listening to you after 5 seconds of silence. If you are somewhere noisey, you will have to shut it down manually when you are done speeking (ctrl c).
• 60 seconds is still your maximum talk time, as imposed by the Google Speech api. (I guess its really more of a short story).

• What is this? Error: Audio data is being streamed too slow. Please stream audio data approximately at real time.
  • I haven't found a good answer yet, but it comes from the Google Speech API. Its apparently a common error for other apps that consume it as well. Just try running your translation again.