Search code examples
google-app-enginegoogle-cloud-platformgoogle-cloud-dataflowgoogle-speech-api

Google Speech + Natural Language pipeline

I am trying to do the following:

I want to create the following pipeline (that will run on google cloud)

  • I have a device : raspberry pi with microphone
  • I send the audio from raspberry pi to google (streaming: real-time),
  • It is transcribed to text (Google Cloud Speech)
  • The text is analyzed (Google Cloud Natural Language)
  • I do some post processing
  • the result (of the processing) is returned to the Raspberry pi

I know that i could do speech recognition request, then with the result from the API, send it directly to Natural Language, etc

But I would like to know if such a pipeline is possible. I tried with Google DataFlow, and with Google App Engine, but I couldn't find a way to make this kind of pipeline with them.

Thank you

Solution

  • If you have only one device producing voice input, a simple solution such as this should be sufficient(replace image with audio and translation/vision api with speech/natural language apis):

    enter image description here

    Cloud storage can trigger cloud functions and in turn cloud function can use cloud pub/sub to trigger other cloud functions.

    See this link for details.