Pervasive Systems Project - Big Ear 2.0

The repository contains the code about the system developed for the pervasive systems course (Politecnico di Milano). It includes the code running on Raspberry pi, the RESTful server and a simple web application.

Usage

• Raspberry Pi
  • Before everything, follow this guide in order to install everything is needed: ReSpeaker wiki
  • Activate virtual environment create in the previous step
  • Inside code execute kws_doa.py

pi@raspberrypi:~ $ source ~/env/bin/activate
(env) pi@raspberrypi:~ $ cd ~/code
(env) pi@raspberrypi:~/code $ python kws_doa.py 

• REST Server

  • Inside rest-server execute server.py

• Web app

  • Read the README inside the web-app folder
  • Execute the web application while the REST server is running

• Localization v2

  • Inside the localization-v2 folder
  • Jupyter notebook source_localization.ipynb where localization is performed

Code Structure

The code uses gstreamer-like elements which can be linked together as an audio pipeline.

The topology is:

Source --> ChannelPicker --> KWS
  |                           |
  V                           V
 DOA           Google Speech Recognition Service