Audio Analyser

This program records audio samples and sends them through a machine learning algorithm which classifies the sounds. The results are send through MQTT to any home automation platform.

Part list

(some are referral links, thanks to DigitSpace for sending me some of the parts)

• Raspberry Pi
• Raspberry Pi heatsink
• USB Microphone

Run this code

• Make sure the machine learning model is running in another terminal via this guide: https://github.com/IBM/MAX-Audio-Classifier
• Run npm install
• Run node main.js

Raspberry Pi

• Install the machine learning model
  • Install Docker
  • Install NodeJS
  • Install Git
• Run the Machine learning model using a custom build
  • Clone MAX Audio classifier repo https://github.com/IBM/MAX-Audio-Classifier
  • Run cd to go into the folder
  • Change to the following branch: git checkout xuhdev-patch-1
  • Build it using docker build -f Dockerfile.arm32v7 -t max-audio-classifier .
  • Run docker run -it -p 5000:5000 max-audio-classifier to start the model
• When the model is running open a second terminal, ssh into the pi and run the code in this repo like described above

Temperature check on the Raspberry Pi

Run this npm run rpi-temp to check the temperate. It can get kind of hot.