dthphuong / baby_cry_rpi

Baby cry detection - Deployment on Raspberry Pi

Recognition of baby cry from audio signal.

The aim is to automatically recognize a baby crying while sleeping. In such case, a lullaby is started to calm the baby down.

This is done by implementing a machine learning algorithm on a Raspberry Pi.

Hardware setup

• Raspberry Pi 2B or 3 (I haven't tried it yet on Zero, the connections being a bit different, it might need some additional setup)
• Usb microphone
• Speakers connected via Jack 3.5 mm

Software setup

This repo should be cloned on the Rapsberry Pi in /opt/.

To make it work:

• Install docker on your Raspberry Pi
• Get the Docker image docker pull mikkl/rpi-raspbian-baby-cry-detection:pi2
• Put model.pkl generated by baby_cry_detection in folder model
• Put the egg file baby_cry_detection-x.x-py2.7.egg generated by baby_cry_detection in folder lib
• Execute script/run.sh from /opt/baby_cry_rpi

For Raspberry Pi 3, replace tag: from pi2 to latest (in script run.sh as well)