Home Secure
The monitoring of dangerous audio events is very important in surveillance systems. We have created a machine learning model on glass breaking and baby crying detection and deployed it on flask.
Created on 24th August 2019
•
Home Secure
The monitoring of dangerous audio events is very important in surveillance systems. We have created a machine learning model on glass breaking and baby crying detection and deployed it on flask.
The problem Home Secure solves
The monitoring of dangerous audio events is very important in surveillance systems.We will make an Android APP using highly predictive and classified SVM model by using MFCC for audio features extraction which will detect the impulsive sound such as a breaking of a glass. It can be used as a safety device such as burglary by sending a notification to client and sounding an alarm.Also it can detect the crying sound of a baby and it will sound an alarm to the attendee when baby cries.
Challenges we ran into
The biggest problem we faced was in collecting the audio samples of glass breaking and baby crying. We collected 1300 audio samples of glass breaking, baby crying and indoor noise. Next we had to find the best audio features to be used from a wide range of audio features like mfcc, spectral contrast, tonnetz, chroma etc.
Technologies used