Speech Recognition Based Criminal Detection Alarm System for Public Railways Transport Using Linear Predictive Coding and Dynamic Time Warping

Ngu, Kevin (2018) Speech Recognition Based Criminal Detection Alarm System for Public Railways Transport Using Linear Predictive Coding and Dynamic Time Warping. Final Year Project (Bachelor), Tunku Abdul Rahman University College.

Text Kevin Ngu.pdf Restricted to Registered users only Download (3MB)

Abstract

In this paper, a system is proposed to fight against criminal cases occur on public railways transport using speech recognition. The main target of this project is to build a system which can detect emergency speech uttered by the victims on the public railways transport. Then, it will trigger alarms and send a signal to the police stations nearby to help the victims as fast as possible. The speech recognition requires more training as it is needed for the syste, to differentiate emergency speech signals from the normal conversation said by people. The system of the project requires the collaboration of software and hardware to make it works. For software part, programming is needed to extract the speech information and compares it with the database inside the programs. Here, Linear Predictive Coding (LPC) and Dynamic Time Wrapping (DTW) are used to extract the speech information and compare it with the database. On the other hand, for hardware part, a hardware simulation implemented with Arduino Uno microcontroller will be adapted to receive the emergency speech signals from the users and send signals to trigger the alarms. For this project, speech recognition techniques are applied with the suitable application so that this system can contribute to the society. After all, the project is able to help to lower crimes and ensure the safety of the citizens who take the public transport frequently or at late night.