VOICE AUTHENTICATION BASED MOBILE ROBOT FOR DISABLED PEOPLE

The security of homes and other important properties is very important. Classic security mechanisms come with a variety of inconveniences including the inability to fit a key into the door lock or unlock when one?s hands are full, keys often are lost as well. Based on these reasons, robot is integra

Project Title

VOICE AUTHENTICATION BASED MOBILE ROBOT FOR DISABLED PEOPLE

Project Area of Specialization

Artificial Intelligence

Project Summary

The security of homes and other important properties is very important. Classic security mechanisms come with a variety of inconveniences including the inability to fit a key into the door lock or unlock when one’s hands are full, keys often are lost as well. Based on these reasons, robot is integrated with a computer system that can move forword, backward, left and right using voice recognition (Voice biometrics). Voice recognition is able to identify a person through his voice. Every human being voice is unique it has distinctive features that are identifying you commercial systems already use voice biometrics to identify the speaker In such systems a statistical model of human voice is recorded in a database to be authenticated when person speaks his voice is compared to the already available database to validate the speaker. Voice quality is affected by environment, speaker health and devices. Hence Crims technology used mathematical factor, joint factor analysis to solve this problem it separates the characteristics of the speaker voice, and this technology helps you to validate the speaker without the consideration of the conditions. There is also used message based service to text the owner of door when the door is being opened, this helps to keep security further strong. The design of this tool is made using Raspberry pi 4 as the processing center and ULN2803 as IC to increase the voltage so that it can move the solenoid that serves to move the robot. Then raspberry gives command to the servo motor to move the robot. Recurrent neural networks (RNN) model are used to recognize the person.

Project Objectives

• To make voice authentication based mobile controlled robot
• To Make voice based dataset for Urdu language
• To remove noise signals from the voice data
• To use machine learning for voice recognition
• To make security system much stronger
• To interface different hardware components with raspberry pi 4
• To recognize the speaker based on voice biometrics

Project Implementation Method

1. Data Collection:

Data will be recorded from speakers in the university office using IPhone and USB based microphone. Before the start of each session, each speaker will be communicated on how to record their voices. There will be almost 20 samples recorded from each speaker, total speaker will be 6 who will have authentication for door lock. 6 persons each uttering 20 samples it makes a database of 120 samples. All the recorded samples will be then transferred to a laptop and converted to .wav from the default format of the allocated devices using audio converter (4dots software) for further processing.

1. Data acquisition and pre-processing:

In this stage, data is acquired by the use of analogue microphone using some recommended properties such as; Sound ID (to represent the speech signal), Duration of recording in seconds (3 seconds recommended), Sampling frequency (22050 Hz recommended) and Number of bits per sample (16 bits recommended)

1. Feature extraction and Data storage:

After pre-processing, some features of the vocal characteristics of the speech are extracted from the speech signal and the speech in form of SOUND ID is stored in the speech system database. The Mel Frequency Cepstral Coefficients (MFCC) feature extraction technique is used for the extraction of features

1. Speaker recognition using RNN model

The decision task for the recognition of speakers from the combined frameworks of RNN and the GMM is based on finding the correlation coefficient measure. This evaluates the goodness of match by comparing the recognized feature of the speech frames of the detected speaker from the RNN and GMM to measure the degree of similarity

1. Interfacing hardware using raspberry pi 4

In this part different hardware components will be used to implement the project on hardware level. In this we will be using raspberry pi as a central controller, power supply and relay switch for servo motors to move and stop solenoid motors. To record the voice USB based microphone will be used.

Benefits of the Project

For several years, researchers have developed methods and techniques that al-low recognizing people through their speech. Since voice is produced through the convolution of an excitation and the impulsive response of the vocal tract model, it is sometimes useful to isolate one of the two components for subsequent digital treatment.

Biometric authentication, unlike passwords or token-based authentication, uses unique biological characteristics to verify an individual’s identity. It’s harder to spoof and generally more convenient for users since they don’t have to remember passwords or carry a physical token that can easily be lost or stolen. The authenticator is part of the individual. Voice for every indivudual is different so they it will be much better to authorize home assets based on individual voice. Research shows that existing security systems have some drawbacks. For example

• Facial recognition based security system is affected by change in lighting
• fingerprint based door lock security system requires high resolution scanner
• and Password Based door lock security systems have the inability to change password during power failure 
• As for key lock based door security keys are often lost or forgotten.

Technical Details of Final Deliverable

Our proposed system aim at a robotic vehicle operated by human speech commands. The system operates with the use of an android device which transmits voice commands to raspberry pi to achieve this functionality. The transmitter consists of the android phone Bluetooth device. The voice commands recognized by the module are transmitted through the Bluetooth transmitter. These commands are given to RNN model to classify the authorized speaker being detected to control robotic vehicle in order to move it in left, right, backward and front directions. The Bluetooth receiver mounted on raspberry pi is used to recognize the transmitted commands and decode them. The controller then drives the vehicle motors to move it accordingly. This is done with the use of a driver IC used to control the motor movements. The Bluetooth technology used to transmit and receive data allows for remotely operating the system within a good range.

Final Deliverable of the Project

HW/SW integrated system

Core Industry

IT

Other Industries

Medical

Core Technology

Artificial Intelligence(AI)

Other Technologies

Robotics

Sustainable Development Goals

Good Health and Well-Being for People, Industry, Innovation and Infrastructure, Sustainable Cities and Communities

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com