Project Name : Voice Lie Detection System

---

SECTION 1 : EXECUTIVE SUMMARY / PAPER ABSTRACT

---

Our project develops an advanced speech recognition-based polygraph system with the aim of providing an innovative alternative to the traditional polygraph for criminal investigations and other areas. The traditional methods, for example the polygraph, have been the subject of criticism on ethical grounds and with regard to their reliability. In view of this, we propose a voice-based polygraph model that uses machine learning to enhance detection accuracy. This provides a more transparent, non-intrusive and adaptable solution for detection.

The project addresses the shortcomings of existing polygraph techniques, particularly in regard to data transparency, model interpretability and cross-domain applicability. They are achieved by integrating a range of machine learning models (including Random Forests, Support Vector Machines, KNN and others) and we utilise the soft-voting integration methods to enhance the reliability and accuracy of the predictions.

From a commercial perspective, the project's voice lie detector system has the potential for a wide range of applications in multiple fields, including criminal justice, corporate censorship and insurance claims. Our system can be offered on a per-use or subscription basis through a software-as-a-service (SaaS) cloud platform model, making it suitable for a variety of users, including law enforcement agencies, healthcare organisations, insurance companies, and corporate users. The system has been developed with the objective of meeting the specific needs of a range of enterprises. It could assist customers in making efficient judgments in different scenarios such as employee selection, internal vetting, and fraud detection.

SECTION 2 : CREDITS / PROJECT CONTRIBUTION

Official Full Name	Student ID (MTech Applicable)	Work Items (Who Did What)	Email (Optional)
Mohan Liu	A0297443U	1. The process of cleaning data, extracting features, training models (such as KNN and SVM), and evaluating models. 2. Develop integrated models and produce soft voting algorithms to integrate the most accurate algorithms, as well as prepare implementation documents. 3. Contribute to the preparation(write and revise) of reports and the design of their layout. 4. Delegate tasks and monitor progress.	[email protected]
Yuhao Zhou	A1234567B	1. Literature review- Web application front-end development 2. Part of web app back-end development 3. Server transmission testing 4. Part of report writing 5.Demo recording	[email protected]
LiXin Zhang	A0279544N	1. Participate in system design discussions and draw system architecture diagrams 2. Participate in model design and write reports on model training part 3. Participate in report integration 4. Produce PowerPoint and video for system design section	[email protected]
Zhiyuan Zhang	A0297736J	1. project reproduction, project Intro, data collection 2. model training 3.related report writing	[email protected]
Wenyu Zhong	A0294636R	1.web application backend development 2. Writing Report 3.PPT creation	[email protected]

---

SECTION 3 : VIDEO OF SYSTEM MODELLING & USE CASE DEMO

.mp4)]
[.mp4)]

---

SECTION 4 : USER GUIDE

Refer to appendix <Installation & User Guide> in project report at Github Folder: ProjectReport

Make sure all developer tools have been installed:

• npm
• Python3
• pip

[ 1 ] To run the back-end server:

$ cd SystemCode/backend
$ pip install -r requirements.txt
$ cd myproject
$ python manage.py makemigrations api
$ python manage.py makemigrations
$ python manage.py migrate
$ python manage.py runserver

[ 2 ] To run the front-end server:

$ cd SystemCode/frontend
$ npm install
$ npm run dev

Go to URL using web browser http://127.0.0.1:4000

SECTION 5 : PROJECT REPORT / PAPER

PROJECT REPORT 1

1. Executive Summary 3
   
2. Introduction 4
   2.1 Project Background 4
   2.2 Project Significance 5
   2.2.1 Current limitations 5
   2.2.2 Proposed Solution and Innovations 6
   2.3 Project Content 8
   2.3.1 Project Scope 8
   2.3.1 Project Challenges 9
   2.4 Business Plan 10
   2.4.1 Business Value 10
   2.4.2 Company Vision and Mission 10
   2.4.3 Products and Services 11
   2.4.4 Market Analysis 11
   2.4.5 Business Model 12
   2.5 Project Objective 12
   2.5.1 Base Line Setting 12
   2.5.1 Web Application Setting 13
   
3. Literature Review 14
   3.1 Relevant Research 14
   3.2 Key Findings 16
   3.3 Methodologies 16
   
4. System Design 17
   4.1 Architecture Overview 17
   4.2 System Components 18
   4.2.1 User Interface Module 19
   4.2.2 Data Processing Module 19
   4.2.3 Knowledge base Module 20
   4.2.4 Reasoning engine Module 21
   4.2.5 Explainability Module 22
   4.2.6 External interface Module 23
   4.3 Reasoning Techniques and Algorithms 23
   
5. Data Collection and Preparation 25
   5.1 Data Sources 25
   5.2 Challenges in Data Collection 26
   5.3 Preprocessing Techniques 27
   5.3.1 Video Data Retrieval Process 27
   5.3.2 Video Data Convert Process 28
   5.3.3 Feature Extracted 29
   
6. Implementation 31
   6.1 Platform and Tools 31
   6.2 Methods and Technologies 32
   6.2.1 Random Forest 32
   6.2.2 Support Vector Machine 34
   6.2.3 K-Nearest Neighbors 35
   6.2.4 Deep Neural Networks 37
   
7. Results and Progress 38
   7.1Preliminary Results from Reasoning Engine 38
   7.2 Performance Metrics Visualizations 38
   7.2.1 Soft Voting Integration Model_A 38
   7.2.2 Soft Voting Integration Model_B 40 7.3 Comparison with Existing Lie Detection Techniques 41
   8 Web Application Development 43
   8.1 Initiation 43
   8.2 Front-End Development 43
   8.2.1 Tools 43
   8.2.2Application Design 44
   8.3 Back-End Development 44
   8.3.1 System Architecture 44
   8.3.2 Database Design 45
   8.3.3 Api Design 45
   
8. Challenges and Future Work 48
   9.1 Obstacles in System Development 48
   9.1.1 Limitation of Data and Features 48
   9.1.2 Problem for the accuracy and generation of reasoning engine 48
   9.2 Strategies to Overcome Identified Challenges 49
   9.2.1 Methods to solve limitation of Data and Features 49
   9.2.2 Methods to solve problem for the accuracy and generation 49
   9.3 Additional Features to be Implemented 49
   References 51