Final Year Project
Open Source Expo by IEEE Student
IEEE Student Chapter launch
Speech Processing in FPGA with C-to-RTL Compiler Technology
In most cases, speech processing algorithms are developed on off-the-shelf microprocessor or DSPs. Recently, the modern FPGA offers virtually unlimited computational resources that can be executed in parallel. Hence, FPGA outperform the DSP in terms of throughput and costs. The design process of FPGA is always considered time consuming and tedious progression. This project explored the technology that shorten the gap between embedded software and hardware, in particular, speech processing in FPGA with C-to-RTL Compiler. Furthermore, issues on the migration and the co-design between software and hardware were also investigated.
Low cost implementation of a Unmanned Aerial Vehicle
Unmanned Aerial Vehicle (UAV) is defined as an aircraft which is capable of flying autonomously without any human intervention. UAVs are no longer serving only military purpose, but also civil application including remote environment research, border monitoring and oceanography . The purpose of the project is to develop low cost an autonomous aerial vehicle which is capable of waypoint navigation. This project includes the study and implementation of the technique of attitude control, design of PID controller, demonstration of Hardware-In-the-Loop simulation and some other features.
Development of Real Time Speaker Recognition System using Matlab
The project developed a speaker recognition system that identifies a speaker’s identity based on the speaker’s voice. These systems have been in use for around twenty years and this application would be useful in telephone-based applications, where there is no way to identify a user based on fingerprint or face. Like fingerprints, a person’s voice has particular unique features and by using this voiceprint, their identity can be authenticated. The main components of the system are the feature extraction, speaker modeling, and pattern matching. Feature extraction extracts unique features of one’s voice and pattern matching will determine if the features extracted match with previously stored features. In this project, feature extractor and pattern matcher are implemented using Matlab. The feature extraction is done by using Mel-frequency Cepstrum Coefficients (MFCC) and the speakers are modeled using Gaussian Mixture Model (GMM). GMM models the probability density function of observed variables using a multivariate Gaussian mixture density. The Gaussian mixture speaker model attains 93% identification accuracy using 15 seconds clean speeches with a 15 speaker population.
Design and Implementation of Artificial Neural Networks in Semi-autonomous Robotic System
The project aims to design and implement a semi-autonomous robotic system that operates based on Artificial Neural Networks (ANN). The ANN is a computational technique which models the way biological neurons work. Application of embedded systems technology such as microcontroller or programmable logic device (PLD) is necessary to control the system. Through wireless communication between the embedded system and a general-purpose computer, the system can be further empowered. The system is a prototype that serves as a foundation for further development for many applications including space exploration (e.g. on Moon) and services (e.g. intelligent wheel chairs and vacuum cleaners).