The pathological voice problems are due to the functional disorder of the vocal system. Laryngeal pathologies include nodules of vocal folds, polyps, ulcers, carcinomas and the paralysis of the laryngeal nerve. More than 25% of the working population requires essential voice usage and hence they are subjected to the risk of these types of voice disorders. Suitable non-invasive systems for early detection and treatment of voice disorders are essentially required for the social betterment.

Due to unhealthy social habits and voice abuses, there is an increase in the rate of voice diseases. The diseases on the vocal folds restrict the movement of the vocal folds and cause more noise and reduction of the regularity of the pitch frequencies. The presence of pathologies in the vocal folds cause significant changes in the normal vibration patterns and results hoarseness and subsequently reduces the voice quality. Thus the acoustical analysis on the voice signal can be used to classify the pathological voices. The aim of this project is to develop simple feature extraction algorithms for extracting the salient features from the voice signals and neural network algorithms for classification of pathological voices.

Automatic Speech recognition (ASR) is the process in which an acoustic signal, captured by a microphone, is converted to a set of words by means of a computer program. The objective of this research is to develop novel and improved feature extracting algorithms for extracting the features from the speech signals recorded from Malaysian Speakers. A simple supervised and unsupervised neural network models will be develop for the classification pathological voices. Finally, a Malay Word Speller will be develop that can spell any Bahasa Malaysia words based on vowel location.

Information and knowledge are expanding in quantity and accessibility. However, people with functional limitations, such as deaf people, often left out of conversation where there are wide communication gaps between them with the ordinary people. The sign language is the fundamental communication method between people who suffer from hearing defects. In order for an ordinary people to communicate with deaf people, a translator is usually needed to translate sign language into natural language. This project presents a simple method for converting sign language into voice signal using features obtained from hand and head gestures. Using a camera, the system receive sign language video from the deaf people in the form of video streams in RGB (red-green-blue) colour with a screen bit depth of 24-bits and a resolution of 320 x 240 pixels. For each frame of images, head and two hand regions are segmented and then converted into binary image. Feature extraction model is then applied on each of segmented image to get the most important feature from the image. Artificial Neural Network (ANN) provides alternative form of computing that attempts to mimic the functionality of the brain. A simple neural network model is developed for sign recognition from the features computed from the video stream. An audio system is installed to play the particular word for the communication between the ordinary people and deaf people.

The current practice of motorcycle engine diagnosis and troubleshooting faults is through computer based system. The trainees are required to understand various engine problems and be able to detect, diagnose and troubleshoot the problems based on sound generated by the engine. Recently existing systems require the expertise of motorbike mechanic. In order to solve such problems, an expert system for motorbike engine diagnosis has been developed to help the motorbike mechanic for early diagnosis of engine faults from the engine sound. Using the powerful microphone, the noise signals from motorbike engine are recorded. The recorded noise signals are analyzed by the extraction of acoustic features by means of digital processing techniques. A neural network model is developed for the classification of faults present in the motorbike by using the acoustical parameters extracted from the noise signals. The NN model is embedded into a processor for easy diagnosis.

To circumvent the damage due to natural and manmade causes in the steel structures is cumbersome. The Detection of the Damage and its localization becomes more difficult with the conventional methods, which are already costly and require expert mechanisms. The convention of NDT (Non Destructive Testing) for damage detection is widely used in the recent researches. The employment of vibration signals in non destructive testing has an extensive usage in the modern detection methods. The objective of this research is to detect the damage present in the structural steel using non destructive vibration testing. The vibration signals are captured in both undamaged and damaged circumstances during an external impact force applied using the impact hammer. The accelerometers placed in different locations of the steel plate are used to capture the vibration signals. Features from the captured vibration signals both during damaged and undamaged situations are extracted. A suitable neural network algorithm is developed for the classification of damages from the steel structure.

Classroom is a place where an individual is educated. When teachers communicate with or give instructions to the students in the classroom, it is important that the messages can be passed effectively and clearly between them. Speech intelligibility is then of prime importance to the outcome of teaching, which affect the future development of the students. Speech intelligibility means how clearly the speech can be heard. Speech intelligibility is the net result of the conditions under which communications takes place. It includes the behavior of talker and listener, the shape and finishes of the room, and the communication system under which the speech sound is propagated.

The effectiveness of this communication, and hence, the effectiveness of the learning environment is influenced by acoustical conditions in the classroom. Good classroom acoustics greatly facilitates learning.

UniMAP ACS is one of InnoSAT payloads and divided into two controllers; Adaptive Predictive Fuzzy Logic Controller (APFLC) and Adaptive Parametric Black Box Controller (APBB).

These controller functions are switched-on one at a time so that the performance of individual system can be studied and verified. It is proposed that initially, the ADCS takes charge from the time the satellite is being launched until it has come into or near its orbit. Then one of the controllers of the UniMAP ACS takes the control over from the ADCS.

The ADCS will wait for flags to indicate that there is no error in running of the controller programs. If after a specified delay no flag is received by the ADCS, then the controller will go into recovery mode. If recovery mode is successful, then the controller will start again. Otherwise, UniMAP ACS shall return satellite attitude control to ADCS. If the satellite attitude control of the UniMAP ACS is successful, it will continuous tracking and control of the satellite orientation for a specified time.