SPEECH AND LANGUAGE TECHNOLOGY GROUP AT UNIVERSITY OF MALAYA
PRINCIPLE INVESTIGATOR: DR. MUMTAZ BEGUM MUSTAFA
DEPARTMENT OF SOFTWARE ENGINEERING
FACULTY OF COMPUTER SCIENCE & INFORMATION TECHNOLOGY
UNIVERSITY OF MALAYA
1. Text-to-Speech Synthesis system
YASMOON AZIYATI AHMAD: Blind A MALAY SPEECH ACOUSTIC MODEL FOR THE VISUALLY IMPAIRED INCORPORATING RELEVANCE SPEECH ATTRIBUTES
This research aims at investigating the perception of the Malaysian blind community on the usage and acceptance of the TTS system as an assistive tool when using computers. The role of the major attributes of synthetic speech towards the usage and acceptance among the VI will be the primary focus of this research so as to determine the expectation of the VI on the existing and future TTS systems. Several existing Malay TTS systems are evaluated by a group of VI. From the data collected, a modified Malay speech acoustic model is developed to incorporate the identified speech attributes. The synthetic speeches generated from the modified speech acoustic model were evaluated by the VI to determine whether the synthesized speeches have met their expectation.
Nor Irji’e Binti Basir: Desktop Access for the Visually Impaired using a Malay Screen Reader
The objective of this research project is to identify the requirements and expectations of a good screen reader by the visually impaired community, develop a Malay screen reader for desktop applications based on the identified requirements, and evaluate the system developed. This research focuses on the development of a screen reader that generates text based output from the user’s activity. The generated text is then used by the text to speech (TTS) system to generate the synthetic speech.
Asma Amalina: Acronyms and Abbreviations to text: A conversion system for Malay
In this research, an interactive web-based converter is developed to overcome the above problems. The proposed system works as an online dictionary that stores all acronym & abbreviation information in a database. As acronym & abbreviation is keep expanding, this system will expect and value user contribution to add new entries and modifies the existing acronym & abbreviation. With the unavailability of such system for Malay, the proposed system will provides many benefits to users by reducing the communication gap between the technology savvy younger generations with the older generation. On top of that, the proposed system can make Malay in par with other languages in term of facilities for language learning and understanding. Apart from being used as an online converter for the convenience of users, the proposed system is also important in the development of Text-To-Speech (TTS) as it provide a solution to current problem faced by Malay TTS.
TAY SUE PHING MALAY NUMBERS-TO-TEXT CONVERSION
This research has developed a Number-to-Text Converter for Malay that has the ability to identify several numerical formats including date, time, currency or measurement and convert them into text accurately. Upon the success of this system, it will be integrated into a functional Malay TTS system that can benefit various users for information retrieval.
2. Automatic Speech Recognition system
Fadhilah: FUZZY PETRI NETS AS A CLASSIFICATION METHOD FOR AUTOMATIC SPEECH INTELLIGIBILITY DETECTION OF CHILDREN WITH SPEECH IMPAIRMENTS
The inability to speak fluently degrades the quality of life of many individuals. Early intervention from childhood can reduce disfluency of speech among adults. Traditionally, disfluency of speech among children is diagnosed based on speech intelligibility assessment by speech and language pathologists, which can be expensive and time consuming. Hence, numerous attempts were made to automate the speech intelligibility detection. While current detectors use statistical methods to discriminate unintelligible speech by calculating the posterior probability scores for each articulatory feature class, the major drawback is that the results are most likely to be based on training and input data, leading to inconsistencies in discriminating speech sounds. As such, the performance of detectors is below that of humans. To overcome this limitation, a new classification method based on Fuzzy Petri Net (FPN) is proposed to improve the classification accuracy. FPN was proposed as it has greater knowledge representation ability to reason using uncertain or ambiguous information. In this research, the speech features of Malay impaired children’s speech are analysed for the identification of the significant speech features in the impaired speech which are related to the intelligibility deficits. This research also presents how the intelligibility classes can be detected by FPN. The results showed that FPN is more reliable in discriminating speech sounds than the baseline classifiers with improvements in the classification accuracy, precision and recall.
This study proposed a solution for the development of an ASR system for speech impaired children with limited source and target data by proposing a multi-layered adaptation technique. For the proposed technique, the first layer adaptation is the cross lingual adaptation, where the large-sized English TIMIT speech database that is phonetically rich and balanced, is used as a source model to be adapted with the medium-sized Malay non-speech impaired children database using supervised MLLR technique. The second layer adaptation makes use of the speech acoustic model generated from the first layer adaptation and adapts it with the small-sized database of children with speech impairment.
SARATU YUSUF ILU: AGE-BASED INTERFACE DESIGN FOR CHILDREN'S SPEECH PRONUNCIATION PRACTICE SYSTEM (2015)
The Speech Pronunciation Practice (SPP) system is very useful for children’s education such as language learning (L1) and second language learning (L2). SPP system offers additional time and materials; and provides a private, stress-free environment in which students can access effectively unlimited input, practice at their own pace and, through the integration of Automatic Speech Recognition (ASR), receive individualized, instant feedback. SPP system also benefits children with speech impairment including language impairment, oral motor disorders, stuttering, impaired articulation or voice impairment. Speech impairment is a communication disorder that affects the performance of a child in communication and education. This research aims at developing a suitable age based user interface design to be incorporated into an SPP system.
3. Academic Planner for University Students
Yeap Chun Sheng: Personal CGPA planning system for undergraduates: towards achieving the First Class CGPA
This research aims to explore the reasons for these failures and reviews the existing methods and tools used by the university students in improving their academic performance and aiding them to do their planning. In addition, this research also investigates and studies the behaviour of two Genetic Algorithm variants, namely Simple Genetic Algorithm and Controlled Elitist NSGA-II in determining its suitability in solving the corresponding optimization and search problem.