A. K. Ariff

Wireless data gloves Malay sign language recognition system

This paper describes the structure and algorithm of the whole Wireless Bluetooth Data Gloves Sign Language Recognition System, which is defined as a Human-Computer Interaction (HCI) system. This project is based on the need of developing an electronic device that can translate sign language into speech (sound) in order to make the communication take place between the mute & deaf community with the general public possible. Hence, the main objective of this project is to develop a system that can convert sign language into speech so that deaf people are able to communicate efficiently with normal people. This Human-Computer Interaction system is able to recognize 25 common words signing in Bahasa Isyarat Malaysia (BIM) by using Hidden Markov Models (HMM) methods. Both hands are involved in performing the BIM with all the sensor connecting wirelessly to PC with Bluetooth module. In the future, the system can be shrunk to become a stand alone system without any interaction with PC.

Heart Sound Analysis Using MFCC and Time Frequency Distribution

This paper presents heart sound analysis method based on Time-Frequency Distribution (TFD) analysis and Mel Frequency Cepstrum Coefficient (MFCC). TFD represents the heart sound in term of time and frequency simultaneously which while the MFCC defines a signal in term of frequency coefficient corresponding to the Mel filter scale. There are 100 normal data and 100 data with disease obtained from the hospital which consists of various kinds of problems including mitral regurgitation and stenosis, tricuspid regurgitation and stenosis, ventricular septal defect and other structural related disease. B-Distribution is chosen from a number of time-frequency analysis methods due its capability to represent the signal in the most efficient way in term of noise and cross term reduction. The advantage of MFCC is that it is good in error reduction and able to produce a robust feature when the signal is affected by noise. SVD/PCA technique is used to extract the important features out of the B-Distribution representation. The coefficient obtained from SVD-PCA and MFCC is later used for classification Artificial Neural Network. The results show that the system is able to produce the accuracy up to 90.0% using the TFD and 80.0% using the MFCC.

Malay Sign Language Gesture Recognition system

This paper describes hardware design and sensors setting and configuration for Malay sign language gesture recognition systems. A set of sensors consists of accelerometers and flexure sensors has been setup to capture the movement or gesture of shoulder, elbow, wrist, palm and fingers. This project is based on the need of developing an electronic device that can translate the sign language into speech (sound) in order to enable communication to take place between the mute and deaf community with the common public. Hence, the main objective of this project is to develop a system that can convert the sign language into speech so that deaf people are able to communicate efficiently with normal people. This Human-Computer Interaction system is able to recognize 25 common words signing in Bahasa Isyarat Malaysia (BIM) by using the Hidden Markov Models (HMM) method. Both hands are involved to perform the BIM with all the sensors connect wirelessly to a PC with a Bluetooth module. This project aims to capture the hand gestures which involve multiple axis of movement. Altogether 24 sensors have been setup in different hand locations to capture hand and wrist movement in different directions.

Malay continuous speech recognition using fast HMM match algorithm

This paper describes the implementation of fast hidden Markov model (HMM) match algorithm in a phoneme-based Malay continuous speech recognition system. The decoding algorithm decouples the computations of state-likelihoods of phone HMMs from the main search which is bounded by syntactical and lexical constraints. This avoids the redundant state-likelihood computations of identical phone HMMs for different word models in the tightly integrated search and thus substantially reduce the decoding time. The algorithm is implemented in the framework of one pass dynamic programming search. For a 541-word task, the fast HMM match reduces the real-time factor (RTF) by a factor of 31.8 from 286.45 × RT to 9.02 × RT, compared to without decoupling. The word accuracy is maintained at 91.6% without loss for a test set perplexity of 15.45 in speaker-dependent mode.

Linear dynamic models for classification of single-trial EEG

This paper investigates the use of linear dynamic models (LDMs) to improve classification of single-trial EEG signals. Existing dynamic classification of EEG uses discrete-state hidden Markov models (HMMs) based on piecewise-stationary assumption, which is inadequate for modeling the highly non-stationary dynamics underlying EEG. The continuous hidden states of LDMs could better describe this continuously changing characteristic of EEG, and thus improve the classification performance. We consider two examples of LDM: a simple local level model (LLM) and a time-varying autoregressive (TVAR) state-space model. AR parameters and band power are used as features. Parameter estimation of the LDMs is performed by using expectation-maximization (EM) algorithm. We also investigate different covariance modeling of Gaussian noises in LDMs for EEG classification. The experimental results on two-class motor-imagery classification show that both types of LDMs outperform the HMM baseline, with the best relative accuracy improvement of 14.8% by LLM with full covariance for Gaussian noises. It may due to that LDMs offer more flexibility in fitting the underlying dynamics of EEG.

Speaker Verification Using Gaussian Mixture Model (GMM)

This paper applies GMM for SV on Malay speech. The speaker models were trained on maximum likelihood estimated. The system was evaluated with 23 client speakers with 56 imposters. Malay clean speech data was used. 20 training of 3.5s utterances are used. The best performance achieved using 256-Gaussian imposter model and 32-Gaussian client model gave 3.01% of EER.

Discriminative tandem features for HMM-based EEG classification

We investigate the use of discriminative feature extractors in tandem configuration with generative EEG classification system. Existing studies on dynamic EEG classification typically use hidden Markov models (HMMs) which lack discriminative capability. In this paper, a linear and a non-linear classifier are discriminatively trained to produce complementary input features to the conventional HMM system. Two sets of tandem features are derived from linear discriminant analysis (LDA) projection output and multilayer perceptron (MLP) class-posterior probability, before appended to the standard autoregressive (AR) features. Evaluation on a two-class motor-imagery classification task shows that both the proposed tandem features yield consistent gains over the AR baseline, resulting in significant relative improvement of 6.2% and 11.2% for the LDA and MLP features respectively. We also explore portability of these features across different subjects.