The study aims to develop an automated isolated word speech recognition for Malay language that relies heavily on the well known and widely used statistical method in characterizing the speech pattern, the Hidden Markov Model (HMM). This paper discusses the development and implementation of an isolated Malay word speech recognition system using HMM as the acoustic model. This research focuses on isolated 5 phonemes word structure such as empat (four), lapan (eight), rekod (record), tidak (no), tujuh (seven) and tutup (close). The proposed system is relatively successful where it can identify spoken word at 88% recognition rate which is an acceptable rate of accuracy for speech recognition.

Isolated malay speech recognition using Hidden Markov Models

This paper describes a design procedure for a fuzzy logic based power system stabilizer (FLPSS) and adaptive neuro–fuzzy inference system (ANFIS) and investigates their robustness for a multi–machine power system. Speed deviation of a machine and its derivative are chosen as the input signals to the FLPSS. A four–machine and a two–area power system is used as the case study. Computer simulations for the test system subjected to transient disturbances i.e. a three phase fault, were carried out and the results showed that the proposed controller is able to prove its effectiveness and improve the system damping when compared to a conventional lead–lag based power system stabilizer controller.

Speaker-Independent Malay Syllable Recognition Using Singular and Modular Neural Networks

article presents a novel technique for the recognition of isolated keywords from spoken search queries. Recognition of the isolated keywords from spoken search queries may be considered as the first step towards the development of a speech-operated keyword-based searching technique. A database of 300 spoken search queries from Assamese language, a major Indian language mostly spoken by the people of north east India, has been created. The system developed during the study has been tested and evaluated with the above mentioned database. In the present study, Mel Frequency Cepstral Coefficient (MFCC) has been used as the feature vector and Multilayer Perceptron (MLP) to identify the phoneme boundaries as well as for recognition of the phonemes. Viterbi search technique has been used to identify the keywords from the sequence of phonemes generated by the phoneme recognizer. A recognition accuracy of 74.67% has been achieved in the present study.

/pdf/search-key-identification-in-a-spoken-query-using-isolated-r3nkr58w9k.pdf

Search Key Identification in a Spoken Query using Isolated Keyword Recognition

This paper presents initial studies of applying isolated speech recognizer (ISR) on different datasets of adultpsilas speech ISR is normally created for a targeted user or language. Even though the targeted user is defined, there are often speakers who are badly recognized. The purpose of this study is to determine whether a different recording specification has an effect on the performance of the recognizer. We used probabilistic models, known as hidden Markov models (HMMs) to interpret a sequence of word. In this study, we tested four datasets that consist of isolated word that utter the different days in standard Malay language. We apply the ISR on the datasets to determine the recognition rate performance and identify pattern of word recognition. The overall result shows that there is strong correlation between the different specification of the datasets and recognition rate. The study shows that, if certain specification is not fully considered during the recording, the recognition rate by the ISR is degraded.

Evaluating unsupervised data in isolated speech recognizer

This paper presents the speech recognition of Malay Alveolar consonants of Malay children using neural networks in a speaker-independent manner. The Alveolar consonants consist of /d/, /t/, /l/, /r/, /s/, /z/ and /n/. The Alveolar consonants are combined with six Malay vowels to form the consonant-vowel (CV) syllable sounds. 100 children are involved in the speech recording with a total of 4200 speech sounds. The speech sounds are recorded at a sampling rate of 20 KHz with 16-bit resolution. Linear Predictive Coding (LPC) is used to extract the speech feature extraction. Multi-layer Perceptron (MLP), which is one of the most popular neural networks, is used to classify the Alveolar consonants. Experiments are conducted to determine the optimal signal length of the consonants, and hidden neuron number of MLP. A maximum recognition rate of 62.14% is obtained at signal lengths of 150ms and 160ms.

Alveolar Consonant Recognition of Malay Children Using Neural Networks

Attempts at creating and using the first database for Malay syllables are presented. The speech vocabulary consists of 16 Malay syllables which are initialized with plosives and followed by succeeding vowels. The paper investigates the use of multilayer perceptron (MLP) and dynamic time warping (DTW) in recognizing these Malay syllable sounds that are quite similar to each other. From the experimental results, DTW and MLP achieved 77.09% and 90.82% respectively in overall average recognition rate.

E.L. Cheah

Papers

Malay syllable recognition based on multilayer perceptron and dynamic time warping