Hua Nong Ting

Speaker-independent Malay vowel recognition of children using multi-layer perceptron

Most of the speech recognitions are based on adult speech sounds. Less research is done in the recognition of children speech sounds. The speech of children is more dynamic and inconsistent if compared to adults speech. This paper investigates the use of neural networks in recognizing 6 Malay vowels of Malay children in a speaker-independent manner. Multi-layer perceptron with one hidden layer was used to recognize these vowels. The multi-layer perceptron was trained and tested with speech samples of Malay children with their ages between seven and ten years old. A single frame of cepstral coefficients was extracted around the vowel onset point using linear predictive coding. The vowel length was examined from 5 ms to 70 ms. Experiments were conducted to determine the optimal vowel length as well as the number of cepstral coefficients.

Self-Adjustable Neural Network for speech recognition

Neural networks with fixed input length are not able to train and test data with variable lengths in one network size. This issue is very crucial when the neural networks need to deal with signals of variable lengths, such as speech. Though various methods have been proposed in segmentation and feature extraction to deal with variable lengths of the data, the size of the input data to the neural networks still has to be fixed. A novel Self-Adjustable Neural Network (SANN) is presented in this paper, to enable the network to adjust itself according to different data input sizes. The proposed method is applied to the speech recognition of Malay vowels and TIMIT isolated words. SANN is benchmarked against the standard and state-of-the-art recogniser, Hidden Markov Model (HMM). The results showed that SANN was better than HMM in recognizing the Malay vowels. However, HMM outperformed SANN in recognising the TIMIT isolated words.

Speaker-independent phonation recognition for Malay plosives using neural networks

The paper investigates the use of neural networks in recognizing the phonation of the speech sounds. The proposed method classifies the Malay plosive sounds of adults and children based on phonation in a speaker-independent manner. The proposed method achieves encouraging result with an average accuracy of 98%.

Fuzzy-based discriminative feature representation for children's speech recognition

Article history: Available online 9M ay 2014 Automatic recognition of the speech of children is a challenging topic in computer-based speech recognition systems. Conventional feature extraction method namely Mel-frequency cepstral coefficient (MFCC) is not efficient for childrens speech recognition. This paper proposes a novel fuzzy-based discriminative feature representation to address the recognition of Malay vowels uttered by children. Considering the age-dependent variational acoustical speech parameters, performance of the automatic speech recognition (ASR) systems degrades in recognition of childrens speech. To solve this problem, this study addresses representation of relevant and discriminative features for childrens speech recognition. The addressed methods include extraction of MFCC with narrower filter bank followed by a fuzzy- based feature selection method. The proposed feature selection provides relevant, discriminative, and complementary features. For this purpose, conflicting objective functions for measuring the goodness of the features have to be fulfilled. To this end, fuzzy formulation of the problem and fuzzy aggregation of the objectives are used to address uncertainties involved with the problem. The proposed method can diminish the dimensionality without compromising the speech recognition r ate. To assess the capability of the proposed method, the study analyzed six Malay vowels from the recording of 360 children, ages 7 to 12. Upon extracting the features, two well-known classification methods, namely, MLP and HMM, were employed for the speech recognition task. Optimal parameter adjustment was performed for each classifier to adapt them for the experiments. The experiments were conducted based on a speaker-independent manner. The proposed method performed better than the conventional MFCC and a number of conventional feature selection methods in the children speech recognition task. The fuzzy-based feature selection allowed the flexible selection of the MFCCs with the best discriminative ability to enhance the difference between the vowel classes.

Computer-based Malay articulation training for Malay plosives at isolated, syllable and word level

This paper describes the use of computer as an articulation training system for Malay plosives at isolated, syllable and word level. The proposed system is more convenient than the traditional speech analyzing tools such as electropalatograph, where the latter requires an external electronic circuit to be attached into the mouth of client. The system is designed in a way that is user friendly and easy to use for the speech-language pathologist or even the client. The client undergoes speech training by just talking into the microphone and the system is able to recognize the sounds and classify them accordingly. Audio and visual feedback is used to help the client to identify his or her articulation errors as well as to make comparisons between his/her articulation models with the standard model. The system can be used for both children and adults.

Speaker-independent Malay isolated sounds recognition

This paper simply describes the use of neural networks in recognizing some Malay isolated sounds of Malay children in a speaker-independent manner. The isolated sounds are Malay plosive sounds, which are comprised of /b/, /d/, /g/, /p/, /t/ and /k/. A three-layer Multi-layer Perceptron (MLP) is used to train and recognize the speech sounds. The MLP output layer has an output layer of 6 neurons, which correspond to the 6 isolated plosive sounds. Network parameters such as hidden neuron number and error function, were investigated to achieve the optimal performance of the MLP. The proposed system was able to achieve the highest accuracy of 84.67%.

Vocal Fundamental Frequency and Perturbation Measurements of Vowels by Normal Malaysian Chinese Adults

The acoustic properties of vowel phonation vary across cultures. These specific characteristics, including vowel fundamental frequency (F(0)) and perturbation measures (Absolute Jitter [Jita], Jitter [Jitt], Relative Average Perturbation [RAP], five-point Period Perturbation Quotient [PPQ5], Absolute Shimmer [ShdB], Shimmer [Shim], and 11-point Amplitude Perturbation Quotient [APQ11]) are not well established for Malaysian Chinese adults. This article investigates the F(0) and perturbation measurements of sustained vowels in 60 normal Malaysian Chinese adults using acoustical analysis. Malaysian Chinese females had significantly higher F(0) than Malaysian males in all six vowels. However, there were no significant differences in F(0) across the vowels for each gender. Significant differences between vowels were observed for Jita, Jitt, PPQ5, ShdB, Shim, and APQ11 among Chinese males, whereas significant differences between vowels were observed for all the perturbation parameters among Chinese females. Chinese males had significantly higher Jita and APQ11 in the vowels than Chinese females, whereas no significant differences were observed between males and females for Jitt, RAP, PPQ5, and Shim. Cross-ethnic comparisons indicate that F(0) of vowel phonation varies within the Chinese ethnic group and across other ethnic groups. The perturbation measures cannot be simply compared, where the measures may vary significantly across different speech analysis softwares.

Acoustic characteristics of vowels by normal Malaysian Malay young adults.

The acoustic characteristics of sustained vowel have been widely investigated across various languages and ethnic groups. These acoustic measures, including fundamental frequency (F(0)), jitter (Jitt), relative average perturbation (RAP), five-point period perturbation quotient (PPQ5), shimmer (Shim), and 11-point amplitude perturbation quotient (APQ11) are not well established for Malaysian Malay young adults. This article studies the acoustic measures of Malaysian Malay adults using acoustical analysis. The study analyzed six sustained Malay vowels of 60 normal native Malaysian Malay adults with a mean of 21.19 years. The F(0) values of Malaysian Malay males and females were reported as 134.85±18.54 and 238.27±24.06Hz, respectively. Malaysian Malay females had significantly higher F(0) than that of males for all the vowels. However, no significant differences were observed between the genders for the perturbation measures in all the vowels, except RAP in /e/. No significant F(0) differences between the vowels were observed. Significant differences between the vowels were reported for all perturbation measures in Malaysian Malay males. As for Malaysian Malay females, significant differences between the vowels were reported for Shim and APQ11. Multiethnic comparisons indicate that F(0) varies between Malaysian Malay and other ethnic groups. However, the perturbation measures cannot be directly compared, where the measures vary significantly across different speech analysis softwares.

Speaker-dependent Malay Vowel Recognition for a Child with Articulation Disorder Using Multi-layer Perceptron

This paper investigates the use of Neural Network in recognizing six Malay vowels of a child with articulation disorder in a speaker-dependent manner. The child is identified to have articulation errors in producing consonant sounds but not in vowel sounds. The speech sounds were recorded at a sampling rate of 20kHz with 16-bit resolution. Linear Predictive Coding was used to extract 24 speech features coeeficients from a segment of 20ms to 100 ms. The LPC coefficients were converted into cepstral coefficients before being fed into a Multi-layer Perceptron with one hidden layer for training and testing. The Multi-layer Perceptron was able to recognize the all speech sounds.

Age Estimation Based on Children’s Voice: A Fuzzy-Based Decision Fusion Strategy

Automatic estimation of a speakers age is a challenging research topic in the area of speech analysis. In this paper, a novel approach to estimate a speakers age is presented. The method features a “divide and conquer” strategy wherein the speech data are divided into six groups based on the vowel classes. There are two reasons behind this strategy. First, reduction in the complicated distribution of the processing data improves the classifiers learning performance. Second, different vowel classes contain complementary information for age estimation. Mel-frequency cepstral coefficients are computed for each group and single layer feed-forward neural networks based on self-adaptive extreme learning machine are applied to the features to make a primary decision. Subsequently, fuzzy data fusion is employed to provide an overall decision by aggregating the classifiers outputs. The results are then compared with a number of state-of-the-art age estimation methods. Experiments conducted based on six age groups including children aged between 7 and 12 years revealed that fuzzy fusion of the classifiers outputs resulted in considerable improvement of up to 53.33% in age estimation accuracy. Moreover, the fuzzy fusion of decisions aggregated the complementary information of a speakers age from various speech sources.