Speaker-independent Malay isolated sounds recognition

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