Applying Hybrid Neural Network For Malay Syllables Speech Recognition

摘要: We proposed a hybrid technique for speech recognition which applying 2 different neural network architecture. The combines self-organizing map (SOM) known as unsupervised and multilayer perceptron (MLP) supervised Malay syllables recognition. used 2D feature extractor acts sequential mapping function in order to transform the acoustic vector sequences of signal into trajectories. output SOM is matrix with same dimension its elements take on binary values. transformation simplifies classification task by recognizer using perceptron. MLP classifies trajectories that each syllable corresponds to. Experiments were conducted 15 10 speakers conventional (MLP only) (SOM MLP). Our has achieved better performance where improves accuracy up 4.5%.