Donggan speech recognition based on deep neural network

摘要

Donggan language, which is a special variant of Mandarin, is used by Donggan people in Central Asia. Donggan language includes Gansu dialect and Shaanxi dialect. This paper proposes a Deep Neural Network based on Donggan language speech recognition method for the Donggan Shaanxi dialect. We designed a text corpus and a pronunciation dictionary for Donggan Shannxi dialect and recorded the corresponding speech corpus. Then we trained the acoustic models of Donggan Shaanxi dialect by deep neural network (DNN). Experimental results demonstrated that the recognition rate of proposed DNN-based method can achieve lower word error rate than that of the monophonic hidden Markov model (HMM)-based method and triphone HMM-based method.