SYSTEM FOR AUTOMATICALLY EVALUATING FLUENCY OF PRONUNCIATION BASED ON CONVOLUTIONAL NEURAL NETWORK-RECURRENT NEURAL NETWORK

摘要

The present invention relates to a system for automatically evaluating fluency of pronunciation based on a convolutional neural network-recurrent neural network (CNN-RNN), to increase evaluation performance. According to the present invention, the system comprises: a voice reception unit receiving uttered voice of a learner; a memory storing a program evaluating the fluency of pronunciation with respect to the uttered voice of the learner; and a processor executing the program. The processor uses the entire uttered voice signal of the learner to evaluate the fluency of pronunciation with respect to the uttered voice of the learner by considering a filter bank operation of the CNN and using an output filter coefficient value as an input of the RNN.;COPYRIGHT KIPO 2020