基于DNN-HMM模型的语音识别的语音导航系统

Speech Recognition based on DNN-HMM Model for Speech Recognition System

随着计算机技术和人工智能总体技术的发展,语音识别系统已成为一个越来越广泛的应用方向。基于传统声学模型的语音识别技术在语音导航系统中得到了广泛应用,但在模型的训练、识别率、鲁棒性方面存在一些缺陷。本文研究基于DNN-HMM模型的语音识别技术在语音导航中的应用,首先介绍了基于DNN-HMM模型的语音识别技术,然后给出语音导航系统的结构和主要功能。最后从识别率、鲁棒性方面对该系统进行了测试和分析,实验结果表明,引入DNN-HMM后的语音导航系统在模型的训练、识别率、鲁棒性方面有显著的改善。

With the development of computer technology and artificial intelligence,speech recognition system has become a more and more widely used.The speech recognition technology based on the traditional acoustic model has been widely used in the speech navigation system,but there are some defects in the training,the recognition rate and the robustness of the model.In this paper,the application of speech recognition technology based on DNN-HMM model in speech navigation is studied.Firstly,the speech recognition technology based on DNN-HMM model is introduced.Finally,from the recognition rate and the robustness of the test and analysis of the system,the experimental results show that the introduction of voice navigation system after DNN-HMM showed significant improvement in model training and recognition rate,robustness.