基于神经网络的语音信号识别

摘要

语言是人类之间交流信息的主要手段之一，自电脑发明以来， 人们就一直致力于使电 脑能够理解自然语言。语音识别技术是集声学、语音学、语言学、计算机、信息处理和人工 智能等诸领域的一项综合技术，应用需求十分广阔，长期以来一直是人们研究的热点。 神经网络是在现代科学研究成果的基础上提出来的模拟人脑结构机制的一门新兴科 学，它模拟了人类神经元活动的原理，具有自学习、联想、对比、推理和概括能力，为很好 地解决语音识别这样一个复杂的模式分类问题提供了新的途径。 本文针时语音识别的特点． BP 神经网络在语音识别技术中的应用进行了探索性研究， 对 进而结合人工智能领域较为有效的方法——遗传(GA)算法。 针对传统 BP 算法识别准确率高 但训练速度慢的缺点，对 BP 网络进行改进，构建了一种基于遗传神经网络的语音识别算法 (GABP)，并建立相应的语音识别系统。仿真实验表明，该算法有效地缩短了识别时问，提 高了网络训练速度和语音的识别率。 关键词：语音识别，神经网络，遗传算法，遗传神经网络,BP 网络 RECOGNITIO THE RSREARCH OF SPEECH RECOGNITION BASED ON THE NEURAL NETWORK ABSTRACT Language is one of the most important means of exchanging information among the mankind．Since the computer was invented，many scientists have been devoted to enabling the computer to understand the natural language．Speech recognition is a comprehensive technology of such areas as acoustics， phonetics， linguistics， computer science，information processing and artificial intelligence，which can be used widely．The research of speech recognition technology has been focused by the world for a long time．The neural network is a new developing science，which simulates the mechanism of human brain and was putted forward by the developing of modern science． is not the overall description of human brain， the abstract， It but simulation and simplifying of the physical neural networks of human beings. The purpose of the research in this area is exploring the human brain mechanisms in information processing，storing and searching．If people can understand these mechanisms，a new way for the research of artificial intelligence，information processing and etc. can be opened up. Artificial neural network is a system which using a physically feasible system to imitate the structure and function of nerve cells in human brain，which has the ability of self—learning，contrasting，reasoning and summarizing .It have offered a new way in solving such complicated pattern classification problems as speech recognition． This paper mainly studies the application of the BP neural network in the research of speech recognition. BP neural network can get higher identification precision, but its training speed is very low, a new recognizing algorithm based on BP algorithm by combining with good effect method in ANN which named genetic algorithm (GA) was proposed and used to improve the BP neural network. Experiments results show that the training speed can be accelerated by the method and the recognition performance is also promoted． words: Key words speech recognition, neural network, genetic algorithm, genetic neural network, BP network 1．绪论 1.1 1.1 课题背景 1．1．1 语音识别概述 随着计算机技术的发展，人与机器之间的交流也越来越广泛和深入，计算机己经渗透 到人们生活的各个方面。在现代社会中，人们逐渐习惯借助计算机来完成各项事务。在这种 情况下， 如何让计算机智能化地与人进行通信， 使人机交互更加自然方便成为现代计算机科 学的重要研究课题之一。 语音识别(Speech Recognition)主要是指让机器听懂人说的话，即在各种情况下，准确 地识别出语音的内容，从而根据其信息，执行人的各种意图。 语音识别是一门涉及面很广的交叉学科， 它是目前发展最为迅速的信息研究诸领域中的 一个。语音识别的最大优势在于使得人机用户界面更加自然和容易使用。随着计算机技术、 模式识别和信号处理技