论文部分内容阅读
语音识别的精度不够高一直是阻碍语音技术得以广泛应用的瓶颈,在具体的应用中充分利用背景知识是解决此问题的一种有效方法.在web语音浏览中,用户的语音输入为某个有限集的元素之一,本文利用这个特点,首先定义了一种文本字符串之间的相似度,利用相似度对识别引擎的识别结果进行后处理,进而给出更准确的识别结果.实验结果表明,采用这种方法,语音识别的正确率能够达到95％以上,为真正实现语音上网提供了有力支持.
The lack of accuracy of speech recognition has been a bottleneck hindering the widespread application of speech technology, and making full use of background knowledge in specific applications is an effective way to solve this problem.In web speech browsing, the user’s speech input is a limited In this paper, we use this feature to first define the similarity between a string of texts and then use the similarity to post-process the recognition result of the recognition engine, and then give a more accurate recognition result.Experimental results show that , Using this method, the correct rate of speech recognition can reach more than 95%, providing a real support for voice access.