Research on Framework of Speech Recognition Combining Text - Speech with Semantic Similarity

With the deepening of the speech recognition research, improving the accuracy of the general recognition engine is becoming more and more difficult. For the noise problem of Chinese speech recognition, in this paper we made a brief review and analysis about the current related articles of the semantic similarity and the text-speech similarity. We compared the advantages and disadvantages of various methods in speech recognition and pointed out the difficulties and the common problems existed in the research. At last, we give a speech recognition framework combining the semantic similarity and text-speech similarity. It provides a new way to improve the accuracy of speech recognition. Keywords—Similarity, speech recognition, natural language processing, semanteme.