Methods and algorithms for the analysis of speech signals

Аннотация

In recent years, many successes have been achieved in the direction of speech recognition. The results obtained make it possible to use voice dialogue instead of text dialogue, which provides speed and naturalness between man and machine. Currently, there are speech recognition systems for English, Russian, Spanish, French, Japanese Chinese and Arabic languages with high accuracy. The above systems are developed in dependence of economic and political relations. Leading speech recognition developers do not pay enough attention to other languages, in particular the Uzbek language. This article is devoted to methods and algorithms for translating Uzbek speech signals into text, the requirements of speech recognition systems, and the problems of determining the effectiveness of automatic speech recognition systems are considered. In addition,methods and algorithms for translating speech signals into text, methods for analyzing speech signals are considered. Solved the problem of analysis and processing of speech signals based on the formation of approximation structures of the correct Hadamard basis method, filtering speech signals based on the basis of discrete cosines.