Article Preview
Top1. Introduction
With the acceleration of the pace of life, traditional human-computer interaction methods, such as keyboard input, mouse input and other commonly used tools, have gradually failed to meet people's demand for intelligent use of robots, and an urgent need for a more convenient, intelligent and diversified Way of interaction. Voice is the most direct and convenient way to communicate with people. Researchers have been incorporating intelligent machines with certain speech recognition capabilities into human voice interaction objects, and implementing human-intelligent machine interaction., The real-time control of intelligent robots through voice is also a difficult and hot topic of research. However, the existing human-robot speech interaction system has a slow recognition speed, high running cost, and poor system stability, which limits the development of humanoid robots to a certain extent. Nowadays, with the development of speech signal processing technology and the advancement of information technology, the interaction between humans and intelligent machines will gradually become a reality. To achieve a more efficient and stable voice interaction system between humans and robots, the practicality of voice interaction is improved. Researching and realizing the intelligence of robots, and using more intelligent service methods to meet the needs of social developmentare essential to improve the quality of human life. At the same time, it is also of great significance for the widespread application and popularization of intelligent robots.
With the continuous development of mobile Internet technology and the emergence of various smart products, these smart products continue to improve the living standards of human beings. At the same time, researchers are more and more inseparable from these smart products. This new technology has become part of our lives. Then, the main interaction methods of these smart products include speech recognition, and the accuracy of speech recognition determines the quality of our experience of using the product (Xinmei Zhong et al, 2018). Noise is a kind of sound that causes people to be fidgety, or the volume is too strong and endangers human health. To reduce noise is to control noise and take measures to reduce noise. In the available environment, human faces variety of noises and it cannot be avoided. Even face-to-face communication is affected by various noises. For example, on the outside, the content of the conversation between two people will be covered by the dialogue between other people outside, which will bring certain communication barriers.In addition, mobile smart devices will encounter more noise. Background noise is collected at the voice acquisition port. It is also affected by channel noise during voice transmission; and it is often affected by background noise at the voice output. The phenomena will seriously affect the quality of people's voice signal exchange (A. S. Kolokolov et al,2019). In some specific cases, some noise can completely overwhelm people's original speech signals, which brings some problems to the subsequent processing like speech encryption and speech recognition. Therefore, how to effectively eliminate noise in various environments is the meaning of speech noise reduction. Just as additive noise and collected speech are an additive relationship, but there is also a non-linear relationship. This analysis is an important direction in the field of noise. Convolution noise can also be transformed into additive noise through specific changes, which can effectively deal with convolution noise Periodic functions can be used to generate periodic noise and this method is often used in digital signal processing.