Research And Implementation Of The Voice-Controlled On Smart Home

With the development of embedded technology and Internet,smart home system based on intelligent voice control is also developing rapidly,and more and more people are entering the ordinary family.It also puts forward higher requirements for the intelligence,accuracy and ease of use of voice control.At present,most of the existing smart home products on the market can merely accurately match accurate instructions and their semantic intentions.Thus,users are unsatisfied with the accuracy and usability of these products because of their inaccurate match between generalized and modal control instructions and their semantic intentions.In a word,these products are not truly intelligent.However,we analyzed intentional linguistic data entries which were generated in common scenarios of smart home.Data analysis suggested that the accurate match between generalized and modal control instructions and their semantic intentions was the key to improving the accuracy and usability of smart home.Therefore,this paper is aimed at solving the problem of inaccurate match between generalized and modal control instructions ad their semantic intentions.As the application scenarios are very complicated,diverse devices are required to match the user's true intention.In this paper,based on the home local area network,the following efforts were made to improve the accuracy of match between generalized and modal control instructions and their semantic intentions:1.A word segmentation dictionary dedicated for smart homes was created to make Chinese word segmentation more accurate.2.Synonyms were added,following their original Chinese words.Original words and their synonyms were processed to make different expressions of the same intention unified,in order to avoid errors in the knowledge graph search.3.An exclusive knowledge graph was created in accordance with types of controllable intelligent devices in the user's practical application scenarios.In the process of intention matching,the scope and possibilities of intention matching were shrunk through searching the user's exclusive knowledge graph.In this way,intention analysis was more focused,maximizing the possibility of matching the user' s true intention.4.The user's habitual data search function was created and used.Relevant functions were updated to provide some reference items for matching the user's true intention.After a large number of experiments,the experimental results show that the method used in this paper can support the accuracy of intention matching of generic instruction and pattern instruction effectively,so as to improve the usability of voice control in smart home.