A method and system for automatically filtering wake-up words

Abstract

The present invention provides a method and system for automatically filtering wake-up words, the method comprising: acquiring wake-up words and corresponding wake-up word audio; acquiring corpus samples containing the wake-up words, and Establishing a voice model; acquiring user voice; comparing the user voice with the audio of the wake-up word to obtain the user wake-up word in the user voice, the position of the user wake-up word, and the situation of the user wake-up word; The position of the user wake-up word and the situation of the user wake-up word are matched with the speech model to obtain meaningless wake-up words that need to be shielded in the user wake-up words; shielding the meaningless wake-up words in the user voice is obtained Voice without wake-up word; analyzing the voice without wake-up word to obtain user semantics. The present invention shields meaningless wake-up words in the user's voice, identifies the generated voice without wake-up words, and improves the accuracy of semantic analysis.

Description

technical field [0001] The invention relates to the technical field of voice recognition, in particular to a method and system for automatically filtering wake-up words. Background technique [0002] With the rapid development of the Internet, people's lives are becoming more and more intelligent, so people are becoming more and more accustomed to using smart terminals to fulfill various needs. Moreover, with the increasing maturity of artificial intelligence-related technologies, the intelligence of various terminals is also getting higher and higher. Among them, the way of voice interaction, as one of the mainstream communication applications of human-computer interaction in smart terminals, is also more and more favored by users. [0003] The smart terminal recognizes the voice input by the user, and then takes corresponding measures. Therefore, the accuracy of the voice input by the user through the terminal seriously affects the feedback made by the smart terminal. ...

AI Technical Summary

Problems solved by technology

[0005] However, there may be multiple wake-up words in the user voice acquired by the smart terminal, some of which are only used to wake up the smart terminal, and some of them have practical significance, but the smart terminal cannot intelligently recognize them, and may be treated as meaningless wake-up words, thus Causes the smart terminal to make an error in analyzing the user's voice

[0006] In addition, due to factors such as the environment or the user's own reasons, the wake-up words in the user's voice acquired by the smart terminal may be incomplete. If there are only four of the set wake-up words in the user's voice, the smart terminal may not recognize them as meaningless wake-up words. In this case, on the one hand, the smart terminal may not be able to wake up in time; The completed wake-up word may be regarded as a meaningful word for semantic recognition, resulting in an error in the smart terminal's analysis of the user's voice

Images