Continuous speech man-machine interaction method and system

Abstract

The invention discloses a continuous speech man-machine interaction method and system, wherein the method comprises: receiving the continuous speech signal inputted by a user; cutting the continuous speech signal into a plurality of short speeches based on the long speech cutting technology ModelVad; identifying the plurality of short speeches based on a dynamic language model, and generating a corresponding identification result for each short speech; performing semantic completion on identification results based on the context semantic parsing techniques, and generating semantic completion results; and generating question and answer statements corresponding to the semantic completion results based on the Agent-Based dialogue management technology. According to the invention, only through once operation by a user can continuous speech identification be realized; through the dynamic language model, obtained identification results can be more accurate, and the identification accuracy of a continuous long speech can be improved. In addition, the continuous speech man-machine interaction method and system can actively guide the user, and improve user experiences.

Description

technical field [0001] The present invention relates to the technical field of human-computer interaction, in particular to a continuous voice human-computer interaction method and system. Background technique [0002] Speech recognition is an interdisciplinary subject. The fields involved in speech recognition technology include: signal processing, pattern recognition, probability theory and information theory, sound mechanism and auditory mechanism, artificial intelligence and so on. At present, users can input voice information through a microphone, and the voice recognition system can analyze the voice information and generate corresponding answers, such as the famous SIRI. [0003] However, the current speech recognition technology mainly adopts a question-and-answer method, that is, the user inputs a speech, and the system provides a corresponding answer for the speech according to the currently analyzed content. Every time the user enters the voice, they need to cli...

AI Technical Summary

Problems solved by technology

Every time the user enters a voice, they need to click once, and the system will recognize it again, which is inconvenient to operate

In addition, the speech recognition system adopts a unified language model, which cannot understand the user's intention according to the context, the accuracy rate is low, and it cannot actively ask the user questions, which is not intelligent enough

Images