Speech processing method and device

Abstract

The embodiments of the invention provide a speech processing method and device. The speech processing method comprises the steps that a target speaking art text corresponding to to-be-processed speechis determined; the target speaking art text is subjected to word segmentation processing; according to a neural network model, the speaking art text subjected to word segmentation processing is subjected to sentence constituent analysis, sentence constituents corresponding to the speaking art text subjected to word segmentation processing are obtained, and the neural network model is obtained bytraining the speaking art text and the sentence constituents; and according to the obtained sentence constituents, the effective speaking art text corresponding to the to-be-processed speech is obtained. According to the speech processing method, meaningless information can be filtered, intent recognition is conducted based on the effective speaking art text, the difficulty of user intent recognition is lowered, the accuracy rate of intent recognition results is increased, meanwhile, the time and labor are saved, and the speech processing method is suitable for practical application.

Description

technical field [0001] The embodiments of the present invention relate to the field of computer technology, and in particular to a voice processing method and device. Background technique [0002] With the continuous development of the economy, the communication technology has developed rapidly, and more and more people start to use the communication system of the communication operator for communication. [0003] At present, if a user encounters any problem when using the communication system of the communication operator, he usually consults the customer service system of the communication operator. The customer service system needs to identify the user's intention, and the existing customer service system generally performs the user's intention identification directly based on the user's inquiry. [0004] However, some meaningless information is unavoidable in the user's inquiry, such as colloquial expressions, such as: "um", "ha", "oh" and other modal particles, which w...

AI Technical Summary

Problems solved by technology

[0005] Embodiments of the present invention provide a voice processing method and device to overcome the problems of difficult identification and low accuracy in existing customer service systems that directly identify user intentions based on user inquiries

Images