Target business positioning method and device for interactive text

Abstract

The invention provides a method and device for locating target services of interactive text, belonging to the technical field of speech recognition. The method includes: identifying the voice data input by the user, obtaining the interactive text input by the user in voice mode, determining the service positioning result of the interactive text based on the semantic understanding of the interactive text, and if the number of service positioning results of the interactive text is greater than one, the voice-based According to the prosodic characteristics of the data, the accented keywords of the interactive text are determined. If the accent value of the accented keywords is greater than the preset threshold, the business positioning result corresponding to the accented keywords is determined as the target business of the interactive text, and the process of business positioning is realized. In this method, the prosodic features and semantic information in the voice data are considered at the same time, and the prosodic features in the voice data can reflect the information emphasized by the user in the voice data, thereby improving the accuracy of service positioning.

Description

technical field [0001] The invention relates to speech recognition technology, in particular to a method and device for locating target services of interactive text. Background technique [0002] With the development and application of speech recognition technology, speech recognition has been applied to various terminal devices, and the use of natural language for human-computer interaction has become an obvious symbol of intelligent devices. With the application of natural language interaction, the requirements for the effect of speech recognition are getting higher and higher. [0003] In the prior art, after a user inputs voice data on a terminal device, the terminal device can perform voice recognition, semantic understanding, and service positioning on the voice data, and then provide corresponding services to the user according to the result of service positioning. Among them, the voice recognition process is to decode the voice data into interactive text, the semant...

AI Technical Summary

Problems solved by technology

[0004] However, in the above process, when performing service positioning based on the results of semantic understanding, if multiple services are determined during the positioning process, but the service requested by the user cannot be accurately located from the multiple services, service positioning conflicts may occur The problem

For example, according to the voice data input by the user, it is determined that the business positioning keyword in the voice data is "i*yi, movie", wherein the service corresponding to "i*yi" is the application download service, and the service corresponding to "movie" is It is a video service, assuming that the service positioning result corresponding to the voice data may be application download service and video service according to these two service positioning keywords, that is, according to the current service positioning method, it is impossible to accurately determine the target service at this time

Therefore, the positioning accuracy rate of the current voice data service positioning method is low

Images