Password understanding method for voice-driven virtual human

Abstract

The invention discloses a password understanding method for a voice-driven virtual human. The method comprises the following steps: determining a to-be-understood password text; inputting the to-be-understood password text into a password understanding model to obtain a password understanding result output by the password understanding model, the password understanding model comprising a word vector training module, a word vector correction module and a cosine similarity calculation module, the word vector training module carrying out word vector learning by adopting a D-Skip-Gram algorithm based on the training set to obtain a word vector space; the word vector correction module establishing semantic constraint conditions based on the virtual human action instruction library to obtain a corrected word vector space; and the cosine similarity calculation module calculating the cosine similarity in the corrected word vector space to realize understanding of the password text. According to the method, the passwords expressed by the same meaning and different languages can be understood into the same action instruction, so that the user can express the intention by the oral expression habit of the user, and the user experience feeling of the voice-driven virtual human can be improved.

Description

technical field [0001] The invention belongs to the technical field of virtual reality, and in particular relates to a password comprehension method for a virtual human. Background technique [0002] With the development of virtual reality technology, people hope to participate in the VR scene they are viewing. Currently, the more common VR scene interaction method is that the user immerses in the VR scene from the first perspective by wearing a helmet, and then uses the operating handle to perform posture changes, Object grabbing and other actions realize interaction. However, the existing interaction method is based on body movements or manual operations, which is not simple enough for the operation of virtual human beings. Directly driving virtual human actions through voice control can simplify the operation process of virtual human beings and reduce real-life Body interaction makes the control of virtual characters more natural, convenient and flexible, and is also ben...

AI Technical Summary

Problems solved by technology

At present, in the research on voice-driven virtual human motion, most of the voice recognition of fixed passwords, and then directly match the virtual human motion variables according to the text recognized by the voice, so as to realize the virtual human motion control. When the user wants the virtual human To execute the action of "raising the head", it is necessary to issue the prescribed password of "raising the head". The avatar cannot execute the "head up" action smoothly when the same action as "look up" has a password expressed in a different language

Virtual reality emphasizes the sense of immersion. If the user can only control the movement of the virtual human with the specified password, it will affect the user experience and the sense of immersion in the VR scene.

Images