A method and device for generating dialogue characters

Abstract

The present application discloses a method and device for generating a dialogue character image, wherein the method includes: the user terminal generates coordinates of a three-dimensional facial grid corresponding to the avatar picture by using a preset first neural network model according to the avatar picture of the opposite end of the conversation point; in the preset scene where a 3D avatar needs to be generated, in each voice collection cycle, the user terminal extracts the voice feature parameters from the audio signal collected in this cycle, and according to the audio signal and the preset second nerve The network model generates corresponding emotional state vectors, and generates each of the coordinate points according to the coordinate points of the three-dimensional facial grid, the voice feature parameters, the emotional state vector and the preset third neural network model For the 3D expression data in this cycle, display the 3D expression of the call peer on the current call interface according to the 3D expression data. The application of the invention can increase the vividness of dialogue characters during a call, and reduce the impact of call overhead on communication quality.

Description

technical field [0001] The present application relates to the field of computer applications, in particular to a method and device for generating dialogue characters. Background technique [0002] In the existing video call or live broadcast solutions, the following methods are generally used to present the image of the dialogue characters: [0003] Method 1: One party in the video call obtains a preset image, selects the corresponding mouth expression in the library according to the characteristics of the voice information, and generates expression frames to send to the other party. These expression frames are two-dimensional, like emoticons. [0004] Method 2: When the video call screen is not smooth, Terminal 1 will crop the current video frame according to the avatar, and then send the cropped video frame and avatar identifier to Terminal 2. These avatars are also two-dimensional. [0005] During the process of proposing the present invention, the applicant found that t...

AI Technical Summary

Problems solved by technology

[0005] During the process of proposing the present invention, the applicant found that the above-mentioned method exists: the quality of network communication is deteriorated and the video call is not vivid enough, which affects the user experience. The specific analysis is as follows

[0006] 1. The above method is to send your own two-dimensional expression frame or avatar to the other party in the video when the video call screen is not smooth. These expression frames and avatar are two-dimensional and not realistic enough to truly generate the user’s image, so that the playback effect of the video object during the video call is not vivid enough, which affects the user experience

[0007] 2. The emoticon frame or avatar generated in the above method needs to be sent to the other party, so it will occupy a certain amount of network transmission resources

The expression frames or avatars mentioned above are usually generated when the video call screen is not smooth, so that in the case of poor communication quality, the further occupation of transmission resources will inevitably lead to the deterioration of communication quality

Images