Virtual anchor generation method and device, electronic equipment and storage medium

Abstract

The invention relates to an intelligent decision making technology, and discloses a virtual anchor generation method, which comprises the steps of splitting an original audio into phonemes and performing vectorization to obtain a phoneme sequence vector, performing first sequence transformation on the phoneme sequence vector by using a bidirectional long short-term memory network to obtain a mouth shape feature sequence, and performing second sequence transformation on the mouth shape feature sequence to obtain a mouth shape feature sequence, obtaining a human face expression parameter, carrying out iterative optimization on a 3D human face model according to the human face expression parameter to obtain an initial human face model, carrying out adversarial training on the initial human face model by utilizing a trained adversarial network to obtain a standard human face model, and inputting an audio to be processed into the standard human face model to obtain a virtual anchor video. The invention further relates to a block chain technology, and the phonemes can be stored in block chain nodes. The invention furthermore provides a virtual anchor generation apparatus and device, and a storage medium. According to the method and the device, the problems that the mouth speaking action of the virtual anchor is discontinuous and the difference of audios of different crowds is not considered can be solved.

Description

technical field [0001] The present application relates to the technical field of intelligent decision-making, and in particular to a virtual anchor generation method, device, electronic equipment, and computer-readable storage medium. Background technique [0002] With the development of artificial intelligence technology, more and more AI technologies have been practically applied in various fields. Among them, virtual anchor is a popular direction in AI technology. Virtual anchor is used in news broadcast, virtual teacher, virtual interview and other scenarios. There is great potential to reduce labor costs as much as possible. [0003] The existing virtual anchor generation schemes usually use the acquired audio for subsequent processing, and do not take into account the large differences in the distribution of audio features. For example, the sound features of male voices, female voices, and people of different ages are very different. The anchor usually has the problem...

AI Technical Summary

Problems solved by technology

[0004] This application provides a virtual anchor generation method, device, and computer-readable storage medium, the main purpose of which is to solve the problems of discontinuous speech movements of virtual anchors and the lack of consideration of audio differences among different groups of people.

Images