Abstract: A speech decoding method is performed by a computer device, the speech including a current audio frame and a previous audio frame. The method includes: obtaining a target token corresponding to a smallest decoding score from a first token list including first tokens obtained by decoding the previous audio frame, each first token including a state pair and a decoding score, the state pair being used for characterizing a correspondence between a first state of the first token in a first decoding network corresponding to a low-order language model and a second state of the first token in a second decoding network corresponding to a differential language model; determining pruning parameters according to the target token and an acoustic vector of the current audio frame when the current audio frame is decoded; and decoding the current audio frame according to the first token list, the pruning parameters, and the acoustic vector.

Abstract: A speech decoding method is performed by a computer device, the speech including a current audio frame and a previous audio frame. The method includes: obtaining a target token corresponding to a smallest decoding score from a first token list including first tokens obtained by decoding the previous audio frame, each first token including a state pair and a decoding score, the state pair being used for characterizing a correspondence between a first state of the first token in a first decoding network corresponding to a low-order language model and a second state of the first token in a second decoding network corresponding to a differential language model; determining pruning parameters according to the target token and an acoustic vector of the current audio frame when the current audio frame is decoded; and decoding the current audio frame according to the first token list, the pruning parameters, and the acoustic vector.

Abstract: Embodiments of the present invention disclose an image processing method and apparatus, and a computer device. The image processing method includes: receiving information sent by a central processing unit (CPU), about a scene within a preset range around a to-be-rendered target object; rendering the received scene to obtain scene depth parameters, where the scene is obtained through shooting by a camera located at a ray light source; rendering the to-be-rendered target object to obtain rendering depth parameters, where the to-be-rendered target object is obtained through shooting by a camera not located at a ray light source; calculating ambient occlusion (AO) maps of the to-be-rendered target object in directions of ray light sources according to the scene depth parameters and the rendering depth parameters; and overlaying the AO maps in the directions of the ray light sources, to obtain an output image.

Abstract: Embodiments of the present invention provide a method for controlling character animation, in which the character animation includes at least two bones and skins corresponding to the bones, the method includes: (a) dividing the character animation into at least two parts, and setting an identification number for each part; (b) establishing a mapping table comprising a corresponding relationship between the identification number and skin data of each part; (c) picking skin data of an operation focus location in the character animation; (d) querying the mapping table according to the skin data, obtaining a corresponding identification number, and controlling the part in the character animation corresponding to the identification number. Embodiments of the present invention also provide a system for controlling character animation. Different parts of the character animation may be picked respectively by dividing the character animation into multiple parts.