A method for encoding a three-dimensional audio signal is provided. The method includes: An encoder obtains a current frame of a three-dimensional audio signal; obtains coding efficiency of an initial virtual speaker for the current frame based on the current frame of the three-dimensional audio signal; and when the coding efficiency of the initial virtual speaker for the current frame meets a preset condition, determines an updated virtual speaker for the current frame from a set of candidate virtual speakers; encodes the current frame based on the updated virtual speaker for the current frame, to obtain a first bitstream; or when the coding efficiency of the initial virtual speaker for the current frame does not meet the preset condition, encodes the current frame based on the initial virtual speaker for the current frame, to obtain a second bitstream.