A method for rendering 3D audio is provided. The method includes: obtaining a to-be-output audio signal, wherein the to-be-output audio signal comprises one or more audio layers; selecting a to-be-filtered audio layer from the one or more audio layers; selecting a rendering Head-Related Transfer Function, HRTF, from a database of HRTFs; applying the rendering HRTF to the to-be-filtered audio layer to generate a filtered audio layer, such that the to-be-output audio signal comprises the filtered audio layer and is a 3D audio signal. In this way, 3D effects are only applied to specific audio layers of an audio signal, thereby efficiently rendering 3D audio without unnecessarily impacting the timbre of the audio signal.