Provided is a method and system of sound processing for a mobile terminal based on hand-holding and orientation detection. The method includes: detecting whether the mobile terminal is held by hand; in response to determination of the mobile terminal being held by hand, detecting whether it is one-hand holding or two-hand holding; in response to determination of one-hand holding, detecting whether it is landscape holding or portrait holding, and conducting corresponding frequency response equalization and stereo processing control; and in response to determination of two-hand holding, determining whether the mobile terminal is in an interaction procedure, and conducting corresponding frequency response equalization and stereo processing control. Compared with the related art, the problem that frequency response of the speakers of existing mobile terminals significantly changes with different holding postures that may eventually lead to poor sound quality is solved, to provide more consistent speaker sound quality in real life.