An apparatus for processing an audio signal and method thereof are disclosed. The preset invention includes receiving a downmix signal including at least one object, preset information to render the downmix signal and preset attribute information indicating attribute of the preset information; rendering the downmix signal by applying the preset information to all data regions of the downmix signal, if the preset information is included in a configuration information region based on the preset attribute information; and rendering the downmix signal by applying the preset information to one corresponding data region of the downmix signal, if the preset information is included in a data region based on the preset attribute information, wherein the preset information is obtained based on preset number information indicating a number of the preset information and output channel information indicating a number of output channel of the rendered downmix signal. Accordingly, one of a plurality of preset information is selected using a plurality of preset metadata without user's setting on each object, whereby a level of an output channel of an object can be adjusted with ease.