Method and device for decoding and encoding multi-channel audio signal

A method and an apparatus for decoding and encoding a multi-channel audio signal are provided to reduce operation quantity required to convert the multi-channel audio signal into a time index/frequency index signal in order to extract or apply spatial information from or to the multi-channel audio signal. A signal on a frequency index of a down-mix signal is divided according to a band. The down-mix signal is converted into a QMF(Quadrature Mirror Filter) domain through at least two analysis filters bank having different filter coefficients according to the divided frequency band(40). The converted sub-band signals are up-mixed to a multi-channel signal by using spatial information(50). After that, the up-mixed multi-channel signal is converted into a time domain(60).