A method for representation of multi-channel audio signals