A Convolutional Framework for Mapping Imagined Auditory MEG into Listened Brain Responses