SLICER: Learning universal audio representations using low-resource self-supervised pre-training