Single- and multi-microphone speech dereverberation using spectral enhancement by