This is an old post, but it surprises me no one mentioned using BBDuk with the file ribokmers.fa (a link to this file, and an explanation of how it has been created, can be found on this post).
BBDuk and SortMeRNA will be largely concordant, but BBDuk will run at a fraction of the time and accepts compressed fastq files as well. My impression is SortMeRNA is slightly more precise, but for cleaning datasets for downstream DGE BBDuk is fine.
Just to reinforce, though: as genomax pointed out, at this level (93%) of rRNA contamination, the best you can do is get back to the bench and redo the experiment, or at least deplete rRNA before doing the library prep again.
You never told us how the libraries were prepared. Did you depleted rRNA with something like RiboZero, or used an Illumina mRNA kit with poly-A capture? Or sequenced total RNA?