I think the alignment tool would be much more powerful if it referred to a machine translation service like DeepL. In other words, it could segment the source file, machine translate it, and then use the results of the machine translation to figure out which parts of the target file match which parts of the source file.