Depends what said double bytes are. Also for some things you can drop it down to 8 bits, also gain a lot more space and memory that way.
If said double bytes are something I can force with u16 unicode then OK. Same if you can do a find and replace later -- in plain ASCII ( http://www.asciitable.com/
) you have things like numbers in hex are 3 followed by the number you want, another is lower case is 20 hex higher than upper (41=A and 61=a, 42=B and 62=b....) so if your chosen encoding is that.
For shiftJIS (note shiftJIS in the real world supports various types of unicode for the lower values, most games however only have the 8100 + ranges seen on http://www.rikai.com/library/kanjitables/kanji_codes.sjis.shtml
, fortunately said encoding includes a copy of the Roman characters which can get you out of a pinch) then there are a few choices. Japanese text editors often have a "J-Ascii" type mode. JWPCE (I see also there is a fork in JWPxp now https://mijet.eludevisibility.org/JWPxp/JWPxp.html
) having such a thing. There are three mode buttons in the right hand side, one is plain ASCII, another is Japanese and the final will force it into the 8??? range that houses Roman characters in Japanese fonts. Save as shiftJIS and slice it out with your hex editor.
In actual Japanese (and Chinese and Korean) writing circles there is another in the form of NJstar https://www.njstar.com/cms/
. I am less familiar with this but it should also be able to do the job.
You could probably write some scripts as well (this sort of manipulation is what most of the likes of python, perl and such live for) but nobody has really shared those. Don't have any examples here, however might as well have a little tool I find handy in these situationshttp://www.romhacking.net/utilities/504/