I'm pretty sure that xlrd decodes it all to unicode() in Python, so that should be a moot point. You would only need to worry about passing it as utf-8 to Vim at that point.