+1

Just because you can do something doesn't mean you should. Doing this sort of data validation within the database will only result in a mess. Even if you have a valid list of names to check against and write the code to "figure out" what they really meant, the chance would remain that two distinct names could be identical after all duplicate letters are removed.

Furthermore, it's not a good architectural philosophy. The application that is the transit between the human who can't spell and the database is the appropriate location for the validation. That is where the 'did you really mean X?' conversation can reliably occur. It is then the database's job to store that data and retrieve it on demand.