This version expects to receive the word in UTF-8 encoding, and it will then use iconv behind the scenes in order to convert to and from the dictionary encoding.
prefix = '/usr/share/myspell/' # Fedora
speller = Hunspell.new(prefix + 'en_US.aff', prefix + 'en_US.dic')
speller.valid?('dog') # -> true
speller.valid?('dpg') # -> false
speller.suggest('dpg') # -> ["dog", "dg", "pg", "deg", "dig", ...]
Copyright 2011 Roman Shterenzon, released under the MIT license