In Ruby 1.9, it's easy. In Ruby 1.9, Strings are encoding-aware sequences of characters, so you can just index into it and you will get a single-character string out of it:
'µsec'[0] => 'µ'
However, in Ruby 1.8, Strings are sequences of bytes and thus completely unaware of the encoding. If you index into a string and that string uses a multibyte encoding, you risk indexing right into the middle of a multibyte character (in this example, the 'µ' is encoded in UTF-8):
'µsec'[0] # => 194
'µsec'[0].chr # => Garbage
'µsec'[0,1] # => Garbage
However, Regexps and some specialized string methods support at least a small subset of popular encodings, among them some Japanese encodings (e.g. Shift-JIS) and (in this example) UTF-8:
'µsec'.split('')[0] # => 'µ'
'µsec'.split(//u)[0] # => 'µ'