diff --git a/pandoc_numbering.py b/pandoc_numbering.py
index ce3348a..298cba1 100644
--- a/pandoc_numbering.py
+++ b/pandoc_numbering.py
@@ -13,7 +13,6 @@
 import re
 import unicodedata
 import subprocess
-from unidecode import unidecode
 
 count = {}
 information = {}
@@ -54,7 +53,7 @@ def toJSONFilters(actions):
     json.dump(altered, sys.stdout)
 
 def removeAccents(string):
-    nfkd_form = unicodedata.normalize('NFKD', unicode(string, 'utf-8'))
+    nfkd_form = unicodedata.normalize('NFKD', u'' + string)
     return u"".join([c for c in nfkd_form if not unicodedata.combining(c)])
 
 def toIdentifier(string):