diff --git a/utils/feedparser.py b/utils/feedparser.py
index 16c4506ad..4654b5787 100755
--- a/utils/feedparser.py
+++ b/utils/feedparser.py
@@ -3740,7 +3740,7 @@ def convert_to_utf8(http_headers, data):
     chardet_encoding = None
     tried_encodings = []
     if chardet and data:
-        chardet_encoding = unicode(chardet.detect(data)['encoding'], 'ascii', 'ignore')
+        chardet_encoding = unicode(chardet.detect(data)['encoding'] or "", 'ascii', 'ignore')
     # try: HTTP encoding, declared XML encoding, encoding sniffed from BOM
     for proposed_encoding in (rfc3023_encoding, xml_encoding, bom_encoding,
                               chardet_encoding, u'utf-8', u'windows-1252', u'iso-8859-2'):