This update should fix the last remaining "Invalid string value" bugs.
In addition to that, the process of getting the title should end in the event that the headers suggest that the content being fetched is either not a valid content type or is too large.
Finally, the code should now degrade gracefully if it cannot find the mb_convert_encoding function; however that function or iconv is certainly recommended and you may still experience varied results if neither of these functions is available. Search your phpinfo for "mbstring" and "iconv" functions:
Thank you to Avensen for letting me know that some URLs in certain character sets were still returning garbage characters.
This lead me to review the code and as well as a bug in the conversion of the encoding, I added some additional code to better detect the original encoding and some improvements to the code in general.