There is a problem or a bug with a windows port of taglib, it does not properly convert unicode characters to non-unicode, but it can be easily fix by adding the following code to the tstring.cpp:

namespace TagLib {
  ...
  wchar_t toWideChar(const char c) {
   wchar_t wc;

    MultiByteToWideChar(CP_ACP, 0, &c, 1, &wc, 1);
    return wc;
  }

  char toAnsiChar(const wchar_t wc) {
   char c;

    WideCharToMultiByte(CP_ACP, 0, &wc, 1, &c, 1, NULL, NULL);
    return c;
  }
  ...
}

then all the castings from char to uchar need to be replaced by using the toWideChar function and vise versa. for example:

String::String(const std::string &s, Type t)
  ...
  for(std::string::const_iterator it = s.begin(); it != s.end(); it++) {
    // *targetIt = uchar(*it);
    *targetIt = toWideChar(*it);
    ++targetIt;
  }
  ...
}

ByteVector String::data(Type t) const
{
  ...
  case Latin1:
  {
    for(wstring::const_iterator it = d->data.begin(); it != d->data.end(); it++)
      // v.append(char(*it));
      v.append(toAnsiChar(*it));
    break;
  }
  ...
}

and so on...