from http://www.honeyplanet.jp/taglib-1.4_wchar.diff

gistfile1.diff

--- taglib-1.5/taglib/toolkit/tstring.cpp.orig  2008-02-05 00:14:45.000000000 +0900
+++ taglib-1.5/taglib/toolkit/tstring.cpp       2008-08-23 19:52:16.000000000 +0900
@@ -208,12 +208,22 @@
   s.resize(d->data.size());

   if(!unicode) {
-    std::string::iterator targetIt = s.begin();
-    for(wstring::const_iterator it = d->data.begin(); it != d->data.end(); it++) {
-      *targetIt = char(*it);
-      ++targetIt;
+    bool cjk = false;
+    //pre-scan: is there any cjk unicode character? if so, convert the string into utf-8.
+    for(unsigned int i=0; i< d->data.size(); i++){
+      if(d->data[i] > 0xff){
+        cjk = true;
+        break;
+      }
+    }
+    if(!cjk){
+      std::string::iterator targetIt = s.begin();
+      for(wstring::const_iterator it = d->data.begin(); it != d->data.end(); it++) {
+        *targetIt = char(*it);
+        ++targetIt;
+      }
+      return s;
     }
-    return s;
   }

   const int outputBufferSize = d->data.size() * 3 + 1;