diff --git a/src/filescanner.c b/src/filescanner.c
index 39a47db4..ff78a9c4 100644
--- a/src/filescanner.c
+++ b/src/filescanner.c
@@ -699,6 +699,13 @@ filescanner_process_media(char *path, time_t mtime, off_t size, int type, struct
   mfi->time_modified = mtime;
   mfi->file_size = size;
 
+  if (type & F_SCAN_TYPE_COMPILATION)
+    mfi->compilation = 1;
+  if (type & F_SCAN_TYPE_PODCAST)
+    mfi->media_kind = MEDIA_KIND_PODCAST; /* podcast */
+  if (type & F_SCAN_TYPE_AUDIOBOOK)
+    mfi->media_kind = MEDIA_KIND_AUDIOBOOK; /* audiobook */
+
   if (type & F_SCAN_TYPE_FILE)
     {
       mfi->data_kind = DATA_KIND_FILE;
@@ -742,13 +749,6 @@ filescanner_process_media(char *path, time_t mtime, off_t size, int type, struct
       goto out;
     }
 
-  if (type & F_SCAN_TYPE_COMPILATION)
-    mfi->compilation = 1;
-  if (type & F_SCAN_TYPE_PODCAST)
-    mfi->media_kind = MEDIA_KIND_PODCAST; /* podcast */
-  if (type & F_SCAN_TYPE_AUDIOBOOK)
-    mfi->media_kind = MEDIA_KIND_AUDIOBOOK; /* audiobook */
-
   if (!mfi->item_kind)
     mfi->item_kind = 2; /* music */
   if (!mfi->media_kind)
diff --git a/src/filescanner_ffmpeg.c b/src/filescanner_ffmpeg.c
index aeefecf3..9cb83448 100644
--- a/src/filescanner_ffmpeg.c
+++ b/src/filescanner_ffmpeg.c
@@ -427,6 +427,10 @@ scan_metadata_ffmpeg(char *file, struct media_file_info *mfi)
 		break;
 	      }
 #endif
+	    // We treat these as audio no matter what
+	    if (mfi->compilation || (mfi->media_kind & (MEDIA_KIND_PODCAST | MEDIA_KIND_AUDIOBOOK)))
+	      break;
+
 	    if (!video_stream)
 	      {
 		DPRINTF(E_DBG, L_SCAN, "File has video (stream %d)\n", i);