diff --git a/yt_dlp/extractor/tokentube.py b/yt_dlp/extractor/tokentube.py
index d6362117f..579623fed 100644
--- a/yt_dlp/extractor/tokentube.py
+++ b/yt_dlp/extractor/tokentube.py
@@ -6,7 +6,10 @@
 
 from .common import InfoExtractor
 from ..utils import (
+    clean_html,
+    get_element_by_class,
     parse_count,
+    remove_end,
     unified_strdate,
     js_to_json,
     OnDemandPagedList,
@@ -35,7 +38,7 @@ class TokentubeIE(InfoExtractor):
             'id': '3950239124',
             'ext': 'mp4',
             'title': 'Linux Ubuntu Studio perus käyttö',
-            'description': 'md5:854ff1dc732ff708976de2880ea32050',
+            'description': 'md5:46077d0daaba1974f2dc381257f9d64c',
             'uploader': 'jyrilehtonen',
             'upload_date': '20210825',
         },
@@ -45,7 +48,7 @@ class TokentubeIE(InfoExtractor):
             'id': '3582463289',
             'ext': 'mp4',
             'title': 'Police for Freedom - toiminta aloitetaan Suomessa ❤️??',
-            'description': 'md5:cd92e620d7f5fa162e8410d0fc9a08be',
+            'description': 'md5:37ebf1cb44264e0bf23ed98b337ee63e',
             'uploader': 'Voitontie',
             'upload_date': '20210428',
         }
@@ -90,7 +93,10 @@ def _real_extract(self, url):
             r'<a\s*class="place-left"[^>]+>(.+?)</a>',
             webpage, 'uploader', fatal=False)
 
-        description = self._html_search_meta('description', webpage)
+        description = (clean_html(get_element_by_class('p-d-txt', webpage))
+                       or self._html_search_meta(('og:description', 'description', 'twitter:description'), webpage))
+
+        description = remove_end(description, 'Category')
 
         self._sort_formats(formats)