diff --git a/management/commands/update_subtitles.py b/management/commands/update_subtitles.py
new file mode 100644
index 0000000..2ca2511
--- /dev/null
+++ b/management/commands/update_subtitles.py
@@ -0,0 +1,20 @@
+import json
+import os
+import subprocess
+
+from django.core.management.base import BaseCommand
+from django.conf import settings
+
+from ...render import update_subtitles
+
+
+class Command(BaseCommand):
+    help = 'genrate kdenlive porject and render'
+
+    def add_arguments(self, parser):
+        parser.add_argument('--prefix', action='store', dest='prefix', default="/srv/t_for_time", help='prefix to build clips in')
+        parser.add_argument('--duration', action='store', dest='duration', default="3600", help='target duration of all fragments in seconds')
+        parser.add_argument('--offset', action='store', dest='offset', default="1024", help='inital offset in pi')
+
+    def handle(self, **options):
+        update_subtitles(options)
diff --git a/render.py b/render.py
index bdaaf2c..d7c60bd 100644
--- a/render.py
+++ b/render.py
@@ -2,10 +2,11 @@
 from collections import defaultdict
 import json
 import os
+import re
+import shutil
 import subprocess
 import sys
 import time
-import shutil
 from pathlib import Path
 
 import ox
@@ -46,6 +47,23 @@ def get_clip_by_seqid(clips, seqid):
         return clips.pop(i)
     return None
 
+
+def write_if_new(path, data, mode=''):
+    read_mode = 'r' + mode
+    write_mode = 'w' + mode
+    if os.path.exists(path):
+        with open(path, read_mode) as fd:
+            old = fd.read()
+    else:
+        old = ""
+    is_new = data != old
+    if path.endswith(".kdenlive"):
+        is_new = re.sub('\{.{36}\}', '', data) != re.sub('\{.{36}\}', '', old)
+    if is_new:
+        with open(path, write_mode) as fd:
+            fd.write(data)
+
+
 def compose(clips, target=150, base=1024, voice_over=None):
     length = 0
     scene = {
@@ -282,8 +300,9 @@ def render(root, scene, prefix=''):
     scene_duration = int(get_scene_duration(scene) * 24)
     for timeline, data in scene.items():
         if timeline == "subtitles":
-            with open(os.path.join(root, prefix + "front.srt"), "wb") as fd:
-                fd.write(ox.srt.encode(data))
+            path = os.path.join(root, prefix + "front.srt")
+            srt = ox.srt.encode(data)
+            write_if_new(path, srt, 'b')
             continue
         #print(timeline)
         project = KDEnliveProject(root)
@@ -304,8 +323,8 @@ def render(root, scene, prefix=''):
                         project.append_clip(track, {'blank': True, "duration": delta/24})
                         break
         path = os.path.join(root, prefix + "%s.kdenlive" % timeline)
-        with open(path, 'w') as fd:
-            fd.write(project.to_xml())
+        project_xml = project.to_xml()
+        write_if_new(path, project_xml)
         files.append(path)
     return files
 
@@ -410,8 +429,8 @@ def render_all(options):
 
         timelines = render(prefix, scene, fragment_prefix[len(prefix) + 1:] + '/')
 
-        with open(os.path.join(fragment_prefix, 'scene.json'), 'w') as fd:
-            json.dump(scene, fd, indent=2, ensure_ascii=False)
+        scene_json = json.dumps(scene, indent=2, ensure_ascii=False)
+        write_if_new(os.path.join(fragment_prefix, 'scene.json'), scene_json)
 
         if not options['no_video']:
             for timeline in timelines:
@@ -517,3 +536,40 @@ def render_all(options):
     print(json.dumps(dict(stats), sort_keys=True, indent=2))
     with open(_cache, "w") as fd:
         json.dump(_CACHE, fd)
+
+
+def update_subtitles(options):
+    import item.models
+
+    prefix = Path(options['prefix'])
+    duration = int(options['duration'])
+    base = int(options['offset'])
+
+    _cache = os.path.join(prefix, "cache.json")
+    if os.path.exists(_cache):
+        with open(_cache) as fd:
+            _CACHE.update(json.load(fd))
+
+    base_prefix = prefix / 'render' / str(base)
+    for folder in os.listdir(base_prefix):
+        folder = base_prefix / folder
+        with open(folder / "scene.json") as fd:
+            scene = json.load(fd)
+        offset = 0
+        subs = []
+        for clip in scene['audio-center']['A1']:
+            if not clip.get("blank"):
+                batch, fragment_id = clip['src'].replace('.wav', '').split('/')[-2:]
+                vo = item.models.Item.objects.filter(data__icontains='2-Whispered', data__title__startswith=fragment_id + '_').first()
+                if vo:
+                    for sub in vo.annotations.filter(layer="subtitles").exclude(value="").order_by("start"):
+                        sdata = sub.json(keys=['in', 'out', 'value'])
+                        sdata['value'] = sdata['value'].replace('<br/>', '<br>').replace('<br>\n', '\n').replace('<br>', '\n')
+                        sdata["in"] += offset
+                        sdata["out"] += offset
+                        subs.append(sdata)
+            offset += clip['duration']
+        path = folder / "front.srt"
+        srt = ox.srt.encode(subs)
+        write_if_new(str(path), srt, 'b')
+