diff --git a/render.py b/render.py
index 8c92656..f58de92 100644
--- a/render.py
+++ b/render.py
@@ -105,6 +105,7 @@ def compose(clips, target=150, base=1024, voice_over=None):
             })
             vo_min += offset
         for vo in voice_overs:
+            vo_low['filter'] = {'volume': '3'}
             scene['audio-center']['A1'].append(vo)
             vo_low = vo.copy()
             vo_low['filter'] = {'volume': '-6'}
@@ -168,6 +169,8 @@ def compose(clips, target=150, base=1024, voice_over=None):
         })
 
         transparency = seq() / 9
+        # 50% of time no transparancy of foregroudnd layer
+        # 50% some transparancy, 25%, 50%, 75% levels of transparancy
         transparancy = 1
         # coin flip which site is visible (50% chance)
         if chance(seq, 0.5):
diff --git a/sax.py b/sax.py
index aa258da..597062c 100644
--- a/sax.py
+++ b/sax.py
@@ -19,14 +19,14 @@ long = {
     "src": long_wav,
     "duration": 3600.0,
     "filter": {
-        "volume": "-3"
+        "volume": "-4"
     },
 }
 noise = {
     "src": nois_wav,
     "duration": 3600.0,
     "filter": {
-        "volume": "-1"
+        "volume": "1"
     },
 }
 project.append_clip('A1', long)