diff --git a/mopidy/audio/actor.py b/mopidy/audio/actor.py
index 7de98075..3910ee80 100644
--- a/mopidy/audio/actor.py
+++ b/mopidy/audio/actor.py
@@ -43,6 +43,7 @@ class Audio(pykka.ThreadingActor):
         self._mixer_track = None
         self._software_mixing = False
         self._appsrc = None
+        self._volume_set = None
 
         self._notify_source_signal_id = None
         self._about_to_finish_id = None
@@ -388,10 +389,16 @@ class Audio(pykka.ThreadingActor):
         volumes = self._mixer.get_volume(self._mixer_track)
         avg_volume = float(sum(volumes)) / len(volumes)
 
-        new_scale = (0, 100)
-        old_scale = (
+        internal_scale = (0, 100)
+        mixer_scale = (
             self._mixer_track.min_volume, self._mixer_track.max_volume)
-        return self._rescale(avg_volume, old=old_scale, new=new_scale)
+
+        if self._volume_set is not None and self._rescale(self._volume_set,
+                old=internal_scale, new=mixer_scale) == avg_volume:
+            return self._volume_set
+        else:
+            return self._rescale(
+                avg_volume, old=mixer_scale, new=internal_scale)
 
     def set_volume(self, volume):
         """
@@ -408,11 +415,13 @@ class Audio(pykka.ThreadingActor):
         if self._mixer is None:
             return False
 
-        old_scale = (0, 100)
-        new_scale = (
+        self._volume_set = volume
+
+        internal_scale = (0, 100)
+        mixer_scale = (
             self._mixer_track.min_volume, self._mixer_track.max_volume)
 
-        volume = self._rescale(volume, old=old_scale, new=new_scale)
+        volume = self._rescale(volume, old=internal_scale, new=mixer_scale)
 
         volumes = (volume,) * self._mixer_track.num_channels
         self._mixer.set_volume(self._mixer_track, volumes)