audio: Switch to time based buffering in tee branches

docs/changelog.rst

@@ -11,6 +11,10 @@ v1.0.3 (unreleased)
   for 2.3 we didn't catch that our previous fix wasn't sufficient.
   (Fixes: :issue:`1153`)
 
+- Audio: Follow-up fix for :issue:`1097` still exhibits issues for certain
+  setups. We are giving this get an other go by setting the buffer size to
+  maximum 100ms instead of a fixed number of buffers. (Fixes: :issue:`1147`)
+
 
 v1.0.2 (2015-04-27)
 ===================

mopidy/audio/actor.py

@@ -164,7 +164,8 @@ class _Outputs(gst.Bin):
         # All tee branches need a queue in front of them.
         # But keep the queue short so the volume change isn't to slow:
         queue = gst.element_factory_make('queue')
-        queue.set_property('max-size-buffers', 15)
+        queue.set_property('max-size-time', 100 * gst.MSECOND)
+
         self.add(element)
         self.add(queue)
         queue.link(element)