diff --git a/mopidy/backends/spotify/playback.py b/mopidy/backends/spotify/playback.py
index 9069ce7e..d7e622fb 100644
--- a/mopidy/backends/spotify/playback.py
+++ b/mopidy/backends/spotify/playback.py
@@ -1,5 +1,9 @@
 from __future__ import unicode_literals
 
+import pygst
+pygst.require('0.10')
+import gst
+
 import logging
 import functools
 
@@ -35,6 +39,7 @@ class SpotifyPlaybackProvider(base.BasePlaybackProvider):
             self.backend.spotify.session.load(
                 Link.from_string(track.uri).as_track())
             self.backend.spotify.session.play(1)
+            self.backend.spotify.buffer_timestamp = 0
 
             self.audio.prepare_change()
             self.audio.set_appsrc(
@@ -54,5 +59,5 @@ class SpotifyPlaybackProvider(base.BasePlaybackProvider):
 
     def on_seek_data(self, time_position):
         logger.debug('playback.on_seek_data(%d) called', time_position)
-        self.backend.spotify.next_buffer_timestamp = time_position
+        self.backend.spotify.buffer_timestamp = time_position * gst.MSECOND
         self.backend.spotify.session.seek(time_position)
diff --git a/mopidy/backends/spotify/session_manager.py b/mopidy/backends/spotify/session_manager.py
index ad0a806e..d372bfa4 100644
--- a/mopidy/backends/spotify/session_manager.py
+++ b/mopidy/backends/spotify/session_manager.py
@@ -46,7 +46,7 @@ class SpotifySessionManager(process.BaseThread, PyspotifySessionManager):
         self.backend_ref = backend_ref
 
         self.connected = threading.Event()
-        self.next_buffer_timestamp = None
+        self.buffer_timestamp = 0
 
         self.container_manager = None
         self.playlist_manager = None
@@ -121,11 +121,13 @@ class SpotifySessionManager(process.BaseThread, PyspotifySessionManager):
             'sample_rate': sample_rate,
             'channels': channels,
         }
+
         buffer_ = gst.Buffer(bytes(frames))
         buffer_.set_caps(gst.caps_from_string(capabilites))
-        if self.next_buffer_timestamp is not None:
-            buffer_.timestamp = self.next_buffer_timestamp * gst.MSECOND
-            self.next_buffer_timestamp = None
+        buffer_.timestamp = self.buffer_timestamp
+        buffer_.duration = num_frames * gst.SECOND / sample_rate
+
+        self.buffer_timestamp += buffer_.duration
 
         if self.audio.emit_data(buffer_).get():
             return num_frames