diff --git a/src/components/views/voip/CallView.js b/src/components/views/voip/CallView.js
index cc5289e7f1..4651e53551 100644
--- a/src/components/views/voip/CallView.js
+++ b/src/components/views/voip/CallView.js
@@ -79,8 +79,7 @@ module.exports = React.createClass({
         if (call) {
             call.setLocalVideoElement(this.getVideoView().getLocalVideoElement());
             call.setRemoteVideoElement(this.getVideoView().getRemoteVideoElement());
-            // give a separate element for audio stream playback - both for voice calls
-            // and for the voice stream of screen captures
+            // always use a separate element for audio stream playback
             call.setRemoteAudioElement(this.getVideoView().getRemoteAudioElement());
         }
         if (call && call.type === "video" && call.call_state !== "ended" && call.call_state !== "ringing") {
diff --git a/src/components/views/voip/VideoView.js b/src/components/views/voip/VideoView.js
index 46c5b354f8..487fd9f6ac 100644
--- a/src/components/views/voip/VideoView.js
+++ b/src/components/views/voip/VideoView.js
@@ -50,7 +50,10 @@ module.exports = React.createClass({
     },
 
     getRemoteAudioElement: function() {
-        return this.refs.remoteAudio;
+        // this needs to be somewhere at the top of the DOM which
+        // always exists to avoid audio interruptions.
+        // Might as well just use DOM.
+        return document.getElementById("remoteAudio");
     },
 
     getLocalVideoElement: function() {
@@ -106,7 +109,6 @@ module.exports = React.createClass({
                 <div className="mx_VideoView_remoteVideoFeed">
                     <VideoFeed ref="remote" onResize={this.props.onResize}
                         maxHeight={maxVideoHeight} />
-                    <audio ref="remoteAudio"/>
                 </div>
                 <div className="mx_VideoView_localVideoFeed">                
                     <VideoFeed ref="local"/>