GNOME Bugzilla – Bug 663842
Don't do extra audio-processing for raw Contents
Last modified: 2011-11-11 14:04:27 UTC
If a Voip Call uses the raw content type, we shouldn't preprocess the audio as whatever handles the raw audio will do that for us. So in that case don't ask pulseaudio to do echo cancellation
Branch at http://cgit.collabora.com/git/user/sjoerd/empathy.git/log/?h=disable-echo-cancellation will attach to bugzilla in a bti
Created attachment 201216 [details] [review] Call: Signal content-{added/removed} instead of sink-{added/removed} Tp-Farstream signals contents, not sinks (like tp-farsight did). So lets mirror that, which allows us to make better decisions about what kind of contents we are given
Created attachment 201217 [details] [review] Don't turn on echo cancellation on the source if it's for raw conferences
Created attachment 201218 [details] [review] Don't turn on echo cancellation on the sink if it's for raw conferences
Review of attachment 201216 [details] [review]: ::: src/empathy-call-window.c @@ +2703,3 @@ + g_object_get (content, "media-type", &media_type, "sink-pad", &sink, NULL); + g_assert (sink != NULL); Looks like the sink is leaked.
Review of attachment 201217 [details] [review]: ::: src/empathy-audio-src.c @@ +501,3 @@ +empathy_audio_src_set_echo_cancel (EmpathyGstAudioSrc *src, + gboolean enable) +{ I'd add a debug when disabling echo cancellation. Could be useful to debug "wtf is echo cancellation not working" bug.
Review of attachment 201218 [details] [review]: ::: src/empathy-audio-sink.c @@ +319,3 @@ + gboolean echo_cancel) +{ + sink->priv->echo_cancel = echo_cancel; same comment about the debug msg. ::: src/empathy-call-window.c @@ +2898,3 @@ + /* For raw audio conferences assume that the producer of the raw data + * has already processed it, soturn off any echo cancellation and any "so turn"
This problem has been fixed in the development version. The fix will be available in the next major software release. Thank you for your bug report.