Difference between revisions of "Talk:Commercial detection with silences"

From MythTV Official Wiki
Jump to: navigation, search
m (A suggestion to improve the visibility of the running job in the Mythweb status page.)
(A suggestion to improve the visibility of the running job in the Mythweb status page.)
Line 16: Line 16:
 
     rec.markup.clean()
 
     rec.markup.clean()
 
     rec.update()
 
     rec.update()
 
+
    #Add this line to show the status in green in the Mythweb status page
+    #Add this line to show the status in green in the Mythweb status page
+
    job.update({'status': 4, 'comment': 'Running.'})
+    job.update({'status': 4, 'comment': 'Running.'})
 
  
  

Revision as of 13:43, 29 April 2015

Anyone know the correct way to edit a skiplist in Python?

Done. wagnerrp

It's stopped working recently

I can't figure out why but silence.cpp seems to be ignoring SIGALRM and so it never stops. If I kill the tail manually (like silence.ccp is supposed to in the alarm handler) it finishes properly. If I send a SIGALRM to silence.cpp nothing happens. I'm on Debian Wheezy and it was working until a few weeks ago so it is probably a kernel or security update. Has anyone else seen this behaviour?

Fixed in silence.cpp v4.2 - I don't know what has changed but the process has SIGALRM blocked when it starts so this change simply unblocks it and all is good in the world again.

A suggestion to improve the visibility of the running job in the Mythweb status page.

I would like to propose a small change to your code to include the following lines at line 192 of the silence.py script.

   # Purge any existing skip list and flag as in-progress
   rec.commflagged = 2
   rec.markup.clean()
   rec.update()
   #Add this line to show the status in green in the Mythweb status page
   job.update({'status': 4, 'comment': 'Running.'})


With this line added the status of the job shown in Mythweb status page shows running in green whilst the job is running, this is only a small cosmetic change but it helps to identify a running job in a long list of finished and queued jobs.

For some recordings the detection does not really start and never ends => jobqueue is blocked!

The cause of this issue is that mythffmpeg can not handle recordings using a variable count of audio streams and/or channels. Changing silence.py, substituting "mythffmpeg" with "avconv" (and installing libav-tools, when not already done) solved the problem for me. Have a look to the mythtv-users mailing list for details: http://www.gossamer-threads.com/lists/mythtv/users/578077

Perhaps this can help in the further develpement of the silence detction tool...

--Sjuk (talk) 15:44, 16 October 2014 (UTC)