<div dir="ltr"><div class="gmail_extra"><div class="gmail_quote">On Wed, Sep 10, 2014 at 10:09 AM, Rob Owens <span dir="ltr"><<a href="mailto:rowens@ptd.net" target="_blank">rowens@ptd.net</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">----- Original Message -----<br>
> From: "Carl Hunter" <<a href="mailto:cdhunter2@yahoo.com">cdhunter2@yahoo.com</a>><br>
> To: "Discussion about MythTV" <<a href="mailto:mythtv-users@mythtv.org">mythtv-users@mythtv.org</a>><br>
> Sent: Wednesday, September 10, 2014 9:22:07 AM<br>
> Subject: Re: [mythtv-users] mythcommflag not working<br>
><br>
> On 9/9/2014 2:09 PM, Gary Buhrmaster wrote:<br>
> > Commercial detection is hard, and is constantly<br>
> > getting harder as broadcasters intentionally try to<br>
> > avoid commercials getting detected (they do not<br>
> > want you to skip the commercials; that is how they<br>
> > make money). The algorithms of old probably<br>
> > could use some TLC. I believe if someone wants to<br>
> > invest (the likely considerable) effort to improve<br>
> > commercial detection the devs have a stash of<br>
> > samples from around the world to work with (what<br>
> > works well in one country, or even for some<br>
> > channels or providers may not work well in others),<br>
> > so if you are serious about working with the code,<br>
> > I would contact them for access to the stash.<br>
> > As I recall, there are some alternative programs<br>
> > out there that try to detect commercials, and<br>
> > perhaps someone could examine them to see if<br>
> > their ideas could be integrated into MythTV.<br>
> > Personally, I just use the skip forward key.<br>
> ><br>
><br>
> I've been using the skip forward key for a few years now. I just know<br>
> it was working well for me and my source hasn't changed.<br>
><br>
> Would an idea be to get mythcommflag from 0.24 compiling / working<br>
> against a 0.27 database to do some comparing or is that not really<br>
> possible since other dependencies have changed like maybe ffmpeg?<br>
><br>
<br>
If anybody's still running older versions of MythTV, I could send you one of my problematic mpg files to test. Or will mythcommflag not work if the video is not in the tester's database?<br>
<br>
-Rob<br></blockquote><div><br></div><div>Yes, that would be quite interesting. My experience is that comm flagging (on SD MPEG2 files from a PVR-500) went from near perfect in 0.24 to catching maybe 1 out of 3-4 in 0.26 and beyond (I skipped 0.25). I assumed it was some change in the integrated ffmpeg but if there were other changes it would be nice to bring them back, ideally with some switches to choose the "old" or "new" algorithm, per channel.<br><br></div><div>Karl<br></div></div></div></div>