[mythtv] New Method for Commercial Detection.

[Chris Pinkham]

* On Tue May 02, 2006 at 06:21:59PM +1200, Robert Dunn wrote:
> The advantages of a system like this is that it does not rely on any 
> 'coincidental' difference between the commercials and the programming 
> such as the volume, as technically it could be taught to eliminate any 
> piece of video/audio, but of course relies on having an up-to-date set 
> of commercial keys, but I have in my work touched on some techniques in 
> keeping the keys up to date automatically.

If the concept works, then training a system like this could also be as
simple as giving a simple thumbs up/down on blocks of video within a
program while watching it or while in the editor.  It seems that where I
am in the U.S. that you see mostly the same commercials during the same
timeslots because companies have specific target audiences.  I think that
this method would have to be used in conjunction with other methods to
determine where the blocks start and end, otherwise it could be very
cpu intensive to analyze a whole hour long show second by second looking
for signatures that match the known lists.  In your testing, how have you
handled this kind of situtation or have you just done analysis to compare
two clips to see if they contained the same audio.  Another thing that
will complicate this is the fact that analog broadcasts will not be
identical every time, so you wouldn't always get a 100% match as you would
with an exact digital copy.

> MythTV seems like the perfect platform for me to build upon, both saving 
> me time, and giving something back to the open source community. Whilst 
> I am not a student, I noticed you are looking for people to develop the 
> commercial detection aspect of MythTV for the Google Summer of Code.

:)  We're always looking for people to enhance commercial detection.  I
haven't spent much time on it recently becuase it works well enough for
me and I've devoted my limited Myth time to other things.  I decided to
put it on the SoC project list because there are several things I've been
wanting to do but haven't had time, and that I thought a student could
probably take care of over the course of a summer.

> Is this something you guys would be interested in having me work on? 
> Please get back to me with any thoughts.

I'm all for anything that makes commercial detection better, with the
key word being better.  I think that this has the potential of working
well in some situations.  I think it would probably be best combined
with the current methods because it will probably never be 100%
accurate by itself because of situations like when a new commercial
that we've never seen before is broadcast as the first or last commercial
in a commercial break.  Another issue that might prevent 100% accuracy
would be differences in audio in the same commercial in analog broadcasts
or when that commercial is broadcast on a different station,

Feel free to make modifications/additions to the flagger if you want,
Lucas gave some good information in his email on doing so.  We all
would like better commercial detection.

--
Chris