From: Joe Legris
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.0.1) Gecko/20020830
X-Accept-Language: en-us, en
Subject: Re: Detection of television commercials
Date: Fri, 06 Dec 2002 17:43:37 -0500
NNTP-Posting-Date: Fri, 06 Dec 2002 17:43:37 EST
Organization: Bell Sympatico
> A long time ago, there was an attempt to do this by looking at the phase
> shift changes in the color burst when switching from one
> source to another. Not sure that phase change exists in modern
> If you come up with something, be sure to patent it. The advertisers
> might pay BIG money for you NOT to license it.
> Richard Haendel wrote:
>>Can anyone provide information on how to do this? I'm not looking to build a
>>"commercial killer" but I need to record and log certain ones for a client.
> Bunch of stuff For Sale and Wanted at the link below.
> Tek 2465, TFT projector, ham radio, 30pS pulser
> Tektronix Concept Books...
I thought about this for a while several years ago and I concluded that
it should be possible to detect the statistical differences between
commercials and program content but there will always be uncertainty,
which translates into Type I and Type II errors. Type I error is to
identify program material as commercial and block it. Type II error is
the opposite, allowing commercials to pass as program material. Type II
errors are merely annoying. Type I errors are completely unacceptable to
most TV viewers.
Nowadays, digital video recorders are being used as time shifters,
alloying deletion of commercials on-the-fly. Televison networks have
been warning that if this practice becomes widespread they'll have to
find new sources of revenue as advertisers go elsewhere. So the question
is, would you rather pay [more] for TV or watch commercials?