🚀 go-pugleaf

In <1992Sep03.042045.5095@ecst.csuchico.edu> rmintz@ecst.csuchico.edu (Rich Mintz) writes:
>I'm interested in building a device (or making my 386sx 
>act like one) that will be able to distinguish television 
>commercials from television programs. I've talked to a 
>few people and have heard of a couple of ways it is said 
>that this can be done.
 
>First, I was told that the audio during commercial 
>announcements on tv is boosted higher than that used for 
>normal programs (for obvious reasons). However, I've 
>tried all kinds of analysis with volume levels, averages 
>over specific periods of time, etc, using my computer's 
>sound card (with stereo capability), with no luck at 
>finding any dependable pattern to distinguish    ads from 
>shows.

There is no control of levels during commercial playback. However,
there are several factors that do make it seem like this is done.
The biggest one is that most program material, when leading  "into"
a commercial, has soft audio levels which trail off. Then the spot
starts, usually with full levels, and the effect is that it is louder.
Also, commercials are usually played-back from a different machine.
Whenever the "source" is switched, the level will change some. 
The machines which playback commercials often do have the levels
set high - but this is not a plot, it`s merely because the
operator doesn`t want any clients complaining about low levels.
Compression is used, but it tends to be significant only on the
obnoxious "shouting" type of spots (e.g., used car).

The bottom line is this: The broadcast TV industry makes its
money from commercials. If there were any reliable way of 
defeating that process, you can rest assured that it would be
fixed in rather short order.

Actually, something that would prove more useful, especially
given the upcomming presidential mud fest, is the bullshit filter.

John Abt