TV commercial auto-mute schematic/kit wanted

You would need AI and Machine learning. Unless you know of something that distinguishes commercial audio from all other broadcast audio that could be used as a trigger? (being annoying doesn't count - that would apply to 90% of programming).
 
There must be some kind of change in the signal when they switch to a commercial. Panasonic came out with a set that did just that in the early 80's. I know because my sister had one. The screen just went dark with no sound for the duration of the commercial. I believe it violated some kind of broadcast copyright laws and was discontinued. Same might be the case now also but there might be a gray area if the visual stays intact and only muting occurs. In any case I'm not interested in marketing anything, just eliminating aggravation😎
 
I think I know how those worked. There was a flashing white square that appeared in the upper left corner of the picture just before a commercial. This was so local stations could be ready for a advertising substitution if required. Most smaller CRTs were set up so that the image was zoomed so that you never saw the square. We had one big TV in the 80's where you could see the square. I think this system has been out of service for decades.
 
In analog days DTMF tones were sent to signal the moment of local ad insert. Occasionally you'd hear them. No idea how it's done now, but there might be some meta-data there to indicate what's program and what is not.
 
I use mythtv for recording, and it has advert detection, but it is not very reliable. There are a number of different ways that it can work (for video).

Commercial detection - MythTV Official Wiki shows how they do it.

One thing you could possibly use is change of levels.... The commercials are almost always louder than the normal program material. Note this would be tricky, you would probably need to detect an increase in the noise floor, as opposed to just a volume change...

Tony.
 
Does anybody know how to build one of these?

On network TV there is the CBS eye, ABC meatball, NBC peacock and Fox has their name. When I
manually remove spots (take 2-3 minutes per hour of program) I look for those items because
they are not present in the ads. I then render the files with the ads removed. Detecting those
items in digital video would take some serious pattern recognition. Since you're trying to do
this real time your best bet is pre-timed mute. You could probably do something with an
Arduino but just how bad is it pressing the mute button?

Stay safe.

 
I use mythtv for recording, and it has advert detection, but it is not very reliable. There are a number of different ways that it can work (for video).

Commercial detection - MythTV Official Wiki shows how they do it.

One thing you could possibly use is change of levels.... The commercials are almost always louder than the normal program material. Note this would be tricky, you would probably need to detect an increase in the noise floor, as opposed to just a volume change...

Tony.
That's exactly what I was thinking. There has to be a change in the noise floor.
 
On network TV there is the CBS eye, ABC meatball, NBC peacock and Fox has their name. When I
manually remove spots (take 2-3 minutes per hour of program) I look for those items because
they are not present in the ads. I then render the files with the ads removed. Detecting those
items in digital video would take some serious pattern recognition. Since you're trying to do
this real time your best bet is pre-timed mute. You could probably do something with an
Arduino but just how bad is it pressing the mute button?

Stay safe.

That's just it. Having to be vigilant to not be annoyed feeds the aggravation. Some people enjoy the ads more than the programming. I'm not one of those people. A lot of them literally ambush you. Constant bombardment. I'm pretty sure I'm not alone in this quest. I did find information about your suggestion with the arduino but that would not be accurate it seems.


BTW, bingo's attack is a good parallel of what I'm referring to
 
Last edited: