[mythtv-users] 'Censoring' decoding 'bad' words off of Closed Captioning

Kichigai Mentat kichigai at comcast.net
Sat Apr 8 03:20:33 UTC 2006


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Apr 7, 2006, at 14.36, James Armstrong wrote:

> I just upgraded to FC4 and the latest IVTV drivers and now have Closed
> Captioning working on local HD and recordings off of Dishnetwork. Has
> anyone thought about 'bleeping' bad words for phrases from shows as  
> they
> are playing be looking at the CC data? There are products that do this
> built into VCR's / DVD players and Windows PC software.

Interestingly enough, I had an urge to insert a commonly used acronym  
in there... ;)

How could that work? Last time I checked, pre-set CCs (as in non-Live  
CCs) don't have time-codes in the stream, based on a per-word basis.  
The entire block of text is rendered all at once. So how could the  
software know when the offensive word is uttered? It would have to  
bleep out the entire phrase, wouldn't it? I can understand censoring  
the CC data (that's easy), but censoring one word?

I suppose the other way to do it is to work on a text-to-speech  
recognizer, but that presents two problems. First is recognizing the  
word. [Expletive], we have enough trouble getting simple Text-To- 
Speech to work recognizing one word (You try using the voice dialer  
on your cell phone, and you'll understand what I mean), but I can't  
imagine that working on a full TV stream, where we have a sound  
track, possibly multiple people talking at once, and a possible SFX  
track, all muxed into one. That sounds like a Herculean task far  
beyond anything our systems are capable of thus far.

The other problem is that CCs aren't always 100% accurate. Putting  
aside problems with interference, and possible typos (FYI: I've never  
really noticed a big problem with that), CCs are often not 1:1  
transcribings of the speech on screen. Often, I've noticed that the  
text on screen has been paraphrased so that the reader has a chance  
to read everything before the next block of text is displayed (I  
can't imagine reading more than two, maybe two and a half lines of CC  
text while paying attention to what's going on in the video stream)

> I think this
> would be a great feature to Myth. Also I took a brief look at the CC
> decoding code and see where V-Chip information is decoded. We can also
> block certain levels of ratings by password. Once the password is
> entered, you can watch shows for X number of hours until there is no
> activity, then it will prompt again.

Brilliant. Now, all we need to know is that all capture cards can  
forward a V-Chip rating to the Myth system. Of course, I'd probably  
suggest two changes:

First, forget messing with the CC. CC can be VERY unreliable, and  
susceptible to signal noise. Instead, use the data you get from your  
EPG data provider. I'm in North America, and the V-Chip ratings are  
included (and parsed) in my listings. I can look that data right up  
in MythWeb, no problem.

Second: Ditch the "for X number of hours" idea. Under that model, I  
could have someone watch something TV-14 starting at 9 P, and allowed  
to watch for three hours. But that show is only a half hour long, and  
South Park (TV-MA) is on next. So, switch to a per-show model  
instead. Simple solution.

>
> Just a thought. I could look into it more if someone would give me
> pointers. Is the CC data stored in the file or decoded in real time
> while watching the show?

It's stored in the MPEG stream, I think. That would explain how some  
DVDs can have CCs (And no, I don't mean subtitles).

> What part of the code puts it on the decodes it
> / reads it from the file / and puts it on the screen?
>
> Thanks,
> James
> _______________________________________________
> mythtv-users mailing list
> mythtv-users at mythtv.org
> http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (Darwin)

iD8DBQFENywEwAwn3hu8KxcRAvLXAKCODSJ9S1BL+3p/uHvMW7yz0OjB5QCfRquU
DGnF5SPLNCHR+hSuUpvZ10w=
=uxZs
-----END PGP SIGNATURE-----


More information about the mythtv-users mailing list