Capturing and authoring for closed captions?

[tomreedtoon]

I would like to know people's experiences with capturing video with closed captions and getting the video, with captions, through the editing process.

To begin with, what capture card do you use? I know there is a very expensive one used at the TV station where I work, intended for logging purposes, that captures with closed captions on Line 21 intact. But I don't want to pay $1000 or so for a capture card.

Second, assuming you do capture the video (let's say to DV AVI) what do you edit the video with, and what do you use to author it to DVD?

I'm used to capturing through the video passthru on my Sony camcorder to DV AVI, which is very convenient for editing out commercials and the like. But as far as I know, no camcorder or Firewire-type converter captures the Line 21 data.

Any advice or suggestions would be appreciated.

[Nelson37]

Search and read, check for McPoodle's site.

ATI Aiw can do it, I believe Hauppage, some others. Many issues, conversions, editing, errors, etc. Basically relatively easy to capture, edit, and convert with free software. It's all here, do a search or two.

[Cornucopia]

Most NoAm NTSC DV cameras do a pretty good job of retaining the CC of analog video sent through them, so does Canopus ADVC products (all in userdata field of video aux stream). Using McPoodle's guides and using good MPEG2 encoders and good DVD authoring apps that allow designation of the line 21 data, will maintain this info all the way through to playable DVD's.
But it isn't easy!

Scott

[Nelson37]

Do you intend to author with the captions retained as closed captions, or convert to subtitles?

Very few apps handle captions at all. Most consumers are unaware of their existence, or at least of the difference to subtitles.

The captions themselves contain numerous grammatical and spelling errors, lines missing, etc.
These errors not real obvious in casual viewing, but once you start dealing with them in bulk the errors really start to add up. Current testing indicates that at least to some extent, these errors can vary with the individual transmission.

Have not determined with certainty that the errors in the captured files are also present in the live screen display. At least some of them definitely are, but I cannot verify that this is true for all.

Graphedit conversions reveal that the display time values are often exceeding tolerances of the display or conversion mechanisms, ie out time for caption 1 is within a few thousands of a second of the in time for caption 2, resulting in caption 2 not being displayed. Times are sometimes identical or even overlap.

[Cornucopia]

re: CC errors

CCs are done 2 ways: live for news/etc transmission, and edited for prerecorded titles (movies, etc).

The 1st one is loaded with errors and mistimings because the person typing in the CCs is doing realtime "translation" or "transcription". So they mess up.

The 2nd one is almost always dead-on correct and timed well (to the best of their ability--the SHERLOCK HOLMES DVD CCs are not to good, done by UCLA students no doubt)

The other thing to consider is that if the original master was analog, the CC info is "digitized" along with the rest of the picture.
And, like any graphic, is DCT compressed, which can degrade some CCs to the point of misrepresentation. Hopefully the compression doesn't go through too many generations.

CCs are supposed to arrive early because it takes the CC decoder a LONG time to read the data (it's spitting it out at such a small bitrate), so it needs lead time. There certainly is an art to getting things to coincide with talking correctly, yet still fit within the constraints of CC technology...

Scott

[Nelson37]

Scott, do you do much capture and/or viewing of CC with premium channel movie broadcasts?

I seem to see less typos, misspellings, and other errors with these, I would say that missing letters, particularly at the end of lines, is by far the most common. There are letter dropouts within words, such as "correct" might display as "Crrct". I would have thought such errors would be caught with simple error-checking. There are also occassional lines completely missing.

This leads me to believe that either the transmission or the capture is at fault. Real-time viewing I do maybe 10% of the time, I am tending toward at least most faults being in the original transmission, IOW what I capture and what I see on the TV is the same thing.

What I found doing both the Graphedit conversion and converting the ATI text file was that the individual character errors are still there, but for the most part I can correct for the overlapping times. There are also occassional sections missing, but fewer with the ATI file than the Graphedit method. Have not been able to effectively compare captured file with live broadcast.

What I am trying to do is gather more data on what types of errors are seen frequently in viewing the live broadcast of CC, and also info on types of capture and conversion used, and errors seen there.

[BobT]

The only video card which I have used for capturing is the ATI AIW 7500. If you capture in MPEG, with this ATI card it will not include closed captions. Some of the DVD recorders will capture closed captions but it is difficult finding which ones since this category was never included for the DVD recorder reviews. I prefer keeping closed captions in my recordings if they are present since I have a hearing problem and rely upon them. The Pioneer DVR-220 and DVR-531H both will record closed captions and these are the DVD Recorders I use. There is no mention of this in the Pioneer manuals but they do in fact capure closed captions in the Mpeg2 recordings. I use MPEG-VCR Ver 3.14 to edit the recordings, remove commercials etc and TMPGEnc DVD Author Ver 1.5.15.49 to make the DVD. The closed captions are then in the completed DVD. It is my understanding that later versions of TMPGEnc will strip off closed captions so be careful of the latest version of this software. I have never tried to convert closed captions to subtitles so I cannot help you there. Most of the commercial pressed DVD have both closed captions and subtitles but it can vary. When TV shows come out on DVD they frequently have only closed captions.

[Cornucopia]

Scott, do you do much capture and/or viewing of CC with premium channel movie broadcasts?

I seem to see less typos, misspellings, and other errors with these, I would say that missing letters, particularly at the end of lines, is by far the most common. There are letter dropouts within words, such as "correct" might display as "Crrct". I would have thought such errors would be caught with simple error-checking. There are also occassional lines completely missing.

This leads me to believe that either the transmission or the capture is at fault. Real-time viewing I do maybe 10% of the time, I am tending toward at least most faults being in the original transmission, IOW what I capture and what I see on the TV is the same thing.

What I found doing both the Graphedit conversion and converting the ATI text file was that the individual character errors are still there, but for the most part I can correct for the overlapping times. There are also occassional sections missing, but fewer with the ATI file than the Graphedit method. Have not been able to effectively compare captured file with live broadcast.

What I am trying to do is gather more data on what types of errors are seen frequently in viewing the live broadcast of CC, and also info on types of capture and conversion used, and errors seen there.

If they're movies, the broadcasting station (sat, cable, OTA) either has a tape master that they're converting live to MPEG, and perhaps re-encoding again to fit the bandwidth multiplexing budget variations. Or they have a digital server that they're real-time recompressing for the same reasons. CC purity isn't a priority with those devices, so it's good that it's getting through at all. I'm sure for those masters, they originally had someone create good CCs.

or

It could be something else...

(Me, I RARELY cap broadcasts at all, let alone stuff I need CC for)

Scott

[Nelson37]

Bob, can you comment on the frequency of typos and missing lines or words?

Have you ever captured or viewed the same show twice and seen differences in the captions?

Most particularly, can you verify or comment on whether the captured captions are or are not identical to what is viewed? Is what you see really what you get?

MMC 9.12, and presumably onward, now captures captions with an MPG file, the VCR format is gone and no longer used.

[ozymango]

For capturing over-the-air broadcasts, if you use the PVR-250 that keeps the closed captioning in the MPEG2 stream, which you could then extract to .srt with mpg2srt. Then you could run that through as a subtitle with various authoring programs, plus you could edit the srt beforehand to clean up any truly weird cc errors.

But I have no idea if the PVR-250 will keep the cc info if you're inputting an outside source.

Though I can't see a good reason to work too hard to preserve the final cc info on your end DVD -- cc is only good on Region 1 discs and even then, every DVD player will play subtitles but not every TV will show CC, so you're still ahead of the game.

EDIT: I have a (mostly mild) case of tinnitus so I LOVE SUBTITLES!!!!

[Cornucopia]

Don't know about the PVR-250, but I use a PRV-9000 (industrial model, for work) and yes, it does retain CC info from direct input (whether Composite, S-Video, or DV). I don't see why those other models wouldn't either.

Every 21' TV newer than '93, some even smaller than that, is REQUIRED by US law to support CCs, so they still come in handy. Especially as it caters to a slightly different audience (partly because of the accepted rules of style/formatting) than those that specificly use subtitles only.
Personally though, I find DVD-type subs a lot easier to deal with, too.

Scott

[ozymango]

Every 21' TV newer than '93, some even smaller than that, is REQUIRED by US law to support CCs, so they still come in handy.
Scott

Kinda sorta, but have you ever had to figure out how to get to the closed captions on some TV sets nowadays? I wish they'd standardize on a simple "Close Captions" button! On my set, to get to the CC menu, it's:

Menu, enter, arrow right, arrow right, select, *don't* hit menu again, that'll take you back to the main screen, hit the "down" button to select the CC option, then arrow left or arrow right depending on if you want CC 1,2,3, 4 or TELETEXT.

See, this is why I LOVE DVD SUBTITLES!!!

[Nelson37]

Few tools for dealing with captions, subtitles have many more tools and options. Just simpler to deal with.

I would really like to know from regular users of closed captions what general level of quality you observe, also general consistency.

[vhelp]

Afternoone Nelson37, guys..

This could be slightly incorrect, but from short dealings with CC and the
capturing (in my case, watching-only through MMC 7.1) when you first launch
or ellect to turn on CC, you may get some garbled up words. But, I have
found that if you wait a moment, it will correct itself and give you a
consistant and clean CC output.

Could it be that some members are experiencing this, and consiquentially,
giving up (in past) and are now comming out of the wood-works and commenting
their unsuccessful feedback ?? I don't know. Anyway.

I have a 13" TV set that I purchased in 2001 and it has several CC formats
to choose. When I ellect to turn it on, it sometimes takes a little time
to que or buffer it into cache, and finally display. As with my MMC 7.1,
I've seen garbled up words in the begining, and missing or misspelled words,
but after a moment (it varies) it straighten's itself out and is clean.

I would guess that its pretty much the same for anything that is capable of
displaying CC -- computer capture card/device or tv.

Captured AVI-- Also, I always thought that the CC text (data) was in the
first few lines at the top of the captured avi file. When I capture any-
thing from my Analog Cable, I see what lines or dashes on the first ONE
or TWO lines of my tv window. My point of this, is this, that the CC data
is captured, but we lack the interpreter or utility to read in these AVI
files and decode them from CC and incorporate them into the AVI image, frame
by frame as CC text. Would be a nice project for someone interested to develop.

An new capture card-- I have to admit, I'm a sucker for ATI, cause I just
purchased yet another capture card by them from CC (Circuit City) for $61
bucks (I had cash on hand) and walked out with it, a couple of weeks ago.
Since then, I've been pretty lazy to set it up.

The new card I bought is the:

--> ATI TV Wonder Pro

-vhelp 4221

[Nelson37]

Well, I will answer my own question.

Closed Captions are Highly Variable. Just captured the same movie twice, major differences between the two CC files. Several lines missing in sections.

This is comparing two ATI CC files, not the original viewing.

So the capture, from ATI, is not the same each time. I will assume that the actual CC viewed is also potentially different with each broadcast of the same movie. Will attempt some multiple viewings along with the capture to verify this and report back.