Question / Help First time user: Captioning / Delaying the output / Recording

[YvesC]

Hi there,

Awhile ago a client asked me to do a live stream (to a projector) of an interview, during a live event, with live captioning... I said "No problem" without really knowing what to do..
I digged a little deeper into the whole live stream thing and now I'm here.

So I figured out what to do on the hardware part:
A7SII with shotgunmic -> connected to a Paralinx ace HDMI transmitter
Paralinx ace HDMI receiver -> connected to a Blackmagic UltraStudio Mini recorder -> connected to my MBP with OBS
MBP with OBS -> connected with the projector through the HDMI out port

BUT I have some questions for the software part:
I have to live caption what people are answering to the interviewer.
--> I already tried with a .txt file where I would just typ in what they say, save the file, delete the line and enter the next phrase... There will be someone coming along to do the live typing.
This works ok but I was wondering if there is an easier/better solution to do this. Any tips, recommendations?
Delaying the output signal.
--> I need to delay the output signal, to be able to do the live captions in time, but I can't seem to find how to do this.
Only the source signal or the stream signal can be delayed. I'm using the "Fullscreen projector" option to go to the projector...
Recording the output source
--> I tried recording the output source while sending the output to "Fullscreen projector" but I noticed this resulted in a choppy recording. What would be the best settings to record the output?

Thanks a lot for the advice & help!

Greetings,

Yves

 

[Narcogen]

There isn't a way to introduce a broadcast delay with OBS currently.

Here is another thread about the same issue. These delays are usually done with specialized hardware.

https://obsproject.com/forum/threads/broadcast-delay-device.93977/

https://www.bhphotovideo.com/c/buy/Video-Delays/ci/5721/N/4028759659

What OBS can do is add a stream delay, but that really doesn't do what you want. You want to hold XX seconds of video to give the person doing transcription time to enter the text, but that's not what this delay does. If you turn that on, everything will be delayed by the interval you set-- so video can go out, say, 20 seconds later than you capture it, but that will just be added to whatever delay the transcriber introduces. What you want is a broadcast delay device that holds the video stream buffered before it gets to OBS, so that the text put in by the transcriber can be added to it before it gets streamed.

Another option would be to use a feature that does speech to text on an audio device and adds those to the stream as closed captions that can be enabled on platforms like YouTube. Check under the Tools menu for the Captions (Experimental) option. These are embedded 608/708 closed captions, and the accuracy is... eh... not really there, to be honest.

 

[YvesC]

Another option would be to use a feature that does speech to text on an audio device and adds those to the stream as closed captions that can be enabled on platforms like YouTube. Check under the Tools menu for the Captions (Experimental) option. These are embedded 608/708 closed captions, and the accuracy is... eh... not really there, to be honest.

Correct, and since the language won't be English I highly doubt that the accuracy will get better... :)
So in other words, it's not possible to delay (-> AKA HOLD) the outgoing video signal.

Ok ok, I can (have to) live with that. BUT I still need to find a good solution to do the live captioning. The fix I have now is really not that good:
Text file where I need to save the file after typing in the line, delete the line, type in the new line, save the file again to update the output...

I actually figured out what to do but the only thing is that I don't know with what program/app/webpage I'm able to reproduce this.
I just need something that will show the line that I'm typing without changing the area I'm typing into. Like a pages document where you hit enter and the line you just typed just jumps above the typing area (instead of going to a new line). I'd have to be able to set a green background on the document so I can chrome key it to just have the text as an end result.

Like this autocaptioner does: https://webcaptioner.com/captioner
^this would be ideal if I could just type it in myself

Any thoughts on what might help me to achieve that?

 

[Chip Chester]

Full disclosure: We do broadcast-style live captions in HD-SDI on-location, for simultaneous streaming and imag. I get in trouble if I add more than one frame to the local signal. Delaying it so someone can type (old-school) would be a tough sell.

It should be possible to do a wipe and lay in the webcaptioner screen as a graphic. That's basic switcher stuff, which I'm pretty sure OBS can make happen.

I was going to say "hey, just call us" until I got to the not-in-English part. Can't help you there, unless you want it Google translated live on-the-fly. We can supply that. The English part will be very, very accurate, and fast. The translation will be fast-ish. The translation accuracy will be up to Google. What are the expectations of your audience, from a speed, accuracy, and cost standpoint? That will drive your choices...

---Or...
Tell us what you did, since it was more than a month ago.