Ahmed Khalifa with Jason Barnard at BrightonSEO September 2019
Ahmed Khalifa talks with Jason Barnard about why you would use captions and subtitles for your video.
Firstly, accessibility. All sorts of people benefit from captioned / subtitled videos, not just deaf people says Ahmed Khalifa: non-native speakers, people watching with the sound off, when the speakers’ accents aren’t clear (think Glaswegian :) … and some people just like to read along. Automatic captions need to be corrected. Machines simply cannot get everything right (especially the scene directions and ambiance descriptions).
Apparently, professionals can write captions almost in real time, including descriptions about background noise. I personally hadn’t thought about how important that can be for context. It’s not just what we say, but the context in which we say it.
Ahmed is deaf, and relies on captions. But all sorts of people benefit from captioned videos… Of course, we get onto Glaswegian accents. Heads up Craig Campbell. Auto captioning is far from perfect. It needs to be corrected. Then we get onto pushing that to transcripts and adapted transcripts. Apparently, professionals can write captions almost in real time, including descriptions about background noise. Guess who hadn’t thought about how important that can be for context. It’s not just what we say, but the context in which we say it. I speak too much without engaging my brain. Ahmed saves the day and makes me sound intelligent. Phew !
Jason Barnard
I’m going to reread your name to make sure I get it right. #SEOisAEO, welcome to the show. Ahmed Khalifa.
Ahmed Khalifa
Actually, yeah, that's quite like intro. People should do that more often.
Jason Barnard
Yeah, but the shame was I actually had to read it.
Ahmed Khalifa
I don't blame you for that.
Jason Barnard
Thank you. Lovely to meet you. I believe you started your career in Brighton.
Ahmed Khalifa
Kind of. I did have an agency experience in Worthing. So I lived there for like a year and I did agency work. I did a few things building up to it, then from then on, I just kept up the momentum.
Jason Barnard
And you worked for a company called Fresh Egg? My first thought when I saw fresh egg was rotten egg.
Ahmed Khalifa
My first thought was some kind of farm or something And that was about seven or eight years ago now. So it's been a while.
Jason Barnard
Okay. Right. Brilliant stuff.
Video Captions
Jason Barnard
You were talking about that earlier on, and shamefully I didn't see your talk… but you gave me a quick idea of what it was all about. Captions on videos and how rubbish they can be when they're auto generated on YouTube and people leave them idiotically.
Ahmed Khalifa
Pretty much. And they don't give it a second thought.
Jason Barnard
Yeah. And, and you were telling me why that's particularly interesting for you?
Ahmed Khalifa
Well, I mean, if I'm going to go from my personal experience, I depend on captions because I'm deaf and I depend on captions just to access videos. But then even if it's not for me, it can be for all for people for whom English is not their first language. They're learning the language. Or it could be some kind of learning disability and they need captions to keep up. It could be attention deficit disorder. It could be even be, for all of us, because you may be an urban transport or maybe you're in a library or whatever, and you just want to watch the video in silence.
Jason Barnard
Or it could be that the person speaking has a really, really thick Glaswegian accent like Craig Campbell.
Ahmed Khalifa
That is one thing that I mentioned in the talk. I had the example of Kevin Bridges, the comedian. He’s from Glasgow, and the auto captions struggled. to get it right - and it was interesting what they come up with. So, of course, a strong Glaswegian accent can affect things (smiles).
Jason Barnard
That leads us really neatly onto the idea that if you make a video, do the transcript. Stick it on your site, then you get both the text - remember, people people prefer reading to watching videos. People talking about this more and more in this industry. Is that a great strategy moving forward?
Transcripts and Video
Jason Barnard
Put a transcript on your site, then you get both the text - remember, people people prefer reading to watching videos. People talking about this more and more in this industry. Is that a great strategy moving forward?
Ahmed Khalifa
It is. I mean, it's a way for you to target both YouTube search and web search. They're two separate things. If you have captions and you've done it correctly, you've done all the editing and it is clean and accurate and correct, then great - YouTube will be able to understand better what the video is about. Then you can use the captions and turn them into a transcript and that can be your blog post where you can embed the same video. So then you target the web search as well. So with one video. You can target both web search and YouTube search.
Jason Barnard
I suppose if somebody exported the auto-thingy - sorry, I can’t remember what it’s called - (captions) and put it on a page and they would read it and they would think, yes, I do look stupid.
Ahmed Khalifa
You’d be surprised. I think more often than not, if you export auto-generated captions, it is not possible to read it.
Jason Barnard
Yeah. I tried doing auto texts for the podcast using Google's language analysis thing, and it was just rubbish.
Ahmed Khalifa
It’s not there for a complete transcript. You're not replacing people who do it manually. It gives you a head start and then you can edit manually. So it will save some time. Doing it word by word - manually stop and type in, play. Listen. Stop it. Play it. - That’s gonna take forever. Although there are experts out there, who can do it quite quickly. They can do it in real time, more or less.
Jason Barnard
I've got a mate who does English subtitles for French films. And he says he more or less types it in real time.
Ahmed Khalifa
People can can do that. It's quite impressive.
Jason Barnard
But you must be typing something slightly after you've heard it and still trying to listen to the current words.
Ahmed Khalifa
And not only that… you may have to put in for example, an explosion in the background, a car crash or whatever. So it’s not just about what the person said, it’s about the sound effects. So that person will also have to think about that. - the sentence ends, new line, then car crash or explosion or whatever.
Jason Barnard
And of course YouTube auto-captioning doesn’t do that.
Captions and Context
Ahmed Khalifa
No, it doesn't. So you've got no idea if there's been a crush in the background. Not at all. And that was one of the arguments I gave in the talk I did today. For example, if you are in a room and there's a background noise of some wind… that could mean anything. It could be the wind outside. A very windy day. It could be the wind noise coming from a TV. It could be wind noise coming down the chimney coming out of the fireplace.
Jason Barnard
Oooh ! Context.
Ahmed Khalifa
Yeah, that’s what I talk about a lot. It's about context. You have to provide context - we communicate not just by content, but with context as well. And that's the whole point of re-touching, rewriting. It's not just the words, you need to really understand the emotion, the story, the context of that video.
Jason Barnard
Oh, that's brilliant. So it's not just the words we're saying, it's the context. And when you watch subtitles films - I live in France and we’ve got quite a lot of subtitled films - you see all these stage directions and descriptions of what's going on.
Ahmed Khalifa
It's one of the most common things people say like “Oh, I hate that channel, I don't really watch it”, but for some reason it's okay for them when it's a foreign film. Okay. Well, it's kind of a similar concept because if you can accept it in a foreign film, then why can’t you accept in your spoken language.
Jason Barnard
And I would imagine that one of the things that stops brands during the subtitles themselves is cost.
Ahmed Khalifa
Yep. And that's a false economy. It's very false, because the are very cheap ways of doing it. First of all, you can do it for free. You can upload your video on YouTube, have them auto-generated and then edit them .
Jason Barnard
Yup. That's free.
Ahmed Khalifa
And there are AI machine that do it for you for 10 cent a minute.
Jason Barnard
Can I just ask a quick question? The services with AI machines doing it for 10 cents a minute… are they better than YouTube?
Ahmed Khalifa
Very similar. I couldn't tell the difference.
Jason Barnard
So YouTube are pretty good within their field, but just not good enough for humans.
Ahmed Khalifa
I just can't see robots replacing human in terms of being able to interpret emotions and context, you know, how can you, how can you expect him to do all the sound effect correctly? I just can't see that.
Jason Barnard
I like the idea of emotions because as well as the words you're using or the… Oh no, that's completely the wrong idea. I was thinking about translations. I was getting very confused.
Ahmed Khalifa
The idea of interpretation IS there. There is some overlap.
Jason Barnard
Oh good! Help me sound intelligent.
Ahmed Khalifa
I can’t do that :)
Jason Barnard
So that the idea that there's overlap between translations and interpreting and sub subtitles.
Interpretation and Accuracy
Ahmed Khalifa
Yeah. People interpret what one person said to another…
Jason Barnard
That was my point. When you're doing the the subtitles, you can simplify it. For example, when I did do the transcripts for this podcast, first thing I did was cut out stuff because it was redundant and change the sentences slightly so that they read better. Is that a bad idea?
Ahmed Khalifa