Live Session (2026)

Hello and welcome back to today's week. Um that doesn't make a lot of sense of audio and video and how everything works. Um as always if you have any questions you know you can always put them in Slack. We do not bite. R rarely at least. but yeah, but if you have any questions today about the things that you watched or anything that you, you know other questions about audio and video and things that or things that we talked in these sessions before, feel free to ask them right now. I don't see anyone holding their hand up, which is totally fine as well. you know, don't need Don't be shy, but also you know, you don't need to ask questions that you already know the answers for. That would be silly. Um oh there Jonathan has a question. Go ahead

Yes, so I was thinking about the 2. 2. 2 pass stop hide and how that works with spinners spinner animations for example when you are for example updating shipping options on a on an e-commerce website and the price is getting calculated so you have an active content underneath let's say the summary and over the summary then it comes an overlay making things lower contrast and then you have a spinner. So how would that work

Good question. So for me let's share the screen here. for me like this would fall under under auto updating content because it's like updating automatically right And and basically this is an essential movement because it like it it conveys that something is happening in the background. So we we think of this as an essential activity both for moving blinking and scrolling and auto updating. Th I have worked with clients who were like, yeah, but we should, you know, maybe maybe it would be good to like be able to pause it and then we usually have implemented that as may more of a user setting. So you say like, oh, you know, I'm I'm I'm I don't want like when updates happen, I don't want them to like happen all the time But in in general this is considered an essential activity. So that would be would be okay. Um I think if you're if your animation, if your updating takes more than five seconds, then I think you have also a different problem, right? So so because Because yeah, I th I think w well this this actually does not have the five second like threshold, but usually we use that for things like flashing and and stuff like that. So, you know, usually the oh here is last more than five seconds. So if it if your update is less than five seconds anyway, then you know we would not in general like fail that. So you know there are there are many ways to look at it, but like it's basically an essential information that this is updating right now. And then also like it it should not take more than five seconds. And if it does occasionally, you know, then that's just how how it goes. Like there's no Well, especially if you have asynchronous tasks and stuff like that. You know, sometimes you're not in control on how long the animation is. Um And what what you can do, you know, f to l lower the impact is, you know, after after a certain time you could stop the animation and have like a text that says like, oh, you know saving still in progress or data still loading or something like that, you know. Um Yeah, because usually you you have the animation to just show that there's activity happening and that it's updating. But in Reality, you know, you also get the situations where something is spinning just for, you know, seconds and seconds and seconds, and then you realize, oh, this like didn't work. And despite showing that there's something happening in the background, there's actually nothing happening in the background. And then you reload the page and you see that like, oh yeah, it has already done it or or something like that. So Yeah, just just having the animation sometimes is not the best way to convey that something is happening anyway, so yeah, it's but it's it's usually not not a big problem. So yeah, we call it essential. although it can be more than five seconds it would be would be considered permissible to have a loading spinner on the page. I'm always a little bit conscious about the fact that like these days a lot of companies are using these like skeleton layouts and stuff like that, which are basically say like, oh everything on the page like blinks and like comes into into view like at some point. Um and you have the skeleton and you know then one thing loads and the other thing loads. And I think that's very busy Um and if I encounter these I usually say like hey just hide the whole interface and have one big spinner instead of like, you know, loading stuff asynchronously like all over the place because it it gets just too busy and like most users all also or don't say like oh the like the the right side widget has loaded, I click on it and I I navigate away from the page. You know, I think that's that's rarely happening in the in those circumstances, but it really depends on the the application and on the on the website that you're you're working with. Um yeah. Okay, thumbs up in the chat as the people say. I don't know if the people say that. Um awesome. Any other questions? Giving you a little bit more time. Alright. Maybe maybe we come up with more questions over time. So I want to To show them some things because like I think despite audio and video being, you know, something that we interact with quite a lot it can be pretty abstract from like going from like okay what are the requirements to like what are best practices that I can do right now and so I Looked at a few best practices and l and and a few good examples to show you today Um and my my friend Nick Steenhout just released his new podcast, The Work Between. Um and it's basically a podcast that, you know, wants to highlight like creative endeavors of people with disabilities, which is like an aw awesome awesome thing to like to highlight. Um and he published a website and Nick is like a very very proficient accessibility person so you know if he does something it's really good and this is no exception and so so I thought like it's a good example to show because it's recent, it's it's new and it uses all like all the tricks in the books that the book that you can have. So this is the the episode and there is a summary at the top which is nice. but of course there's also transcripts in there and you know and then show notes at the end. so show notes would be where you have your like profile of who is on the show and like links to like further other things like his Ellie Rules podcast which is also very good. Um so you get like these things in the in in the show notes that are basically general purpose for everyone. but then you also have like the accessibility specific features. and what he uses here is a software called Able Player. Which is a free open source thing to to embed. It's very ugly and like pretty hard to use, but it has like some amazing features that we're gonna gonna see in a second. So First, you can, you know, just play the episode and I will just mute mute it because I think that makes most sense. Um Cause then I can talk about it as well. So you you have the normal like you know place sound and then you can show captions. and normally you would say like captions with Um podcast is a little bit unconventional. I think you have to pause and play. Showed the caption just before when I tried it, but it doesn't now, so there used to be like a little caption thing at the bottom. Let's tr see if it works now. Oh did it not work because I Switch the down the volume? No, it just works. So this is this is basically captions like you would have on a video. And for some people this might be the preferred way to to see it. and I I talked to Nick and he was like, hey, I have this like captions and then I have the time transcript, which I will talk about in a second, and then I have also the static transcript. And I was like, yeah, just keep all of them because like s other Some people will prefer one over the other, so you know options is good. Um then here you can listen faster and slower. to the audio which can be useful features for some people with cognitive disabilities so there are some folks with ADHD who like need quicker speech because they just can't follow if it's too slow their their mind wanders and then they go like, I don't know. And i the other the opposite thing is someone who might not be acquainted with the language fully, you know, or might be have developmental challenges in like understanding speech. so or Or maybe bad hearing, you know, it might all be things where it's like, oh I'm gonna slow this person down. And that's ver very personal preference and you can do both here. So going down to fifty percent for example. Um yeah, and then here 's the time transcript here, which is basically like The text of the whole episode which is relatively slow because r a relatively short because it's introduction episode Um and it highlights the the sentence basically that he currently says and then it like jumps to the next one once he's done, which is going very slowly because I slowed down the speech. That makes sense. Um but you also can like click on a sentence and then it like jumps back, which is which is useful and and you know some people really prefer this type of time transcript And then there are people who go like, oh, I heard this podcast, but I want to like talk to someone about some content in it or I want to like just have y you know, a read through or something like that. Or I want to search in it And that's basically where the static transcript comes in. It's all the same text, of course, because it all comes from the same file or from the same data. But that's just like paragraph after paragraph of like what was th said and I assume once he talks to a guest it will say like, you know, Nick and like guest colon, like probably guess name. I would be would be polite. And he's Canadian, so he's he's a very polite person. And and yeah and and so you have these like different modes of like using the same content over and over and that's not a requirement for every everyone and every like single you know publication but it helps to make stuff accessible to like a lot more people than if it was only audio or only you know, only writing or something like that, multi multimedia makes a lot of sense in these cases. Um and if you if you're thinking like, oh but you know, things like Apple Podcasts or Overcast or like probably also Android audio podcast players they now create their own transcripts That's generally like it's a great accessibility feature that we have that and I will talk about like automating automated transcripts in a bit and captions as well but the problem is that they have a significantly higher error rate than like humanly transcribed text and that can be problematic and I mean this it has this word which is an like native american you know c Native Canadian. Um a word for like where this territory is and who the people are and I would not know how to like pronounce it so I will just go to the time transcript and click on it and have Nick read it to us. Probably means I should share sound with you as well Oh, I have to switch on volume on.

Before I get into what the show is I want to say that you're not going to be able to do that. Before I get into what the show is, I want to say whose land it's made on This podcast is recorded on the traditional, ancestral, and unceded territory of the Stolo Coast Salish people. the people of the river, whose presence in the Fraser Valley stretches back thousands of years. I named this

I'm also annoyed how much nicer Nick's voice is compared to mine because he has like this really soothing, soothing voice. So yeah, so Stolo. is apparently the how how you pronounce it. But you know, I I I would not know it from like just how it's written here. But also like I think Transcription software would have like a really hard time to like transcribe it because it's not in and and that goes also for words in different languages or if you have loan words from different languages, it's not in the training data. So you get into like I could I could see that this would be transcribed into solo coast. Which like that's that's not a thing. That's a Star Wars character. And Star Wars characters are much more prevalent in like training data and stuff like that. So yeah, I think I think this is actually a very good example where it's like that's why it's essential to when you do transcripts and captions that you have a human review it. Which fair point I don't do for these that I publish. Um because we we have limited time to do these these trainings so it's more like a search and like r research thing but But yeah, it's it's it's really important like if you if you do something like this, like in professional capacity you should always make sure that you have good humanly reviewed transcripts. I I I don't think there's like it's a bad thing to like have the grunt work Done by an LLM that does the transcription. I think I I see a lot of value in that. Um but in the end you have to listen through the whole in this case podcast. like if it's if it's an hour podcast, you have to put in at least an hour to like listen through everything or like, you know, half an hour like listen it at it at two X the speed and just read through the transcript and make sure that it works. Um because that's Unfortunately the only way to make sure that it really, really matches what was said. And you know, and it has a lot of problems like automated captions with accents and other like non-standard speech. So if you're you know slang speech or if you have a certain inflection it can it can be really difficult. I'm lucky my German German English is like apparently very easy to transcribe. So I I usually have like a relatively high hit rate. But Nick, who is Franco-Canadian and speaks English as his like main language or like f for this podcast and stuff. He always like sends me like terrible Siri transcriptions on his phone and like it's it's it's it's all really bad. So it really really depends on the training data and that you are fitting. into a certain mode which is usually you know white and male and like stuff like that and then you have a good time with automated transcriptions and otherwise it can be really dicey. So So yeah, you have to you you know just just be careful because it can be sometimes very Yeah, v very hilarious to be to be honest. Like there can be can be hilarious issues. But yeah, that's that's the new new podcast from from Nick and he puts all the accessibility effort into it. Which is why I want to showcase it. I also want to talk a little bit about making audio and video media accessible. I think I talked about that in the videos. I have to admit that I didn't like closely review them. But I think the main thing that you should take away from this resource is the audio content and video content part because basically how you treat audio and video is or can be two ways you can either like Just talk and just you know let it all out or just film something and then afterwards describe. or transcribe or capture what has been happening in the in the video or on the on the page or whatever. Um but it's much less effort to go in and do it beforehand like plan your video have maybe like either a rough outline or even a complete script of what you're going to say before you record because then you can like use that script and feed that into youtube for example and youtube will just align your script that you read into the camera will align that with the timing of what you say And it will not change any of the words or anything like that. So it's a very easy way to say like, oh, I have these like, you know, seven sentences to say and I Write them down before then I put them into YouTube YouTube aligns everything and then I'm happy with the outcome And it's correct because I just read it from the page. so that's that's a really good way to do that. Also important that you have things like a good microphone and that, you know, you don't have a lot of background noise which like you know, makes makes it not only harder to transcribe but also to just hear what you're saying. So people with like you know, low hearing and stuff like that, they will have a much easier way if you have good a good recording setup. That doesn't need to be like a you know a a fancy like studio microphone. like some some of us do. Um but you know you maybe a headset, maybe just make sure that that you can be heard well. Um you should speak clearly and slowly, which, you know, is always a challenge because you d hear yourself differently than other people hear you. So yeah, it really pays off to like to practice and to make sure that like you really open your mouth when you're when you're speaking and mumbling everything, you know, under your breath. because that's that's it's getting challenged. Um Use a clear language for a script. as I said, it's great to have a script and just read from it. Um and you know you can improvise as well. Like you don't need to like stick to every sentence and every letter. Um I think it it it helps often. But if you use a script you also see like where you can like improve your language, where you can make things easier, simpler to to understand. And that's a great benefit. And then you can also include in your speech things that are visible in the video. So You know, sometimes I see videos where someone is explaining something and then they just say like oh and then Let's click this and that and then this happens and this and that is not described, so you don't know what it is. So in WCAG patterns you would need audio description for that that basically inject jacks somewhere and says like oh it's the settings and then speech buttons. And you know, if you have a script and it's just like, oh, I click the settings button and then I click the speech button then you don't need to provide additional audio description. That's like one of the big shortcuts of like making sure that you don't need to do post-production as well. Um yeah, and then there are like on the video side there are a lot of things that are Should be should should be pretty clear, like a avoid causing seizures. making sure that you can see the speaker clearly that you know You don't you're not you're not hidden behind something. Um that's that's a good thing to do. make sure that if you're using text it's readable. Um and you know if you're in some circumstances you need to have sign language, error audio descriptions and other visual information. on the page. So so for example when you're planning your shot you know and you know that like oh I want to have like captions on the bottom of the screen like it's you know common on YouTube or on streaming services then you want to make sure that like the person that is on the screen is framed in a way so they don't get overlaid by the caption. Um and that's like the whole like skill of themselves. And then you have, you know, you might have an optional sign language interpreter that is like on the bottom left of your screen and it's always there. And then you know you have to make sure that like, oh, you don't interfere in that space. So lots of things that you can think about. And so I think this like planning page making sure that you you have your content properly planned and then like properly executed like that goes a long way Um and then everything else becomes relatively simple, to be honest. Um I want to go next. Yeah, I think I want to like brag a little bit about my own YouTube channel, which I did in 2021 and Just you know, thing is I put together a couple of videos and one of the first ones is this big WCHAC overview video, which premiered in September 2020 And now has 34,000 views, which is like a lot of people. And like I'm really happy with like, you know reaching so many people. So that that's v very well where you can have like a big impact on on a lot of people right doing that. But I did a few things. So the first thing is so I didn't didn't have a great setup for this. It doesn't matter. Um I wrote a whole script. So I I did sentence by sentence, , wrote a whole script, put that on my iPad, and had like basically a teleprompter underneath the the video or the video camera so I could like look into the camera more or less. That's all the first try, right? Um and like read every sentence. and then I went into YouTube 's user interface and I uploaded it. And then on under subtitles , which is where you find all the captions. So they have English captions and then they also you know you can have multiple languages in there as well and you can see that there are timings in there. I have not put those timings in. I would be way too w w w yeah, and we wouldn't just wouldn't do it. Um oh that's not what I wanted. But you can upload a file with timing or without timing, and if you do it without timing, it will just sync your subtitle to your video 's audio. and that will take a while, like, you know, half an hour or something like that. But then you will have like a really nice formatted or at least like draft of your captions, you know, and then you can go through and hear this is pr pretty much the most you want to have as a caption, right? Because like if you have too much text It's too much to read and then like it gets harder to think up. Um and you can just go through and like split stuff up, you know. You could like If if you want to split this up, you press the plus key and then you insert it and you adjust the the timing on there as well. So it's it's very, very good to do this. and I think unfortunately YouTube is the only place to do it. I don't think there's a good open source way to do it or like I I don't know. Um and and yeah and you can do things like, you know, here I put in like a carriage return. and and the line break just to make sure that like the R here is not on the upper upper What is it called? Line. because like it f it felt like it's better it's easier to read if if this is in like a new line. So I did that and now I split it up. I didn't didn't want to do that. Um and I think you do shift return to split it up. But yeah you can you can you know you can play around with it, make sure that like it lines up nicely, it's not too long, it's not too short. Um and that makes a lot of sense to me to do it like that I don't want to have changes. So yeah, and and and that was like a really good way to like as a single person, like doing all the script writing, recording. Um transcription theoretically, and putting stuff online, cutting, cutting it, editing it, everything as one person. That was a very straightforward way to do it and was still like So much work. Like you know, just writing the script usually took me like this was a long one, like 14 minutes. The other ones are all shorter and guess why because like writing a good script or even like a mediocre script is like a lot of work. So so yeah, but I wrote the script in one one day and then I basically recorded it the next day. Um and Yeah, did did a lot of cuts and stuff like that and worked pretty well. Um, but in in the end was like a lot of effort for what what I got out of it But but I think if you are like in a professional context with like, you know, having a little bit of support, that is certainly something that makes it much easier than like trying to just talk into camera for like, you know, extempor extemporaneously for some time and then have it transcribed either automatically or by a person Um and and yeah, and that makes makes most sense. And then I did put transcript on my website as well. And this is the other thing, you know, I'd like to to own my content and and at least have like some control over it. So I created the transcript and I exported it and then basically as an HTML file with headings, with you know subheadings, with you know lists and stuff like that. So it's really easy to read. That took a lot of like took a bit of effort to make it like Look nice, but I think it's also worth it to make it like a really really nice experience for people who cannot see the video Um hello. Oh, this still works. I I was like, oh, this is broken, but no, it still works. Good, good. Um Always surprising myself today. Um yeah, but this is basically the approach that I think makes the most sense. So you you you go in there with the plan, which is not Often not how most of filmmaking goes and even if you're doing like informative stuff You know, it's often like, oh, we're on location and we're just filming what's happening and then we do like a voiceover and then you know we may might have an interview. And of course you don't tell the interview person what they have to say. That would be rude. Um and so yeah, it's like it's like you have this like, okay, some things need to be transcribed after the fact. You know, you can't be you can't have everything pre-planned. But if you can, it makes it really, really much easier. Um yeah. oh first welcome to W why are you autoplaying? You should not. Um Yeah, we did the the webinar the other day and also has like proper captions, at least they look proper to me. Um exaslab did the transcription so it's really a good good example as well i can put that into the into the list of links as well. I think it's a good overview of over the EN. I don't think there's anything that you haven't heard. about it because like I'm I'm just reiterating but I will put that in as well. Um There is the BBC has a whole long list of subtitle guidelines. I don't want to go through most of it because it's very detailed. Um and they also have different things for online and broadcast, , which is interesting. Like I I think this is more interesting than like really something that you will use a lot. Um but they have things like du Preserving style like using mother versus mum. So th there was a long time f especially for TV. captions were always abbreviated and like, you know, more s summarization instead of verbatim the text that was spoken. and this has now changed a lot. Like you would do you you would be more trying to like be as close to the spoken language as possible because it's just it just makes more sense. but there was the well, s some people suspected that accessibility would other that people who couldn't hear would not be able to read long sentences and stuff like that. So they tried to make it really short. Yeah, the keywords that can be easy liberate read, so you can basically sync up while you're reading it. that makes a lot of sense. you don't want to edit out strong language, so if someone swears in your TV show, they should be swearing in the like captions as well. Um just Yeah, just just make sure. If it's bleeped in the sound as well, then you also need to reflect that of course in the captions. So always try to to like basically captions what i what is in the visual or what is in the audio. Um Yeah, and then you have line length. This is what I was talking talking about, like you for broadcast platforms, so this for your TV stuff, the length is a fixed 37 characters and then for online you can use sixty eight percent of the width of a landscape sixteen to nine video or 90% of a 423 video or 90% of a square video or ninety percent of a vertical video. So you get like okay to avoid like long line length, they try to like box that in a bit. Um subtitles should contain single sentences. you don't want to have too many lines. I think they have one or two. That's what they say. And this is like English specific I think, you know, because in German we have like these long words, sometimes you go you rather go to three lines instead of like making too many changes of the text. So it really depends on your language. Um Yeah. Usually it's centered as well. du doo, short sentence, long sentences. So they go into like splitting it up and then also timing like how many seconds per words you should have and stuff like that. And that is very like detailed stuff. Um They also go into identifying speakers, so you can do it by color if that works. Um and this should be used in most cases. So Often you see like one person reflected in white and another person reflected in in yellow or like in green and that helps to to d identify different people especially if they are in concurrent situations like you have a very quick exchange and one says, No, yes, whatever, you know, you have like these back and forth. Um then sometimes you even use like a placement of the the text to indicate like who says what. So there can be a lot of a lot of interesting things here. Like here they do, did you see Jane? I thought she went home. So this is a response. Yeah, I think that's a little bit like up. Yeah. I I mean I I think they d they don't have enough contrast, so that you go into like whole nother set of issues there Um but it can be, you know, ca can be useful like and and they also want to have overlap as much as possible just to make sure that your your eye doesn't need to like fight find someone, you know, elsewhere. Um can use dashes and stuff like that. Um yeah, and then you can use off-screen voices, you can use arrows. They have a lot of details, as I said. on there. Um and white on black is the preferred color for the BBC, which makes sense, I think. They also talk about typography and positioning and intonation and emotion. So if you say charming as a sarcastic statement charming, then you can use the exclamation mark at the end or you can use a question mark for sarcastic stuff. You can also use uppercase for stress. and so on and so forth. Like yeah, there's there's a lot of stuff in there. Um really, really I really think it's an interesting document because it's so detailed. Um it goes into humor. Yeah, try whenever possible to keep punchline separate from the preceding text because you might skip, like your brain might just skip to the punchline first. Um and and stuff like that Yeah, I think that's basically it. There's also file format stuff, of course, because like It makes makes a lot of sense. I mean, currently basically everything is SRT, , but it's not Not for the BBC because they have STLs and they also have XMLs for online. It's like it's a little bit messy. But that's that's how it is. Um couple of years ago W3C won an Emmy Award for introducing the TTML format, which is a text-based subtitle format. So Technically because I worked there I have won an Emmy. That's fun fun fact. I I think it's fun. Um So we have two things in the in in the chat. Jonathan says I've had a 50-minute video that I've had to manually add captions for. It took two hours. Yeah, if you're not like in the training or you don't do it often and you have a video that has not That is not based on a script. It can easily take two hours, especially if there's like sounds or or background and or it's hard to understand. Um it can be like a real well there are professionals for that for a reason because it it can take a long time Do you have any examples of good ways to implement audio descriptions? It's a double-A criterion, but I feel like I rarely see it online. So, great question. Um Every streaming service has heaps of audio described content online. If you look up something in Apple TV, if you look at something in Disney Plus, any any of those series that are modern streaming area series, they usually have audio descriptions and they are very good. If I can make a recommendation for audio descriptions, I very much enjoyed the Netflix Daredevil Season 1, which is on Disney Plus now. Um it's really really good, really great audio descriptions, and because it like it's about a blind man being a superhero, it's also like fun and and makes sense in the in the context. Um but there are there are a lot of good audio description content out there there if you if you look for it. So usually you go into the subtitle menu and there you also have the different languages and then you go English audio description and if you're lucky there's also one in your language. But that's lot rarer than it should be. Good ways to implement them are basically the only thing that you can do is find spaces in the audio of the video and then record audio that like describes what's happening on the screen. if that's important information to know. Um, if it's not important information, you don't have to describe it. Like if it's just like a person like this what we are doing right now, I'm just talking to you and you know everything that happens on the screen is mostly optional, not not everything, then you don't need like a separate audio description for, you know, that's why I always try to describe what I'm doing when when I do it. So we avoid audio description. And that's the ultimate best practice to do it. Like just be aware of like what's happening visually and verbalize what's happening. Um yeah, but if you yeah, if it's just b-roll iCAD in the background of an inform informative speaker voice, no AD is necessary. Yes, that's correct. Like you only need audio description if what is described is important to understand what is said, basically. If everything is said, then you don't need audio description. Yeah, that's that's how it that's how it goes. Um and also like if you have you you need to implement audio description properly, you need space in between the the narrative or like dialogue or something like that. That can be hard in web content because it's so compressed. In these cases I recommend doing audio description before and after the video. So you you would say like for the audio described version you have you have someone who who who says like who sets the scene basically and then you let it play out and then you describe the outcome at the end. That can be a way to do it. Um so I can quickly show The Apple video designed for every student. Um so and this is of course an an Apple commercial and it it's a song called I'm Not Remarkable and they have an audio-described version of this song online so there's singing and then there's also audio description in the middle. I can show that. I think you go to Audio track and there you have English descriptive. So that's how YouTube does it. And not everyone can have this because YouTube is yeah Does stuff. So let's let's just listen to for three minutes of a song.

A blind student holds an iPhone. Dear student, congratulations. You have been accepted as a new guy. Students cheer. A student signs in American Sign Language. Subtitles. I'm going to college. I went into college. A student taps accessibility in their iPhone settings. On a college campus, the blind student holds an iPhone and says, Starting route to lecture hall.

I'm not remarkable. I'm just finding my way.

Turn left.

It's not a walk in the park. But maybe I got the grades. Don't wanna be in my

A student types on Mac with her toes.

If you wanna be inspired, there's a library down behind

A student uses magnifier on Mac

You know if admiration sometimes smells a bit like peer I can be strong, I guess so wrong Ah I don't always talk so pretty I'm not remarkable Not heroic arbre. I've got no super powers. I can't even do this stupid equation.

A student draws an iPad.

We all have to

She holds Apple Pencil with her toes.

Funny and new face, I've got a place, I've got a life, I've got a name. And I need stuff, and you need stuff, and all our stuff is not the same. And some days I get so tired, she gets grumpy. Yeah, well I have a special need, I'll call it.

Students hold coffee and bounce in their seats energetically. In class a student uses live captions on iPad. A blind student plays piano.

Do I have to be remarkable? I'm just having a try. I'm not fragile or indestructible unlike everyone I sometimes cry and I'm not like the rest But then the rest of you aren't like the rest is

the student's party I'm not a scene, I'm not a scene, I'm not a I'm not a winner. I'm not born, I'm not less. I

guess you're average human.

A student uses hearing aid on AirPods Pro. Confetti falls at a house party. Students with and without disabilities dance. They wake in a daze the next morning. How long

remarkable

A row of students kicks out their feet.

I've got a face, I've got a place, I've got a life, I've got to find it.

Outside, students dance. Students in cats

And we will fail and we'll get hurt, we'll prevail, and it'll be joyous, and it'll suck, and I'll be lonely, and I'll be loved. And on the way we will discover we're not the same. But in all the effort, there's a life out there that I'm with you most to miss. I'm only remarkable because everybody

is The Dancing Students Pose They smile at each other, then disperse. Titles. Designed for every student. The Apple logo.

So yeah, this is this is like a really good example of and a challenging example for audio description, right? Because you cannot describe every single thing that's going on on in the video. There's just too much going on. Like there's the dancing and the singing and the you know the people looking all different. Like you cannot describe everything in in detail so you want to give like an impression of what's happening and I think that's also a very good good thing to like keep in mind that like you know you don't need to be super exact, but sometimes it it makes a lot of sense to be exact. Like for example at the end there's like the the mission statement and then the logo and Saying that makes a lot of sense because then you know actually, you know, who like produced this video, and then that makes a lot of sense. So so yeah, that's That's what I recommend to to do or or to look at. So it's yeah, it's it's it's more an art than a science, I would say. Um but basically If you have things that are less artsy like explainer videos and stuff like that, then you should just make sure that everything that is visible and is important is is properly explained and you know if you close your eyes you would see it. in front of your your mind's eye if you have one. Um that is probably the best way to describe it. All right. I have one little thing I think that we can look at and then I would also be Pretty much done, I think, for today. Awesome. Good that it's helpful. And that is Mac Whisper, which is the software that I use to do the captions for this lecture and like all the the other things that I do when I need captions. Um and I'm gonna share the screen sh share that as well because it's it's a really nice software to to use for these for these things. Um It's called Mac Whisper. It uses like different LLMs. So I can't can't show you the home thing because I have secret stuff in there but basically as you can see I'm just re actively recording this this meeting because I wanted to to to show this as an example so I'm gonna press stop I usually do the description for the for the chess transcripts website on on my website and for the captions on the the videos that I upload, I do them afterwards from the video so I don't I don't usually have life captions. That also has to do with like I don't want you know someone saying something getting like into a transcript right away like if there's something that you know is like I don't know we get someone doing like a zoom bomb That used to be a thing, where people would just randomly get into into Zoom meetings and be like assholes. Um Oh that that's gonna be fun in the transcript. Um but you know, and and and if something like that happened I would cut it out first and then would do the transcription because it has nothing to do with the with the thing. So that's why I don't do live transcripts usually. But basically what this now did is it recorded my speech and your speech if you spoke up. Um and this is all locally on my computer because I'm super uncomfortable with like sending voice data prints to Other companies. I think that's a little bit of a of a bad form. So this is all like locally. Um and so it uses in this case the parakeet transcription and it takes some time as you can see it goes through but it should in a second be done with it like can I already open it Yeah, and you can see it took track from Zoom and from my microphone and should mesh them together. It should be done in a second. life life stuff. So there we go. Um yeah and then we have all the the things in there. Like it even says recording in progress, which is what Zoom that and then we can like see all the things all the things I just blather around. Um I wonder if it did pick up the Oh yes it did actually. It even did transcribe the video. I didn't know if that would come into the thing. And here we have speaker six is basically the audio description. So now I could go in here and I could say this is audio description. Um and press return. And then that is now audio description here. And then microphone is me. Oh Eric. And then that is all in your in the list of things, you know, and then I could could put your name in and and stuff like that. I usually don't do that. I just say speaker one, speaker two, pay speaker three. And as you can see, it's pretty much pretty much spa bang on when I speak. which are which is why why I said like my my language, my speech patterns are apparently Very g easy for for these transcriptions things to to transcribe, but like sometimes it's much more difficult. Um and then you you know, I had like I said WCAC sometimes and it just goes like all over the place with like transcriptions. So basically have to like review everything and yeah and that's that's what's happening so you have the transcript here you can also show it in segments so you you see like basically every single like part of the sentence that it transcribed which can be useful And then you can export that into different formats. So the subtitle format that you would use usually is SRT. Um because it's the most common one that you YouTube uses it. The chess canvas uses it so SRT is the thing and it looks like this so you get like you know every section has a number and then it says from when to when and and then what is said. Um and you can include the speaker names which then it adds to every line which is annoying so I never do that And you can also also do favorited segments only, which I have never done. I don't know why that's useful. WebVTT is a different format, it just has doesn't have the one, two, three before it. Avid never used that, so I don't even know what that is. And then you can have like simple markdown. I don't know who uses that. But that's usually the subtitles that I export. And then I also export the transcript and I do it in HTML. And then I just throw it on my server and the script that I wrote takes the transcript and like makes sure that everything works and it has the proper dates. Um And things in it. And that's like that's how to do it. I always also add the show timestamp, so you can have a little little like checkbox show timestamp, hide timestamp. for it. Um but yeah it's it's pretty straightforward and these transcripts these days are really good and really quick. I mean you know this was a How long? Like forty minutes of of mostly me talking? Or fifty-six minutes of me talking and you have seen like how quickly it did it, like in your side of like two or three minutes and the the thing that I'm using is the parakeet version three. model. I've I find that this is like super quick, does not take a lot of like energy and stuff like that. Should also work on most less powerful models. I mean this is a MacBook Pro from 2001. So it has, you know, there's It's not it's not the the quickest thing, but you know you can also use other models and you can see if they are good or bad. Um I think there is a non pro version that is just free and then the pro version is like fifteen euros or something like that. So it's not It's really not bad and all the models are free. So I think there's no no reason to not at least provide some kind of like transcript for your media, right? Even if it's not perfect, which is also fine. Like don't let perfect be the enemy of the good. Like if you only have time to check for like major errors then do that and you know pro provide some kind of transcript. I think then it's like more useful as like a transcript than like captions because yeah they can can be just more distracting Um yeah, and then I want to show like another thing because we talked about or I talk. Oh

good.

I talked about the captions and how they can be formatted and here I have like one of the best I want to say captions that I've seen in my life. because how on how the captions reflect the speech that is happening. So this is from a three-year-old Tom Scott Plus video. Um and basically they do the hot sources thing where you know you answer questions. Like they did it in a quiz and if you ask Answer wrongly, you have to taste hot sauce. I don't know why you do this. Like it makes doesn't make a lot of sense to me. Um But we will see that Tom Scott has answered the question wrongly and he need he takes some hot sauce and then you know take a look at the caption and what's happening there. Ha!

Yeah, it's just under five minutes. That's it's over for me. It is. It's not for me. That is a significant ab I'm I'm you've got it going all out. You've got it. You've got to I can't I can't go half-assed on this, can I? No, I guess not. It is YouTube after all Oh, seems pleasant maybe. I see why he said it's just wet. There's nothing

Oh seriously

I was almost disappointed. Sugar go and have a walk by the paint. Oh That takes ten seconds and then That was interesting. Your reaction at the beginning was almost disappointment in e

So I r I really love the ow with the captions here in the middle. Um really makes a lot of like sense to do it to do it like that. But like You know, you have to be a really good captioner to like not only see this and hear it and then like use different sizes of text to like really embody what was happening in the audio. So so I I think this is always a great example to just show like Yeah, usually you want to be like very strict and very business-like, but you can have fun with these accessibility features as well, right? Like it's not there's not like one way 100% to do it. Like people who need to rely on captions also can have fun. And yeah, that's why I like to show this example. Um and not all captioned software can display can display it like that. So you know you have to to look at your format and like how it works. But yeah, I I thought that was a very funny thing. I also saw that like Tom Scott as the host of the program has a white text and then the guest has yellow text and stuff like that. So very almost BBC like. Um yeah, and really makes a lot of l lot of sense to like be like you know have a feeling for like how those how how that content gets created and what you want to say with it for that Alright, that was the last thing I had. Are there any other questions that you want to ask? This is your chance. Doesn't look like it. So thank you very much. See you next week. I will double check that it has been published because I know that there have been some problems. with that but yeah thanks for being here and see you all in seven days bye