Follow


I suggested that we should have on the stream I'm on, Sword and Quill. The solution we arrived on was WebCaptioner.com, since there's multiple people on the stream and WebCaptioner would separate them. So far so good.
But that service uses Web Speech API. And only has that API. Even other Chromium browsers, like , do not, as seen in this issue:
github.com/brave/brave-browser
Basically, there's two reasons why Brave doesn't have this. 1. Because they would have to pay for the right to use that API, and it is quite expensive. And 2. Because what this API does is send the audio to GOOGLE'S SERVERS where a transcript is created.
1/2
@accessibility

So, because I do care about making the world around me slightly more accessible, and because this is the best solution we've found so far, I'll have to install Google Chrome and and reconcile myself to everything I say being sent to Google, which I've so far attempted to avoid.
This is what most proponents of miss, as well as most people promoting , , , , -- whatever the hecking heck you want it called.
Privacy and accessibility rarely mix. Open source and accessibility rarely mix.
That's why my friends who do bother to mess around with and post so many rants about most desktop environments not being accessible. That's why gets constant praise for , where as all the solutions on lag behind.
I wish I could just ... go back to not caring.
2/2

Show thread

Update on the caption situation: The captions did more to help us feel better than actually help anyone. Sometimes the speech recognition was pretty good, but at other times it descended into complete nonsense.
Amusing? Yes. Useful? No.
The fact that most of us have English as a second language and so don't have one of the "common" accents certainly didn't help.
still has to be done by people. This is another good example of this. That developers cannot just throw everything at neural networks so the rest of the humanity does not have to be burdened by our existence. It doesn't yet work for image recognition, and it doesn't work for speech.

Show thread

And yes, I'm being a damn hypocrite myself, by constantly asking for image descriptions, but never bothering to caption audio. The first takes less effort, sure, but both should be done.

Show thread
@Mayana Just wait till you encounter HCaptcha. This is nothing compared to it.

Yeah, accessibility and privacy don't really mix well, especially when you are limited and you either have to buy into an ecosystem, or miss out.

@erion Oh. I have, I have. And when someone posted a toot about people needing to switch from Google Captcha, I tried reminding people to please not switch to hCaptcha instead. Some asshole revived that thread just yesterday to do this:
uwu.social/@easrng/10568573101
Completely ignoring everything I said and just whining on about Google. Why the fuck did I need to be tagged for that, then?
It basically sent the message of "I don't give a shit about accessibility because Google is worse" and look ... can't they both be terrible? 😩

@Mayana How nice. HCaptcha and Accessibe are the two accessibility nightmares today.

@kravietz I have not used it yet. I'll get back to you once I've done so.
@erion

@Mayana @erion

I have been using it for a year or so - although "used" is a bit of an overstatement since it just sits in my Firefox add-ons tab and does it job without any intervention. Which means I very rarely even see hCaptcha prompts - as opposed to reCaptcha which routinely pisses me off beyond imagination as Google seems to be making it extra annoying when you load it from Firefox.

@Mayana I use an iPhone because it’s what’s compatible with my hearing aids. I stubbornly tried to use LineageOS in the first months after getting hearing aids, but it was such a hassle that it was mostly unusable.

@lilletale @Mayana I'm curious to know what you found hard to work with? I might be able to take back some of your suggestions.

@objectinspace @Mayana The Bluetooth connection was rickety af and would seem to space out for extended periods. I had to wait minutes for the phone to even connect to the hearing aids, and changing sound settings took too long, forcing me to skip to the back of the line because I couldn’t hear the cashier and similar problems. With iPhone I can change the sound settings when I enter a room and be able to hear within seconds.

@lilletale @Mayana That is too bad :( Do you think it was because of Lineage/the phone or a problem in Android itself or the accessibility service?

@objectinspace It could have been all of these, and probably was a combination of several things. I didn’t spend too much time troubleshooting because it was affecting my quality of life and ability to function, so I looked up what works instead and swapped to that.

@Mayana "That developers cannot just throw everything at neural networks so the rest of the humanity does not have to be burdened by our existence." yup, that's exactly it! people don't wanna bother with us but if something bothers them, then suddenly it's a fucking problem... and so, they don't want to be reminded of our existence so they just throw us their failed toys whenever they remember we exist.

@Mayana
Yeah, it's a hard problem.

We're demanding SciFi-level captions and people don't have the infrastructure to do that. They could pay professional stenographers but that's expensive AF. The solution? Submit it to a megacorp's servers. They got the tech to make it work, but sheesh. Then as you said it doesn't really work.

I don't think accessible live captions will be available in years; maybe even decades - at least not until hardware-based AI implementations (e.g. neuromorphic chips) become mainstream.

But you know what might work? Volunteers. Maybe not for live streams, but if they can caption and submit, then an editor makes sure the captions submitted are accurate.

However, we need a specialized kind of software to make that work, and I don't think there is one yet.

Indeed it's a hard problem.

@yuki Yeah, you're right. Transcribing requires hard work and time, and that's why it is so expensive. I've been comparing it to image descriptions, when really I should be comparing it to audio descriptions of movies.
It's quite odd that there aren't any platforms for this -- surely there should be something after all this time? Maybe at least small niche things? But then again, seeing as Google decided to remove people-submitted captions on YouTube, making a big platform is not really an option; the big players don't give a shit.

@yuki Actually, I have seen a few small-scale attempts around the internet. For example, the Nerdfighteria Wiki, where volunteers make transcripts of Vlogbrothers videos. Those two have managed to bring together a really nice community in general.
But that's just a wiki for one channel. Still not a volunteer transcription platform.

@Mayana @yuki I don't know how good it is (I don't think it can help for live), but this thread might interest you regarding what Laura from Small-Tech has been using: https://mastodon.laurakalbag.com/@laura/105639494718140594

@silmathoron Oh, thanks for sharing! I can send this to Zein to possibly add to the podcast version.
But to be honest, depending on the price, this may not be an option for us. This is only a tiny D&D stream/podcast, we're not that interesting. So unless someone who was hard of hearing actually showed interest, I'm not sure how much we'd all be willing to pay for ...
But then again, such people aren't likely to take interest in the podcast *unless* it has captions. And from my own experiences with accessibility issues I know that most disabled people don't feel like speaking up for every tiny thing. So ... we'll see. 🤷
@yuki

@Mayana @yuki yes, I understand it's no simple issue... I just thought it could let you know of both people trying to make their stuff accessible and of a potentially interesting tool...
Tough, as with many other issues, I do think that this should be solved by governments considering this a public-interest technology and developing open-source code that anybody can use.
Of course, it won't be perfect but if anyone could make an immediate transcript, though imperfect, it would lower the barrier.

@silmathoron I understand that! And to be clear, I am very thankful you shared this. It'll no doubt become useful sometime; if not now, then perhaps later. 🙂
Agreed, that would be wonderful. Perhaps someday ...
@yuki

@Mayana @accessibility yeah sadly this is indeed the case. My latest moment with this happened yesterday when I found out that the new iPhones have a special action in Voiceover which helps you find people around you and can lead you to them and describe them roughly. I wish this wasn't the case but at the moment, Apple Leads mobile accessibility by such a huge margin and that not even just for using the device itself, but also the environment around you.

Sign in to participate in the conversation

A fun, happy little Mastodon/Hometown instance. Join us by the fire and have awesome discussions about things, stuff and everything in between! Admins: @Talon and @Mayana.