El Reg shows you how to run Zyphra's speech-replicating AI on your own boxTobias Mann (The Register)
reshared this
reshared this
Peter Vágner reshared this.
Peter Vágner reshared this.
Yo!
I have been building a universal fedi client called #DhaagaApp
It supports all 3 biggest decentralized protocols (#mastoapi, #bluesky and #misskey)
It has pretty timelines, unique client-side features and a modern, aesthetic design.
It's available on Play Store and #IzzyonDroid
Do have a look, and share with your friends if you like it. ^^
I also post frequent devlogs and am always excited to receive feedback.
suvam.io/dhaaga
github.com/suvam0451/dhaaga
Dhaaga - An indie SNS app that blends sleek design 💅, useful features ✨ and fun ways to discover and connect 🎉 - suvam0451/dhaagaGitHub
reshared this
This demo shows what could be the future of GNOME installer. Instead of having a live image that can install another instance of the system, what boots into ...YouTube
reshared this
I wonder who might be willing to fund an overhaul of Java accessibility on Windows, particularly for frameworks that run on top of AWT like Swing and the desktop version of Jetpack Compose. The old Java Access Bridge for Windows was necessary back when there was nothing but Microsoft Active Accessibility (MSAA). But today, Java should implement UI Automation.
The decrepit state of Java accessibility on Windows has real consequences for users. mindly.social/@valiant8086/114…
It's too far back in my timeline, but I just installed blip, a free file transferring application on Windows. It works on other platforms too. The link I used was on a blog post where they discussed how it worked with Voiceover on the Mac. On windows? 100 percent inaccessible. I can't navigate anything at all. I pushed enter and it took me to a license agreement page. That's all I can get it to do. What a blah. I imagine it works fine on Android. How is there anything on Windows that is just 0 navigation at all. I can't move the navigator object in NVDA or anything. Absolutely nothing, like it's a video game. That's pretty rare to find something that bad.
Peter Vágner reshared this.
#XMPP enthusiasts out there: what would you say the ultimate Achilles heel of the XMPP ecosystem is, at present? Fragmentation of clients? What?
My sense is that it's this: when one goes to store an XMPP address in one's addressbook, there doesn't seem to be standard way to store an XMPP address. #Android doesn't have that as an allowable field, and #Thunderbird and #Nextcloud have an "Instant Messaging" field, where the type can be set to "XMPP". But are these two compatible with each other when trying to sync between them? Edit: Yes, but there's a catch: *the XMPP address must be prefixed with "xmpp:"*
So "user@foo.bar" is not an OK XMPP address, but "xmpp:user@foo.bar" is.
Then to make matters worse, now there's a wish to change the labeling of "XMPP Address" to "Chat ID": gultsch.social/@daniel/1140129…
It might be a long time before the address synchy-ness ever works again between Android <-> #Davx5 <-> Nextcloud <->Thunderbird
Note: Android allows a "Jabber" type for an IM address, where you *don't* prefix the address with "xmpp:".
(#DeltaChat gets to gloat hard here, as they have plain-old email addresses)
#prosody #conversations #gajim #dino #snikket #monocles #monal
Peter Vágner reshared this.
If you're still recommending #Signal, you may have missed the tech oligarchs' takeover of the US government. The best time to recommend European alternatives was 8 years ago; the second best is now.
FediVerseExplorer likes this.
reshared this
Dear @Thunderbird: Free Your Inbox your developers and the community is awesome. It's already in the review process!
Peter Vágner likes this.
Installs or updates a window subclass callback.learn.microsoft.com
reshared this
reshared this
I just migrated my personal #XMPP server based on #Prosody to its own #selfhosted server. I could not believe how easy it was, including transferring configuration, accounts, stored messages/files, etc.
BTW, the new server is just tiny, with 1GB of RAM, and it still seems to use only a fraction of it.
I suppose this is the result of dealing with an amazing protocol (XMPP) coupled with an excellent piece of technology (Prosody) 👏
reshared this
Did you know that you can easily (re-)setup DAVx5 via QR code by simply using this type of URL to automatically pre-fill the DAVx5 Login screen:
davx5://user@server.example.com/path/
You will need an appropriate QR code scan app like Binary Eye. Read more here: manual.davx5.com/integration.h…
Pro tip: save the QR code somewhere and whenever you want to re-setup simply scan it again 
. Also great to help other not so tech savvy users by simply giving them a QR code!
Peter Vágner reshared this.
Peter Vágner likes this.
Peter Vágner reshared this.
This one passed me by.
As of Ableton Move V1.3, released a few days ago, you can now sample from audio sent to the USB C port.
This is useful enough as it is stated, I.E. you can make a sound with a synth on your Mac or PC, then instantly sample it into Move without rendering and uploading a wave file, or connecting a line input and sampling it through the Move DAC.
However, for those of us who rely on accessibility, there is another hidden benefit.
Let's say you're on an airplane or something. You have an iPhone or Android with USB C, and you want to access Move Manager at move.local, so you can use the built-in web-based screen reader. Traditionally, this needs a Wi-Fi connection, but since you're on a plane, that doesn't work.
The Move's USB C port also acts as an Ethernet connection, so you can theoretically just plug your phone into USB C, then open your browser and go to move.local.
Yep, that works fine, except... oops, you're also a Voiceover/talkback/some other screen reader user, and audio gets routed away from the phone to absolutely nowhere... at least, that used to be the case.
Now, you can select USB C in the sample mode, and it will pipe audio from whatever is sent to that port to Move's built-in speakers or the headphone jack.
Thus, it's now possible for blind and visually impaired operators to use Move's screen reader when completely offline, with no way to access Wi-Fi without having to deal with a bunch of splitters, cables and mixers.
I bet they didn't even think of that use-case, but I'm happy to see it.
Since sample mode doesn't talk yet, the best way I've found to make this work is to connect your device to Move's USB C port, wait until speech goes away, press sample, then click the wheel until your device starts making noise through Move's output.
There are now three options that are toggled when clicking the wheel on the first item in sample mode.
The default is to record from Move's mic/line-in, second is resampling, and third is USB C.
The USB C option, far as I can tell, only appears when a device is connected to the USB C port.
reshared this
Microsoft can't make you switch to the new Outlook.Aman Kumar (XDA)
reshared this
Happy I Love Free Software Day! 💕
Unfortunately, this year I could not join nor organize any in-person celebration, BUT of course I want to share my gratitude to the many, countless #FreeSoftware services I don’t merely use, but actually depend on.
Last year, I decided to focus only on #YunoHost, because it would have been crazy to list all the projects I use and I love.
This time, even if I will most certainly forget someone, I am challenging myself to mention all the #LibreSoftware my life is powered by.
Without further ado, THANK YOU to:
Lastly, but most importantly, the biggest thank you goes to all the free software libraries and dependencies the above mentioned #software are made of/built with, including #C, #JavaScript, #Python, #Rust, and all community-maintained programming languages.
I am super sorry if I forgot someone!
#OpenSource #ILoveFS #SoftwareFreedom #Fairphone #Android #LineageOS #FSFE #OpenStreetMap #PhotoPrism #LibreOffice #Readeck #Eleventy #11ty #GNOME #Signal #forgejo #Codeberg #MusicBrainz #ListenBrainz #MetaBrainz
Information and insights concerning the configuration and maintenance of Tommi’s server.Tommi (Tommi’s mind)
Peter Vágner likes this.
reshared this
So is @mozillaofficial's #distilvit a #LLM that could be used to embed in a CMS to create alt text for uploaded images?
image-to-text model for PDF.js. Contribute to mozilla/distilvit development by creating an account on GitHub.GitHub
Peter Vágner reshared this.
hey, neat! The project has a codecarbon file for tracking resource usage in it!
The repo is written in python l so it would be not too complicated to reuse the code in say… wagtail.
If there is a php or WASM version of the hugging face transformers library then you could run it in say… drupal I think.
For at least the last 15 years, the translations of GNOME into Czech have been in excellent condition. With each release, I would only report that everything was translated, and for the last few years, this was also true the vast majority of the documentation. However, last year things started to falter. Contributors who had been carrying this for many years left, and there is no one to take over after them. Therefore, we have decided to admit it publicly: GNOME currently has no Czech translators, and unless someone new takes over, the translations will gradually decline.
Personally, I started working on GNOME translations in 2008 when I began translating my favorite groupware client – Evolution. At that time, the leadership of the translation team was taken over by Petr Kovář, who was later joined by Marek Černocký who maintained the translations for many years and did an enormous amount of work. Thanks to him, GNOME was almost 100% translated into Czech, including the documentation. However, both have completely withdrawn from the translations. For a while, they were replaced by Vojtěch Perník and Daniel Rusek, but the former has also left, and Dan has now come to the conclusion that he can no longer carry on the translations alone.
I suggested to Dan that instead of trying to appeal to those who the GNOME translations have relied on for nearly two decades—who have already contributed a lot and are probably facing some form of burnout or have simply moved on to something else after so many years—it would be better to reach out to the broader community to see if there is someone from a new generation who would be willing and energetic enough to take over the translations. Just as we did nearly two decades ago.
It may turn out that an essential part of this process will be that the GNOME translations into Czech will decline for some time.Because the same people have been doing the job for so many years, the community has gotten used to taking excellent translations for granted. But it is not. Someone has to do the work. As more and more English terms appear in the GNOME interface, perhaps dissatisfaction will motivate someone to do something about it. After all, that was the motivation for the previous generation to get involved.
If someone like that comes forward, Dan and I are willing to help them with training and gradually hand over the project. We may both continue to contribute in a limited capacity, but the project needs someone new, ideally not just one person, but several, because carrying it alone is a path to burnout. Interested parties can contact us in the mailing list of the Czech translation team at diskuze-l10n-cz@lists.openalt.org.
reshared this
reshared this
UTF-8 🤦♂️
I already knew about the "confusables", e.g.: e vs. е. Which look ~same but are different.
But you can also smuggle arbitrary byte streams in any character via "variation selectors". So this emoji: 😀󠅧󠅕󠄐󠅑󠅢󠅕󠄐󠅓󠅟󠅟󠅛󠅕󠅔 is 53 tokens. Yay
reshared this
reshared this
reshared this
Call with a universal credit agent a couple of days ago.
'So can you upload a CV?'
'Yes sure, I can do that.'
'How will you do that, can you use a computer?'
'Yep, no problem.'
'How do you do that, do you use bra-lay?'
'What?'
'Bra-lay.'
'Oh, you mean Braille?...'
'Yes, yes. Braille. That's it.'
'Well, I can read braille but I will use a screen-reader.'
'Oh, you use a screen-reader?'
Yep, that's how my call went. I've heard of some weirdness in my time but 'Bra-lay' for Braille, that, is a special kind of special right there.
reshared this
Peter Vágner reshared this.
)
🤖 Just a command runner. Contribute to casey/just development by creating an account on GitHub.GitHub
reshared this
I did a static analysis on the DeepSeek Android app
tl;dr it does aggressive device fingerprinting, root detection, has anti-tampering mechanisms, bundles native code and has dynamic code loading and execution facilities
none of which should be necessary for an app like this
more here: michael.bacarella.com/2025/02/…
reshared this
#Movuary day 3 brings you a classic example of why I should not be allowed to make music like this, whatever this is supposed to be. I don't know how to classify it, and I wrote it. Creapy broken jazzy dub and bass? I dunno...
HQ download:
borris.me/audio/movuary2025/03…
reshared this
reshared this
Peter Vágner likes this.
reshared this
Hi, The public share link has been removed with the 3.15 client version. Could it be possible to bring back this feature ? Below, a screen from a right click on a folder/file with the 3.14.3 client...GitHub
🇨🇦Samuel Proulx🇨🇦 likes this.
reshared this
📣 Shortwave 5.0 is now available, bringing background playback and completely revamped stream recording!
Peter Vágner likes this.
reshared this
Peter Vágner reshared this.
reshared this
My girlfriend bought me a new dress last night and I went a little off the rails trying to describe it in the alt-text 😉
reshared this
reshared this
We are excited to announce the National Federation of the Blind Radio Network (NFBRN). Broadcasting 24/7, this dynamic station brings together a diverse mix of content from the organized blind movement, including legendary speeches, Braille Monitor articles, recent podcasts, convention highlights, uplifting stories, and live events. And there is more to come. Send your feedback to communications@nfb.org.
NFBRN is available on any Amazon device by opening "Nation's Blind."
Peter Vágner reshared this.
NFBRN, the National Federation of the Blind Radio NetworkNFBRN, the National Federation of the Blind Radio Network - Internet Radio Directory - RCAST.NET
Be wary when adding additional context only for #screenReader users. An example:
Say you're working on an e-commerce site, and some products have two prices to show how great a sale discount is. The before and after is made visually apparent via some aspect of text formatting, and you want to make it explicit for screen reader users too.
The first step is to ask if this is necessary. If a user encounters two consecutive prices and one is lower than the other, they may intuitively understand what's going on without any explicit signposting, and can verify how much they're gonna pay during the checkout process. Only your users can provide this verdict.
If it's determined that some additional context is helpful, you could format it as something like: "Was $14.99, now $8.99" (optionally swapping the prices). It's short and punchy in braille and speech, perfectly descriptive of the situation at hand, and mirrors how it may be spoken out loud on an ad.
Resist the temptation to go further than this. You do not need to say "original price: $14.99, current sale price: $8.99". This is much longer and more verbose, while adding nothing. It also implies that you think screen reader users need to be told what a price is and explained the concept of a sale, even though you're not doing so for other audiences.
You also don't need to spell out the word "dollars", format the price in words, repeat the product name, and so on. If you find yourself with screen-reader-only text like: "The current price of 500 Grams of Premium Oolong Tea was fourteen dollars and ninety-nine cents, and is now on sale for eight dollars and ninety-nine cents", it has gone way too far.
In short: Set out to identify the problems that actually need solving, and only solve those problems.
FediVerseExplorer likes this.
reshared this
reshared this
reshared this
Quentin
in reply to Jamie Teh • • •Jamie Teh
in reply to Quentin • • •Jamie Teh
in reply to Jamie Teh • • •Quentin
in reply to Jamie Teh • • •Kara Goldfinch
in reply to Jamie Teh • • •Kaveinthran (inactive)
in reply to Jamie Teh • • •I always struggle to go to the options section of ditto, when I press the ctrl+grave key and right click, there are all sorts of menu, but not options. How do you go to options? cc @KaraLG84
Kara Goldfinch
in reply to Kaveinthran (inactive) • • •Jamie Teh
in reply to Kara Goldfinch • • •Kara Goldfinch
in reply to Jamie Teh • • •Jamie Teh
in reply to Kara Goldfinch • • •Kara Goldfinch
in reply to Jamie Teh • • •Kaveinthran (inactive)
in reply to Kara Goldfinch • • •Jamie Teh
in reply to Kaveinthran (inactive) • • •Jamie Teh
in reply to Jamie Teh • • •Kaveinthran (inactive)
in reply to Jamie Teh • • •Jamie Teh
in reply to Kaveinthran (inactive) • • •Kaveinthran (inactive)
in reply to Jamie Teh • • •Kaveinthran (inactive)
in reply to Jamie Teh • • •What other features that you usually use, I am just trying to discover hidden gems here, I should try the num without dashes thingy.
Kaveinthran (inactive)
in reply to Jamie Teh • • •