mastodon - Link to source

After years designing and scaling web and mobile applications using React and Node.js, I am now actively looking for a full-time, remote FullStack position.

My focus has always been on solving tough technical problems and building systems that last. I enjoy creating maintainable solutions and have a background in developing products that significantly boost efficiency and customer retention.

Any leads or introductions to teams building great products would be genuinely appreciated!

#FediHired

#fedihired

mastodon - Link to source
Ranting about how stupid cultural superiority is.

Sensitive content

mastodon - Link to source

TIL that Kenyan workers have been used so much to train AI systems, that standard writing by Kenyan people is often flagged as AI generated while it is not (which means that they can get discriminated for jobs / exams etc)
marcusolang.substack.com/p/im-…

Edit: many people raised below that the article is talking about texts written in very classically trained English detected as AI generated, which is the case for many Kenyans. It is documented that many Kenyan workers have been hired to train LLMs, but I made an assumption that it was the reason for this detection while it may not be. Sorry about that, thanks for the feedback (and feel free to continue the discussion here)


I'm Kenyan. I Don't Write Like ChatGPT. ChatGPT Writes Like Me.

I'm calm. I'm calm. I promise.
Marcus Olang' (this man's mind)
This entry was edited (1 week ago)

reshared this

mastodon - Link to source

Welcome to our last In-Process blog post for 2025! nvaccess.org/post/in-process-1…

In this edition:
- Holiday Season Trading Hours
- NVDA with Digitech Reece
- World Blindness Summit Presentation
- Finding Things

Do check it out, have a wonderful break if you are having time off or a holiday, spend time with loved ones, and we look forward to catching up with everyone in 2026!

#NVDA #NVDAsr #ScreenReader #Accessibility #Christmas #NewYear

Screenshot of YouTube cover image for NVDA with Digitech Reece video, showing Reece working at his computer, with "Watch on YouTube" and play buttons visible)

NV Access | In-Process 16th December 2025

www.nvaccess.org
#Accessibility #screenreader #nvdasr #nvda #christmas #newyear
in reply to NV Access

mastodon - Link to source

NV Access

Following on from that last In-Process, I (Quentin) am mostly off until the a week or so into January. I will pop in from time to time just to make sure everything is running smoothly, but if you need a reply from us specifically, unless it's urgent (eg the website is down), we likely won't get back to you until the new year. If you need help with NVDA, please do check the resources in the "Holiday season trading hours" article in the blog. Happy New Year everyone!
This entry was edited (1 week ago)

Zach Bennoui reshared this.

mastodon - Link to source

The death of the internet as we know it is real and very much happening in front of our eyes

"Google AI summaries are ruining the livelihoods of recipe writers: ‘It’s an extinction event’"
theguardian.com/technology/202…


Google AI summaries are ruining the livelihoods of recipe writers: ‘It’s an extinction event’

AI Mode is mangling recipes by merging instructions from multiple creators – and causing them huge dips in ad traffic
Aimee Levitt (The Guardian)

mastodon - Link to source
President Trump signs an executive order designating fentanyl as a weapon of mass destruction. whitehouse.gov/fact-sheets/202…

Fact Sheet: President Donald J. Trump Designates Fentanyl as a Weapon of Mass Destruction – The White House

DEFENDING AMERICA FROM A CHEMICAL WEAPON: Today, President Donald J. Trump signed an Executive Order designating illicit fentanyl and its core precursor
The White House
in reply to victor tsaran

mastodon - Link to source

Matt May

@vick21 Fentanyl is objectively terrifying. Pretty narrow band between a recreational dose and a fatal one. Cheap to make, hard to detect in transit, and now it's being fabricated all over.

Basically, any pill or powder being sold on the street could be cut with fentanyl, if not 100%, because it's cheaper and easier to procure than other drugs. That's how most people die from it.

@victor tsaran
in reply to Mahmoud

mastodon - Link to source

Thunderbird: Free Your Inbox

Thanks for trying the Android app, and we're sorry notifications aren't working! Please see the article at Settings > Getting Help > Settings > "Configure push...", then enable push and Notifications in Manage folders > [folder]. Also enable notifications under Settings > [account] > Notifications > "New mail.." and your phone's notification settings in Settings > Notifications > Messages. (Also, make sure Thunderbird doesn't have reduced battery usage!)

mastodon - Link to source
Eloquence for Android is available to me, with early access. If you want to find out if it's available to you, just search the play store for "Eloquence" and look for the one called "Eloquence text to speech." I didn't need to click a link or do anything special; I'm not on CodeFactory's radar at all.
The news is pretty much all good. There's a new rate boost option that will help a lot with reading apps that arbitrarily cap the maximum speech rate at 300 or 400 percent, which is basically every reading app I could find. This was a major pain point of the previous version of Eloquence, and it wasn't even CodeFactory's fault, so it's nice to see this added as an option.
Otherwise, it should be familiar to anyone who used the old version. I can't remember if there was a way to remove intonation pauses in the previous version but this new one has a checkbox for it. It has a sample rate selection, and like before, I actually enjoy the 16KHZ version. The higher sample rate on iOS causes Eloquence to sound like it has no teeth, but the Android version has always sounded more natural, and it adds much-needed clarity when using low-quality earbuds.
As for the subscription, it's $3/month for me. I'm not sure if that's Canadian or US dollars, so it might be less than $3 for Americans. There's also a $30/year option. Considering Eloquence was originally a one-time purchase of $20, this seems like a good balance. Android is the affordable option for a lot of people, and $3 USD goes a long way in other countries, so I imagine this was their attempt to get as many people subscribed as possible. Let's not forget that the price for subscriptions can change at any time, so this could be an introductory price or a forever price.
My main question is whether a failed license validation will cause the synth to stop functioning, leaving the user with no way to use their own phone. CodeFactory also resells Vocalizer voices, and those are lifetime licenses, but I've had two different phones become unusable after being turned on with no internet connection. In one case, I had to get sighted help to connect to the wi-fi where I was staying, so CodeFactory could connect to its licensing server to--idk, verify the definition for the word "lifetime" I guess. So I sincerely hope the fallback mechanism has been improved here, because rendering someone's phone unusable is a serious potential safety issue.
I know people will probably crack this, for good reason, but I would urge everyone to only install this hypothetical crack to circumvent the licensing restrictions and keep the subscription active. CodeFactory did create this port and if we all just steal it for free, they'll make no money from it. That's not a win for anyone, and it will cause them to abandon it the way they did the first one.

hometown - Link to source
Big news from the Video Game History Foundation - they’ve come into a huge collection of previously-inaccessible Sega Channel data. Largely unseen since it originally broadcast in 1994-1997. gamehistory.org/segachannel/

The Secrets of Sega Channel: VGHF recovers over 100 Sega Channel ROMs (and more) | Video Game History Foundation

Our project to preserve the history of Sega Channel — including over 100 new Sega Channel ROMs.
Phil Salvador (Video Game History Foundation)

mastodon - Link to source
Eloquence on Android first impressions: CodeFactory still doesn't seem to know how to make an Android TTS voice properly, because it does not interrupt how it should at fast rates, same as Vocalizer. With RHVoice, if I flick through an app quickly, it starts speaking the start of every string as I flick. Eloquence, though? If I flick twice super fast it'll just keep reading the very first string until it's done, and then move onto the final one. #YouHadOneJob
#youhadonejob

Dennis reshared this.

in reply to Jamie Teh

mastodon - Link to source

Andre Louis

@jcsteh @Scott @KaraLG84 I wish I'd recorded it, but once they sent me a build that was so responsive that turning one of the controls at high-speed read the beginning of everything you passed through. I said 'I want that one! give me that one!' They didn't. Now if you turn too quickly, you just miss things as I guess, not being a programmer, it has a poll rate or some shit, so if you're over it, it skips. I want a responsiveness slider. Give me a responsiveness slider.
@Jamie Teh @Scott @Kara Goldfinch
in reply to Andre Louis

mastodon - Link to source

Jamie Teh

@FreakyFwoof @Scott @KaraLG84 Rule number 1 in speech related accessibility: responsiveness is of utmost importance, almost over everything else, unless prioritising it would cause the information to be inaccurate. I don't know why so many people struggle to understand this, especially since the same people would probably be horrified if their buttery smooth scrolling had even the slightest hiccup or dropped a single frame. Apple could take a lesson here too.
@Scott @Andre Louis @Kara Goldfinch
in reply to Andre Louis

mastodon - Link to source

Jamie Teh

@FreakyFwoof @Scott @KaraLG84 I used to think this delay was some communication lag between KK and the host, but it isn't, because transport buttons, etc. respond very quickly. With OSARA, the DAW buttons end up reading faster than KK's own controls do. That means it's very likely an artificial delay, which is utterly infuriating. It annoyed me enough that I spent a little bit of time trying to see if there was some way I could reverse engineer their protocol and write my own helper, but I didn't get anywhere.
@Scott @Andre Louis @Kara Goldfinch
in reply to Chi Kim

mastodon - Link to source

Jamie Teh

@chikim @Scott @FreakyFwoof @KaraLG84 Makes sense. That's honestly what I thought, but I got the impression somewhere from NI that it used auto mapping features in some DAWs or something and I assumed those mappings were somehow more sensible. Apparently, I misled myself. On the flip side, this should be fairly straightforward to implement in REAPER. Famous last words.
@Chi Kim @Scott @Andre Louis @Kara Goldfinch
in reply to Jamie Teh

mastodon - Link to source

Scott

@jcsteh @chikim @FreakyFwoof @KaraLG84 Hmm I thought they were doing some nice sorting or something as well. So two questions:
1. Is using the controller to twiddle DAW params still subject to the same delayed speech response?
2. If there's nothing complicated happening, does that mean ReaKontrol could do this with hardware other than MK3 Jamie?
@Jamie Teh @Chi Kim @Andre Louis @Kara Goldfinch
in reply to Scott

mastodon - Link to source

Jamie Teh

@Scott Regarding 2, I wondered that too. The issue is that ReaKontrol doesn't get feedback from the knobs and page buttons unless you're in the mixer view, at which point the user probably wants to use them for the mixer. We could probably have a "fake" mixer view which gets enabled when you press the 4d encoder while in the mixer view or something like that, but it's definitely weird. Plus ReaKontrol doesn't get anything when you touch the knobs, only when you turn them, so you couldn't query the value. @chikim @FreakyFwoof @TheQuinbox @KaraLG84
@Quin @Chi Kim @Scott @Andre Louis @Kara Goldfinch
in reply to Jamie Teh

mastodon - Link to source

Jamie Teh

@Scott Personally, I don't buy NI's argument that this stuff is only possible with Mk3. Mk3 does more of it on the keyboard hardware itself, sure, but that's no reason they couldn't have done it with Mk2. They'd have to move the display and accessibility rendering out of the KK plugin and into a separate service, but it should still be possible. I think they just chose not to. Of course, I'm not an NI dev, so maybe there's something I'm missing. @chikim @FreakyFwoof @TheQuinbox @KaraLG84
@Quin @Chi Kim @Scott @Andre Louis @Kara Goldfinch
in reply to Jamie Teh

mastodon - Link to source

Zach Bennoui

@jcsteh @Scott @chikim @FreakyFwoof @KaraLG84 Yeah maybe it's just a matter of resources. I get the impression they're trying to make their actual software more accessible without the keyboard as well, latest Kontakt update is a good example of this. I can now at least browse presets comfortably without using KK, but I hope they'll make editing accessible from the GUI as well.
@Jamie Teh @Chi Kim @Scott @Andre Louis @Kara Goldfinch
in reply to Scott

mastodon - Link to source

Jamie Teh

@Scott If there aren't, that seems like something that could probably be done in ReaKontrol - it would just pretend that only the parameters you chose exist in the order you chose them - but the question is how to set up those templates. I actually wonder whether it might be worth exposing a fake plugin which gives you all the parameters you added to the TCP. That way, you could use track templates to get you just the parameters you regularly use. @chikim @FreakyFwoof @TheQuinbox @KaraLG84
@Quin @Chi Kim @Scott @Andre Louis @Kara Goldfinch
in reply to Scott

mastodon - Link to source

Jamie Teh

@Scott Oh haha, I didn't realise they didn't have a UI for it. Doing it with config files wouldn't be too hard, but how non-OSARA users would get the parameter numbers is an interesting question. Honestly, it probably wouldn't be that hard to build a UI for it either. I'd just rather do... just about anything else. @chikim @FreakyFwoof @TheQuinbox @KaraLG84
@Quin @Chi Kim @Scott @Andre Louis @Kara Goldfinch
in reply to Scott

mastodon - Link to source

Jamie Teh

@Scott You could do that I guess, though it makes the config a bit harder to deal with.
VST: ReaEQ (Cockos) = Freq-Low Shelf | Gain-Low Shelf | BW-Low Shelf
Maybe it needs to be YAML or something like that so we can do items on separate lines:
VST: ReaEQ (Cockos):
- Freq-Low Shelf
- Gain-Low Shelf
- BW-Low Shelf
But anyway, this is a way off, so no idea why I'm musing on it.
@chikim @FreakyFwoof @TheQuinbox @KaraLG84
@Quin @Chi Kim @Scott @Andre Louis @Kara Goldfinch
in reply to Andre Louis

mastodon - Link to source

Jamie Teh

@FreakyFwoof @Scott @KaraLG84 I just sent an email to NI. Excerpts below:
Even with Mk2, I've always noticed that the responsiveness of the speech with the keyboard is rather sluggish. I always assumed this was a limitation specific to the hardware. However, with the NI Accessibility Helper, I'm noticing that the latency is high even when using the PC keyboard to navigate the zones, get help, etc.
Subjective measurement is tricky for developers, so let's get scientific. I used a SAPI5 voice implementation which allows me to get speech directly into the NVDA screen reader. NVDA can measure the time from a key press to when speech occurs. Here's what we get:
[NVDA log]
As you can see, there's a delay of over 300ms between the time a key is pressed and the time speech occurs. To ensure this isn't a delay elsewhere in the system, I tried outputting directly from a Python program to SAPI5:
[NVDA log]
Here, you can see that the time is only 45ms. That's still not ideal, but that's a great deal lower than the >300ms we see with NI Accessibility Helper.
As a totally blind screen reader user myself and someone who spent 10 years developing a screen reader full time, I can't emphasise enough the critical importance of responsiveness. It can make or break a delightful experience. Many users won't even understand exactly what is bothering them, but it will irritate them over time. To make this practical, if someone hears 10000 messages from NI Accessibility Helper (not unreasonable if they're browsing a lot of instruments and making a lot of parameter changes in a long session), at 300ms of latency per message, they've wasted 50 minutes of their time just waiting for speech that they could have spent making music.
Perhaps the concern is that the user will hear a lot of truncated speech if they scroll fast and so an artificial delay has been introduced. On the contrary, hearing truncated speech is far preferable to high latency and screen reader users are already very familiar with this when they move fast. If this is not a concern you are willing to dismiss or there are certain voices that have caused problems, perhaps a responsiveness setting could be introduced.
If this isn't the concern, it'd be great if this responsiveness problem could be investigated.
@Scott @Andre Louis @Kara Goldfinch
in reply to Brian Tatosky

mastodon - Link to source

Andre Louis

Three of the four I'll agree with. Zoom AI though, a silent note-taker, yeah. I'm here for that. I teach on zoom and at the end of an hour when I've covered more things than I can remember, having a recap for the student at the end of it is wonderful.
It may not be 100% accurate but it's better than me trying to take notes during, and remember what happened after.
The rest can go right away though, for sure.
I know my students have found it useful too.

mastodon - Link to source

As I was already complaining about #WUR I can add another nice episode from their IT:
As some of you might know, @thunderbird added native support for exchange. I immediately wanted to connect my uni-email and get rid of my wonky outlook-brave-webapp.
Of course, it did no go as expected: an error message told me that admin approval is needed.
So far so good, let's ask IT if they could change this, as thunderbird is the e-mail-clients with the oldest and best track record you could possibly have.

Reply from them: "We cannot provide access because we don't know what is done with the data gathered from your Office365 Account. You will need to use Microsoft Outlook."

#wur @Thunderbird: Free Your Inbox
in reply to Puntarella

mastodon - Link to source

Thunderbird: Free Your Inbox

Sorry for the delay! Thunderbird being a desktop client means the communication between the user's terminal and the organisation's mail server is direct without using a third-party server or a web application backend. It's probably best to point your IT department, for now, to our privacy policy to show what we do (and don't do!) with your data: thunderbird.net/en-US/privacy/

Thunderbird Privacy Policy

Thunderbird is a free email application that’s easy to set up and customize - and it’s loaded with great features!
Thunderbird
in reply to Mikalai

pleroma - Link to source

feld

@mikalai @collectifission it's a bug that affects me every single day. They have nothing more pressing than fixing what they've already made and actually building a fully integrated login manager so they can behave like a real OS.

All this rewriting everything every 2 years is nonsense. Maybe a lot has changed since I last used KDE 3 but I'm not seeing the fruits of their labor

@Emil Jacobs - Collectifission @Mikalai

mastodon - Link to source

My Ubuntu Summit talk is up! Where I talk about:
1. How Desktop UX is effectively dead
2. Why I hate the term UX/UI with the heat of 1000 suns
3. How OSS can actually innovate in #ux

youtube.com/watch?v=1fZTOjd_bO…


Are we stuck with the same Desktop UX forever? | Ubuntu Summit 25.10

This talk focuses on that evil little term “UX/UI,” which is responsible for so much confusion and tension in open-source projects. Not only does it unnecess...
YouTube
#UX

mastodon - Link to source

i was informed yesterday that Atlassian has inserted an F1 racing game into Confluence.

I was asked if I (as an admin) could disable it because folks were finding it's pop ups annoying. Here is what I found:

"To disable the Car Racing Easter Egg feature, contact Atlassian Support. There is currently no in-product toggle for end-users to turn off this feature themselves. Support can assist with disabling the feature for your site or organization if required."

mastodon - Link to source

I'm now running into what I think of as "ghost" open source projects due to genAI.

You'll search for a project that does X, find a public repo for a project that says it does X, it has a beautifully formatted README and well-organized code, but when you install it, it doesn't work.

Then you find the note: "Written with the help of AI, never tested, please send me spare hardware so I can test it."

...
.....

Dead internet.

reshared this

glitchsoc - Link to source
I just discovered a quite nice open source #android app for #PublicTansport connection look up. It's called #Bimba and uses #Transitous api.
f-droid.org/packages/xyz.apiot…
App screen for entering origin and destination points.
App screen with found connections amd map.

Bimba | F-Droid - Free and Open Source Android App Repository

FLOSS public transport passenger companion; a timetable in your pocket.
f-droid.org
#android #transitous #publictansport #bimba

Jiří Eischmann reshared this.

in reply to Ondřej Eliáš

glitchsoc - Link to source

Štěpán Škorpil

Ještě nevím, zatím ji testuju.
UI pro zadání query je rozhodně měně přehledné.
Le líbí se mi použití Trasitous api, tzn mělo by to jít použít i v zahraničí.
Na druhou stranu Trasitious ani v Čechách nemá kompletní data od všech dopravců, tzn v českém pokrytí má Idos navrch - idos patří Chapsu který měl státem určený monopol na data o jízdních řádech. Rozbil to dlouhými soudy až Seznam, který data chtěl taky.

mastodon - Link to source

Mixxx Stable 2.5.4 Released!

We're proud to share our early Christmas present with you: Mixxx version 2.5.4, a new stable release. This version bundles various minor corrections, improvements and fixes for the issues you reported. We recommend that all users upgrade to this version.

Check out the full announcement:
mixxx.org/news/2025-12-14-mixx…


Mixxx - Mixxx 2.5.4 Released

Download the most advanced FREE DJ software available, featuring iTunes integration, MIDI controller support, internet broadcasting, and integrated music library.
Mixxx DJ Software

pleroma - Link to source

KDE Dolphin has a thing called kio-fuse which will mount a fuse filesystem to provide paths to access files from other software that isn't KDE-ified.

This works fine as long as your fuse isn't broken. But also there are apps that get marked as supporting SMB natively... like VLC...

so what I've encountered is that VLC playing videos out of SMB shares you browse in Dolphin is broken on FreeBSD; Linux distro users complain about this sometimes too because they need additional packages

VLC can support SMB, but on FreeBSD it's not built with that by default. (ffmpeg needs SMB support enabled too)

Firefox works though!

This entry was edited (1 week ago)

mastodon - Link to source

Just published an article about how an expert learns - from sight-reading to performing music... on cello using eye tracking technology. it's open access, so freely available: journals.sagepub.com/doi/10.11…

Even if you don't read the fancy bits, the progression of the heat maps to show the engagement of the eye and mind are very cool ! #research #music #cello

#music #research #cello

mastodon - Link to source

If, like me, you absolutely hate the recently introduced sidebar-like overlay "side-panel" that opens up when you click tickets in #GitLab, I finally found the obscure place where you can turn it off.

It's not in your user preferences or kebab menu, it's… tucked in that little "Display options" icon-only button between the tickets searchbar and the sorting order combobox:

Screenshot of GNOME's GitLab instance showing GNOME Calendar's issues tickets list, with the "Display options" menu open after clicking its icon-only button. At the bottom of that menu, there is an "Open items in side panel" toggle item that can be switched off. You're welcome.
#gitlab

mastodon - Link to source

This weekend, I tried out TalkBack 16.2. And rather than rant about the one step forward and a good 3 steps back it took in Braille support, or how it can't take an actual screenshot and describe the screen unless there's one accessible element onscreen it can grab on to, or any of the other stuff I routinely rant about, today, I'll just say thank goodness for the iPhone.

I'm so glad I didn't sell or give away my iPhone 16 or Apple Watch. I'm so glad I got the AirPods Pro 3. I'm so glad it's evident that at least a few blind people work at Apple and are listened to at least when it comes to their most popular product, the iPhone. I'm so, freaking, glad that Apple at least keeps the lights on for Braille on their platforms. And I'm so glad screen recognition is a thing so when I'm playing a video game, I can have the menu read out even though there is no accessible element on the screen. I'm so glad that with AirPods, Voiceover isn't too sluggish and that you can't really tell that, when you swipe passed the visible end of a list, like the settings app, VoiceOver scrolls the screen for you. I'm so glad that at least that bit of magic is still there. You don't *have* to know these things. In Braille, it *feels* like a Braille-first interface, as far as we've made one in the past 30 years. And when VoiceOver gets AI image descriptions, I'll be thankful for those too. In the meantime, hey Siri, describe my screen. I'm so glad that I can almost always keep my iPhone in my pocket and just use Braille. I'm glad I can type my passcode in Braille and it just works! I'm so glad I can read a book and it pans to the next page. I'm so glad I can bind just about any VoiceOver command to arbitrary Braille keys. If I want dots 1-3 with Enter to go to next link, I can make it do so! If I want Space with Enter with O to open the Braille Access app chooser, there it is! Oh also I can press Command, Option, Control, FN, and Shift modifier keys with Braille commands. Just sayin'. At this point, iOS with Braille would be a way more blind-friendly OS than Android. And I think we should make this more clear: Braille-friendly is blind-friendly. Because if your interface is flexible enough to work with Braille, it'll be flexible enough to work with speech.

#android #iOS #VoiceOver #apple #google #tech #technology

#android #apple #google #technology #ios #tech #voiceover
in reply to Devin Prater :blind:

pleroma - Link to source

Jason J.G. White

An alternative for braille display support under Android is the Android-specific version of BRLTTY, reputedly available from the Google Play Store. As i don't have an Android device suitable for testing, I haven't investigated its capabilities further. I've used it under Linux, of course, but that's a very different environment from Android.

In general, TalkBack seems to be very much a corporate project. The source code used to be available (and perhaps still is), but as far as I know, there's no community of developers outside Google (except for Samsung's version - also a corporate project).

mastodon - Link to source

There's a new Amazon bot launching today: `Amzn-SearchBot`

"...used to improve search experiences in Amazon products and services...such as Alexa and Rufus. Amzn-SearchBot does not crawl content for generative AI model training."

Doc page shows user-agent & links to source IP list.
Looks like it *does* obey `robots.txt` at least.

developer.amazon.com/amazonbot

#Bot #WebDev #Amazon


About AmazonBot

Customer facing page of Amazonbot crawler which all web content publishers can refer to.
Developer Portal Master
#webdev #Bot #amazon
⇧