It is true, though. Dynamic analysis is nowadays the simplest way to find the actual implementation, and after a while you learn where things probably are. So many function pointers.
about a billion years so I needed a SHA256 function in Lua. Existing bindings to OpenSSL didn’t cover it, so I set to add it. I was beaten back by waves of confusion (which apparently got worse, somehow??). In the end I bound to WolfSSL. Not as fast but actually designed to be embedded. It was like ten lines of C.
Ein Artikel über die Debatte zum Bürgergeld. Ich mag den Beitrag. Vor allem, weil er von individuellen Schicksalen erzählt. Keine "Totalverweigerer", die vermeintlich lieber vom Staat leben. Sondern Menschen, die auf die ein oder andere Art nicht damit gerechnet haben jemals in diese Situation zu kommen.
Neue Regeln mit neuer Grundsicherung: Heute debattiert erstmals der Bundestag über die geplante Reform des Bürgergeldes. Drei Menschen, die davon leben müssen, berichten von ihren Erfahrungen und Sorgen.
📣 Unterstütze den Zugang zum !@metalab für Menschen mit Sehbehinderung oder Blindheit! Komm zum Do-It-Blind (DIB) Treffen 🛠️ am Montag, 19. Jänner, Treffpunkt metalab um 16:00; Ende 20:00 metalab.at/calendar/event/8381… #a11y #make #blind #inklusion
Note to authors about using "AI" art to promote your work
Sensitive content
I really hate to say this bluntly but I'm going to say it. I will not boost your stuff if you're using "AI" art in the cover or in the promotional material.
There's a lot of other people who won't either.
I would rather see the cringiest possible clip art than see hot "AI" garbage. I'd rather see the same five public domain photos off of Pixabay and Unsplash again and again than see "AI" garbage.
We don't want LLMs fucking with the writers' side of things. Don't use generated images to fuck with the artists' side of things.
Finally made the switch to @Tutanota Mail. Not 100% done, have a lot of accounts under [them who shall not be named] Mail, but I believe this is a good first step.
All this CSAM stuff on X with Grok has been making the rounds, and it's properly and appropriately sickening. But you know what we haven't heard from? We haven't heard from MasterCard or VISA blocking payments to Xhitter. We haven't heard from Collective Shout...
Say, yeah! Hey Collective Shout! Weren't you all about protecting the children? Weren't you all against abusive sexual material? Weren't you trying to be soooooo feminist with your targeting of LGBTQIA+ content? Weren't you intent on fixing these things?
Where's your campaign against Xhitter, where's your letter writing against the payment providers still supporting Musk in the middle of all this? After all, clearly Grok was trained on images of young girls. Why aren't you speaking up? Why aren't you shouting?
Maybe Collective Shout are afraid of the backlash they'd get from MuskE.
Maybe they're just having too much fun using the product.
Right now it relies on eSpeak only for one thing: turning text into phonemes. The actual sound generation, timing, and prosody all live elsewhere. The big question ahead: how hard would it be to decouple from eSpeak and build our own text→phoneme layer? The answer is: non-trivial, but very doable — especially for languages like Hungarian, where spelling is regular and stress rules are simple. The hard part isn’t DSP. The engine already works. The real work is linguistic: normalization, phoneme rules, edge cases, and a lot of listening and iteration. It’s the kind of challenge that rewards patience more than clever tricks — and honestly, that’s part of the appeal.
Confession: I've been a web browser engineer for over 8 years and shadow DOM still really hurts my head. In its simplest form, it's... simple enough. But nested shadow roots, descendants of shadow-including ancestors... 🤯
I recently had a need for a recursive C++ lambda that captures variables. This isn't possible in the intuitive way; you get: error: variable 'someVar' declared with deduced type 'auto' cannot appear in its own initializer. It turns out you can achieve this (albeit a bit horribly) by passing the lambda to itself! See this article for details: artificial-mind.net/blog/2020/…
There are algorithms that have natural recursive representation and are very cumbersome or painful to write iteratively. With the advent of lambda expression...
Lol, I think this post encapsolates my distaste for C++ in a single go. Still fond of it, since it was my first programming language, but also, just ick :). Good luck, sir.
Oi. I feel so bad for creating more projects for myself. But eh not really. It gives me a lot of learning, and actually if anything I can look into how NV SpeechPlayer does synthesis forming itself, something I can't do so well with other synths. Even if it wasn't completed and early stages, and later they got into Espeak, doesn't mean it's a throwaway project.
@kaveinthran Yeah, part of me is curious what improvements I could make to the synthesis itself, but getting ahead of myself a bit. Just getting it working / built for more modern NVDA is a good first goal LOL. The key thing you have there is the SpeechWaveGenerator which to make from scratch would take me something like weeks or months. So already I think just having it for re-use is huge. @NVAccess @jcsteh
Was bei der Debatte über Besteuerung von Unternehmenserben oft zu kurz kommt: Öffentliche Infrastruktur ist das Fundament, auf dem erfolgreiche Unternehmen überhaupt erst entstehen konnten. Straßen, Sicherheit, Bildung, Rechtsstaat - ohne das alles gäbe es diesen Reichtum nicht. Oft wird das in politischen Debatten nicht explizit benannt. Vielleicht, weil es zu offensichtlich ist für einige.
Trotzdem sollten wir es immer wieder klar sagen. Gerade marktradikale Akteure setzen gerne auf die Erzählung vom „sinnlosen Staat“ vs. „effiziente Unternehmen“, um gegen Steuern zu wettern. Ein Extrembeispiel für solche Narrative sind libertäre TechBros aus dem Silicon Valley. Dabei wäre die ganze IT-Branche undenkbar ohne Vorarbeit aus der öffentlich finanzierten Forschung & massive staatliche Förderung. Dieses wichtige „Detail“ wird gerne weggelassen, weil es nicht zur Ideologie passt.
Forked and maintaining it here: github.com/tgeczy/NVSpeechPlay… Thanks @NVAccess for splitting this repo off before the merge into Espeak, at the very least this will be kept up to date to work in NVDA 2025.X and later 2026. Along with other wrappers I maintain I'm adding this one to my projects. Let's see what comes of it tonight, at the very least reviving it to some extent will be a success I think. Since I now have the source it may eventually be possible to modernize this DLL as a 64-bit module as well, as a later goal.
I did try an early build of SpeechPlayer in Espeak, and in 2021 it already sounded the same as it does today. So what we know as "edward" and the various "klatt" voices are using the ideas from NV Speech Player, but what happened was, rather than maintaining it as another project, NVAccess decided to merge the ideas for how SpeechPlayer worked back into Espeak. This avoided them from needing to maintain another project as well, and merge back research or ideas into Espeak itself. A noble goal, but in the process they destroyed a bit of the experimentation that made NV Speech Player so unique.
I can't sleep. I'm really missing NV Speech player and it's keeping me awake, that aching feeling in my heart. When you could change all those parameters in the voice dialog related to NV Speech player. Before its ideas were merged into Espeak code you know. I just can't live without that beautiful slider interface, making the voice sound all weird. (And thanks to Jake's site for still keeping this one archived, truly)
@threlm4280 ha. Not sure. that driver hasn't been updated since 2014, it would take some refactoring to even get it to python3 state and if even indexing or modern NVDA say-all works is another question entirely. The good news is that a lot of the source is up on Jake's site so it would be easy enough to try to revive.
@Bri @threlm4280 Initially I was yes. For example: github.com/vortex1024/SpeechPl… is just a cached copy of Espeak from 2021 but identically sounds the same to current Espeak so once I tried that driver and compared I became quite sad. I then dug around and found the 2014-era code from Jake's site which I think is more of what I'd be looking for, the actual NV Speech Player DLL before it got integrated.
@threlm4280 OK, your amount of work is a lot less than you'd think. Last update was 2021. It's py3 compatible, you'd just need to implement whatever NVAccess's latest add-on changes are and build. github.com/nvaccess/NVSpeechPl…
You're not the only one. I'd like to see that be compatible again. It's not at all like any of ESpeak's variants. There's nothing like it no matter what you try with them! ☹️
Demonstrators in Montreal discuss their recent experiences with police aggression. As systemic barriers prevent victims from seeking justice, some victims of police violence have taken matters into their own hands.
Fender apparently bought Presonus in 2021, which I didn't hear about until just now, and it looks like they are perhaps killing the Presonus brand, starting with their DAW, recently rebranded as Fender Studio Pro.
I am way out of the loop when it comes to this gear stuff, I guess. That's what happens when you get locked into an environment where you can't explore such things for several years. So it goes.
I don't really know what to think about that. Fender hasn't been the most kind caretaker of brands over the years, that's for sure.
My Fender... uh... Presonus Studio Live 16.4.2 digital console is stuck in storage possibly forever, and I'm honestly not sure if I would want to use it these days anyway on account of it's audio interface being firewire 400 instead of something more modern. That made a lot of sense back then, but not anymore.
There was a bug recently introduced in the latest Accessiweather nightly build which improved app startup performance. The explain weather button wasn't visible even though everything was configured. Bug now fixed on the dev branch in tonight's nightly build in around 30 minutes or so when it runs.
i'm probably going to be told tomorrow that i'm being laid off, so if anyone has a lead on a decent remote embedded software engineering or unix/linux systems job, i've got a resume to send you.
i don't like discussing the specifics of who my past clients have been in public. however, my most recent position was contracting with the kernel team of a major CDN and cloud services provider that i know many people on fedi are also current or former employees of, maintaining their internal linux distribution (an customized Ubuntu downstream) for their custom server platforms, porting forward and cleaning up patchsets to newer versions of the distribution, fixing issues in toolchains and libraries, etc.
i've also done embedded software work for a large variety of devices including medical, cellular technology, and industrial control equipment, in addition to some consumer products. i've done board bringup and BSP development with Yocto/Openembedded, Buildroot, FreeRTOS, etc, on most major SoC and microcontroller families. when doing business logic, I am most comfortable in C and Lua, but fine working with C++ codebases, and can learn to work with other languages if a project needs it.
Finally home after having spent the last 36 hours in the music building. I have these little taco sticks from 7 11. I'm going to eat those and then find a pillow.
Hi @Bri I'm just trying out FastSM and I really like it. I was wondering if it's possible to also include mentions in the notifications buffer by chance? This is one thing I've come to rely on in Pinafore/TWBlue as it makes checking notifications super easy. You could still keep the separate mentions buffer but I just thought it could be easier to see all notifications/events in one place on the notifications buffer if mentions (replies) are there too. Thanks for putting this together, it is a really cool Mastodon client.
@Kaliah There's just one problem, I can't reply to a mention from there as it says the post is not found. I seem to be able to reply from the mentions buffer though.
@Kaliah Yep. Right now, keymaps support is basically just, load these keys from a file, no editing unless you do it manually, etc. Whenever I get around to the keymap, I'll fix this.
@Kaliah Here's another thing I found. If I use the mentions in notifications buffer option, replies such as the one I'm replying to will get cut off. For instance, like this: Bri🥰 (@Bri) mentioned you: @Kaliah @BTyson thanks for finding these bugs. I've been doing a metric shit ton with this app so it... 01:11:39 AM
@Kaliah So if I run this from a cmd, it likes to put the errors.log in the same folder that I'm at with cmd and doesn't seem to update the one in the main folder.
@Kaliah alright y'all, last build for now, which implements yet again, everything we just talked about. CMD fix, keymaps support with a win11 keymap option, replies in notifications buffer no longer being cut off, readme fix, reply to mentions in notifications buffer fix, no more .version in soundpacks list. github.com/masonasons/FastSM/r…
Ah no worries. I was just thinking like a TC ish thing where a keystroke put up help. Also again and I'm sorry for being so naggy, but maybe make it so where sounds aren't cutting each other off? Or make a toggle for that? I'd prefer they didn't do that.
I figured it out, you'd just moved the keymap default lol. Broken behavior though, it doesn't seem to be saving my chosen soundpack for my second account.
If mentions are separate, associate the sound to mentions buffer. If mentions sending to notifications, associate the sound to the notifications buffer
@Kaliah Yep. If you want to, you can even take the FastSM folder out of appdata, put it in the FastSM folder itself, rename it to userdata, and run the app entirely portably
@Kaliah @NicksWorld I will say I've taken great liberty in creating a context menu that shows the most rellivant options for what you're focused on. For example, if you're focused on a post, context menu will show options related to the post, and, in the case of posts with for example a poll, you'll see options related to the poll. Etc etc.
@Kaliah @NicksWorld Here's the deal! Nope, I don't have any team, it's just me working on this here thingy. But also, none of ya are bothering me at all. :)
@NicksWorld @Kaliah I think that might have something to do with the fact that we might not be performing quote operations correctly, despite doing it exactly as the mastodon API claims to want us to do it. Sooooooooo, I dunno
@Kaliah OK, I'm actually done now. - New earcon sounds (image, media, notification, poll, pinned) - Fixed invisible mode to play all the new earcons - Removed soundpack copying from application.py - Added pinned field to models - Fixed streaming to respect mentions_in_notifications setting
@Kaliah Hey, I appreciate you developing this application. I am enjoying it so far. It certainly is different, but that's just because I'm not used to it yet for example, I don't know how to use the invisible interface.
@Kaliah It does? I quoted a post not long ago and then red it in the sent buffer but it didn't read what I quoted. I was mostly doing a test to see if it would work because I haven't seen quoted posts on my home TL.
@Kaliah OH sweet! Thanks friend. Sorry if I'm asking so many questions this late at night. Here's hoping you're not the only one doing this and you have a team, because it must be a pain having to go through so many questions.
@Bri It really is. Did DJ got to know us, and did not treat us any differently just because we were blind. We are in contact with him, and if he finds a new place and we can get there, hopefully we can make it happen. But, even he doesn’t have a karaoke home at the moment.
FoxVK
in reply to SuspiciousDuck • • •SuspiciousDuck
in reply to FoxVK • • •@FoxVK
..looking for a template hold on mate and meanwhile can u look at my ****code hh