Canceled another subscription recently, the money that's left will probably go into donating to NTFY or other projects, another awesome software I seriously enjoy using. Support your FOSS developers, even if it's just a small amount. Especially if you'd pay for the tool, like I did with pushover before.
Yes. The fact that the variations between clients and the browser interface are *tiny*, and are limited to platform specific areas, is one of Tuta's HUGE strengths.
Beim dänischen Digitalministerium sollen alle Angestellten ohne Microsoft auskommen. Stattdessen werde man Linux und LibreOffice nutzen, sagt die Ministerin.
jestli myslíš Mastodon Pokud si potřebuješ obnovit heslo na mastodon.arch-linux.cz: 1. Otevři přihlašovací stránku instance. 2. Klikni na „Zapomněli jste heslo?“. 3. Zadej e-mail, který máš u účtu.
Pokud ti e-mail nepřijde: • zkontroluj spam / hromadnou poštu • případně se mi ozvi a podívám se na to
Sorry guys. I had to remove the timeline gaps system. It was just not working correctly at all, and was only causing issues and annoyances. Not sure if I will ever reimplement something like this in the future, however if it's of concern, you can disable caching entirely, at which point the app would load a fresh copy of the timeline upon every client start like before.
@jonathan859 couple of situations, now that we have timeline caches. If you close down the app and reopen it after a while, a timeline gap can occur. This basically means there is a gap in between the newest post in your timeline that was saved, vs the oldest new post retrieved.
Thank you for using FastSM! We update the app hourly to keep your experience running smoothely! For best results, keep your app up to date! FastSM will never ask you to review the app. Mainly because it can't, because there is no review system attached to GitHub. Anyway, This update includes but is not limited to bugfixes, performance improvements, new bugs, unintentional performance degradations, issues, disappointments, problems, issues, and many more issues. Please keep your updates turned on to experience all of this, and more! Our developers will never tell you that they are taking a nap, or mowing the lawn, unless, of course, they are indeed taking a nap, or mowing the lawn. Why our developers would be mowing the lawn in a changelog, who actually knows. Anyway, more bugfixes and performance improvements time! Please keep your updates turned on for more bugfixes and performance improvements! Thank you for using FastSM! We promise that as we continue to add more bugfixes and performance improvements, eventually we will add so many bugfixes and performance improvements that we will be forced to rename the app to SlowSM, or potentially even BuggySM. So stay tuned for this eventuality. Anyway, again we hope you enjoy using FastSM!
Hi, I'm probably also going to switch. However, I wonder if there's a way to filter boosts and replies from the home timeline as with Tweesecake? I see a filter boost option in the context menu, but I'm not sure if that is related to the timeline or to the user whose post I am focused on. Also, is there an option to hide the username so that just the nickname of the poster gets shown (vice versa is fine as well)? Great work so far, keep it up!
@garo Yes to all of this (Checkout timeline filters). You can filter many things out of whatever timeline you want, not just home. Also templates for removing things you don't want
@darren_duff Oh shoot. Hmm. When you manually updated, what did you do, replace your old files with the new ones, or completely delete the old copy and start fresh with a new one?
@darren_duff I only had that loop once, when I didn't wait long enough for the update to take place. Next time I updated, it worked normally. Maybe I'm just lucky.
@frog67 @darren_duff Oooooh, you know what, mine does the same thing. The window doesn't actually come up, it just focuses. It's hidden though. If you just alt tab away, it's hidden and doesn't show up in your alt tab list, it just focuses whenever you open it
I don't think I like this new world, you guys. Every time Youtube Music plays me a song I haven't heard before, I wonder if it's AI. Hmm, that's a lot of releases in a short time. Is the voice quality a bit odd because the track quality is low, or because I'm imagining it, or because it's AI? Wait, those rhymes are really weak. Did an LLM just spit those out?
There are two primary reasons I like watching the output from Claude or Gemini, or any model for that matter: #1. you can learn some really cool shell tricks, #2. I love watching how it transforms my thoughts, e.g. from a markdown spec file, into a plan. Even the most careful planning on my part always has some gaps, high verbosity, etc. For that reason, I always request the model to ask followup questions.
In today's world of autocorrect, why do so many places where people write from their phones always include contractions with missing apostrophes? You'd think any modern operating system would see "isnt" and immediately correct it to "isn't". Same for "ive", "dont", and so on. I can understand the apostrophe being multiple taps away, and so not getting added manually, but why wouldn't software put it back in?
I'm tempted to just remove the VLC based audio playback. We have yt-dlp support anyway, and VLC just seems to be causing issues more than it's solving them. Thoughts? Should I keep it or remove it?
Depends on how you look it it. Whenever I couldn't sleep as I child, I went and grabbed my Gameboy to play Pokémon for a few minutes/hours to make my brain tired and slept afterwards. Worked quite well actually, and still does today.
#TimeToTehOn...Good morning. I like FastSM rather more than tc, only probably because it loads faster, and takes up less RAM space. However, I won't be running it networkly like I can TC, for the simple reason that I cannot localise user data, or not that I know of. I don't use Telegram that much on the pc anyway, because most who have my details there, also follow me on mastodon, so no use boring people on telepaw if I already posted here anyway. BlueSky I don't have, and probably will never have either...Between my facebook/mastodon.africa/fluffy.family accounts I already have 3 social media related accounts, and that seems to be enough for now. However, knowing me, I may just fall for Bluesky, like I fell for the Seika Mini2 Smart Braille Notetaker...so I'll never say never, simply because I know myself well enough to know that curiosity might indeed push me to get bluesky regardless of the current reluctance. Anyway, have a nice day on this humpday.
@Bri Aha, thanks again. I thought because this is a beta, such things would not be possible, but thanks for the information that I can indeed localise data.
A veces me pregunto medio en serio si debería haber un partido que representase los intereses de los empleados públicos, y la calidad de los servicios públicos. Luego se me pasa.
The focus thing works, except when you try to arrow through the list of branches, it also moves focus to the commit field again. So probably should only default to focusing on the commit list when that dialog is first loaded.
It occurs to me that I haven't yet tested FastSM's ability to upload media to Mastodon. So I hereby present something which I call, "Girl, You Know it's False." I take the intro to the Milli Vanilli song "Girl, You Know it's True," where a man and a woman are talking to each other on opposite stereo channels, then I deliberately use and normalize the channel where each person is *not* talking. What you hear is a really echoy conversation, with almost certainly tape hiss and even some 50 Hz hum in the background. Then at the end the two orchestra hit type things also sound weird. Anyway here goes.
Well, I got GPT5 Pro mode again, asked it to help refactor my driver because I don't do a good enough job myself of that, and now it's been thinking for 88 minutes. Either it's going to give me a big disaster pile, or I somehow got it stuck. Longest time I've seen it think was like 65 minutes haha. But what the heck can you do for 88 minutes! This better be good. I feel like I'm asking a real coder to do the work, but worse, because it's not an actual real coder but a program, so it's more disappointing.
@stormproductions I agree. Gpt sucks. It's trash and it's gotten horrible particularly since uncle Sam keeps rushing to put new models out to try to keep his userbase without properly testing said models to be sure there's, y'know, actual improvements.
And to nobody's surprise, every iteration of gpt 5 has been a complete disaster.
@xogium @stormproductions I'm a noob vibe coder so nothing I say matters, but all three of my NVDA addons were done with GPT 5 and they all worked after a couple hours of back and forth, for what that's worth.
@stormproductions @xogium I honestly haven't found it difficult with this. I talked through what I wanted, it happened. Does Gemini have a decent GUI? I don't feel like doing all that from cli if I don't have to.
@FreakyFwoof @stormproductions @xogium Gemini's cli has a special --screen-reader flag that is really nice. It's why I keep them around honestly it's the nicest to use.
@serrebi @FreakyFwoof @stormproductions @xogium oh interesting! May give it a try at work, we have access to both. I have Google One plan which I share with my partner so we both get Gemini, they prefer it and honestly OpenAI has never allowed for family sharing like that, quite unfortunate. I feel like I need both in my toolbox. OpenAI (pro for now, we'll see if I make enough use of it this month to count) for those "think through and patch files direct inside container for you" moments, Gemini for the "review my code I wrote and that it's not trash and passable to an AI to work on", and they both look at different areas of the code. GPT will get it done even if you give it a bad idea to get done, Gemini's for sure a lot more cautious. So I can't decide on either these days, usually my workflow is augmenting one's response with the answer of the other, which makes me wonder what a powerful tool could be built if it just passed your prompt between the 2 AIs to solve, I'm sure someone's thought of that by now.
@FreakyFwoof @stormproductions @xogium I'm terrible for pointing this out to you, it probably doesn't apply but they give you an individual youtube premium sub with ultra. I wish it was family , then I'd think about it at least.
@serrebi @stormproductions @FreakyFwoof @xogium yeah, even in the "AI pro" plan :) + the 2 TB Gmail is nice, well, that's why I was initially subbed to it but they migrated people over with a slight price bump. economically speaking I think if I wanted to get the most use out of AI, Gemini does win by a longshot, even if GPT's now trying to compete with their $8 "go" plan LOL
@stormproductions @FreakyFwoof @xogium Gemini makes good early decisions for me when it comes to accessibility, but if I need to fix something like in the table output or something nitty griddy like that, I still find myself going to Codex.
@serrebi @FreakyFwoof @xogium I have installed Gemini's CLI, but I'm either too stupid with CLI's in general, or I don't even know how to navigate with NVDA. I tried typing /settings so I can see what settings I can tweak, and I couldn't even see the darn list of settings with NVDA. Any tips? I mean, any shortcuts for either NVDA or Gemini's CLI that I have to learn to start using it properly? Where do I actually pass the screen reader flag? My questions probably sound sily and stupid for somebody used to CLI's, but I truelly can't get the hang of it.
@serrebi @stormproductions @FreakyFwoof @xogium ahaha OK, that does make more sense. Yeah, even Codex always asks you to confirm things like pull requests so a flag for that isn't bad.
@serrebi @FreakyFwoof @xogium Okay, just tried the --screen-reader flag, it opened up, and said "(Use Enter to select, Tab to change focus, Esc to close)". I tried /settings, and tab just says something like checked 1, 2, 3. I can't really understand how to navigate, LOL.
@stormproductions @FreakyFwoof @xogium Use up and down arrows to modify settings, escape to save, I think. It's been a while since I modified settings. It repeats the last settings item as you arrow if I remember correctly.
@serrebi @FreakyFwoof @xogium Up and down arrows just keep telling me to use Enter to select, Tab to change focus, Esc to close. They either changed something, or I'm too dumb to understand how this works.
@stormproductions @serrebi @FreakyFwoof @xogium hm, if you review it in object nav does it put any symbol next to the highlighted choice? You might manually need to see if the lines change that way, just review the console's text, this is really good insight for me too though when I start using it for work.
@serrebi @FreakyFwoof @xogium If I use object nav I can't understand exactly what option is selected, thus I don't know what's gonna change if I hit enter.
@stormproductions @serrebi @xogium I haven't tried this myself, but try going up a few times, hopefully it doesn't rap. If not, then you can just count down and work out where you are.
@xogium @FreakyFwoof @stormproductions @serrebi oh gosh. that's aweful. Do you know if it behaves any better with Mac OS VoiceOver? For work I'm using both side by side, so perhaps I'd have more luck with Gemini Cli under Terminal? Even though VoiceOver's terminal support isn't known for being great, I do wonder if it would do better.
@FreakyFwoof @xogium @stormproductions @serrebi If I'm not mistaken, TDSR can be made to work with WSL2 on Windows. There's TDSR-server that needs to be added. It's an interface between TDSR and NVDA or JAWS.
Get tdsr, type tdsr github in google, and it's by tyler, if you know tyler, pretty easy to install, both for mac and linux I think, not sure if there's a windows version.
@tardis @xogium @FreakyFwoof @stormproductions @serrebi yeah, I freakin' love TDSR, Tyler did a really good job on it and it lets you review screen chunks and doesn't do that weird skip mid-utterance VoiceOver does. Love love love it for work machine :D
The text mode lie: why modern TUIs are a nightmare for accessibility
The mythical, it's text, so it's accessible
There is a persistent misconception among sighted developers: if an application runs in a terminal, it is inherently accessible. The logic assumes that because there are no graphics, no complex DOM, and no WebGL canvases, the content is just raw ASCII text that a screen reader can easily parse.
The reality is different. Most modern Text User Interfaces (TUIs) are often more hostile to accessibility than poorly coded graphical interfaces. The very tools designed to improve the Developer Experience (DX) in the terminal—frameworks like Ink (JS/React), Bubble Tea (Go), or tcell—are actively destroying the experience for blind users.
The Architectural Flaw: Stream vs. Grid
To understand the failure, we must distinguish between two distinct concepts often conflated under “terminal apps”: the CLI (Command Line Interface) and the TUI.
The CLI (The Stream): This operates on a standard input/output model (stdin/stdout). You type a command, the system appends the result below, and the cursor moves down. This is linear and chronological. For a screen reader, specifically kernel-level readers like Speakup, this is ideal.
The TUI (The Grid): This treats the terminal window not as a stream of text, but as a 2D grid of pixels, where every character cell is a pixel. It abandons the temporal flow for a spatial layout.
Case Study: The gemini-cli Madness
Let's look at a concrete example: gemini-cli, a tool written in Node.js using the Ink framework. On the surface, it looks like a simple chat interface. But underneath, Ink is trying to reconcile a React component tree into a terminal grid.
When you use this tool with Speakup (Linux) or NVDA (Windows), the application doesn't just fail; it actively spams you.
Because the framework treats the screen as a reactive canvas, every update triggers a redraw. When the AI is “thinking,” the tool updates a timer or a spinner. To do this, it moves the hardware cursor to the timer location, writes the new time, and moves it back.
For a sighted user, this happens instantly. For a screen reader user, this is what you hear:“Responding... Time elapsed 1s... Responding... Time elapsed 2s... [Fragment of chat history]... Responding...”
It drives the screen reader mad. The cursor is teleporting all over the screen to update status indicators, spinners, and history. Speakup tries to read whatever is under the cursor at that exact millisecond. You end up hearing random bits of conversation mixed with timer updates, making it impossible to focus on what you are actually typing.
Worse, lets pretend that you've somehow managed well with speakup so far, but that you want to do some work with nvda. Maybe paste an error you're getting on windows. So you open your terminal, ssh into your linux box, attach to your screen session and paste your text.
The result is an immediate crash of the screen reader (NVDA) or massive system instability. Why? Every time you type a character or paste text, the application triggers a state change. The framework decides it needs to re-render the interface. Because the conversation history is part of that state, the application attempts to redraw or re-calculate the layout for thousands of lines of text instantly. The more messages you have in a conversation, the more this will happen. And no, you can't just avoid this by using insert+5, the key combo supposed to avoid announcing dynamic change of content.
The Lag Loop
Furthermore, frameworks like Ink running on single-threaded environments (like Node.js) suffer from massive performance degradation when the history grows. If you paste a large block of text, the system has to calculate the diff for thousands of lines.
This causes input lag. You press a key, and you wait. You can wait up to 10 seconds for a single character to echo back. The system is too busy calculating how to redraw the screen to actually process your input.
Why The “Old Guard” Works (nano, vim, menuconfig)
Sighted developers often ask: “If TUIs are bad, why do you use nano, vim, or menuconfig?”
The answer is not that these tools handle the cursor perfectly by default. The answer is that they allow you to hide the cursor entirely.
1. Hiding the Cursor (nano, vim)
In tools like nano or vim, usability depends on turning off features that track cursor position. If you run nano with options that show the cursor position (like --constantshow), or if you use vim without specific configuration, the experience is broken.
When the cursor is visible and tracking is active, Speakup prioritizes the cursor's location update over the character echo. Instead of hearing the letter “a” when you type it, you hear “Column 2”. You type “b”, and you hear “Column 3”.
These older tools succeed because they allow you to disable this noise. You can configure them to suppress the visual cursor or status bar updates, forcing the screen reader to rely on the character input stream rather than the noisy coordinate updates. Modern frameworks rarely offer a “no-cursor” or “headless” mode; they assume the visual cursor is essential.
2. Single Column Focus (menuconfig)
Tools like the Linux kernel's menuconfig work because they enforce a strict, single-column focus. Even though there are borders and titles, the active area is a vertical list. The cursor stays pinned to that list. It doesn't jump to the bottom right to update a clock, then to the top left to update a title. The spatial complexity is kept low enough that the screen reader never gets “lost.”
3. The Lost Art of Scrolling Regions (Irssi)
Irssi is the gold standard for accessible chat, but not because of luck. Irssi was built over 20 years with a custom rendering engine that utilizes VT100 Scrolling Regions.
When a new message arrives in Irssi: 1. It tells the terminal driver: “Define a scrolling region from line 1 to 23.”2. It sends a command: “Scroll up.” The terminal moves the bits up. 3. It draws the new text at the bottom of that region.
Crucially, it handles this in a way that minimizes interference with the input line. It relies on the terminal's hardware capabilities rather than rewriting every character on the screen manually. Modern frameworks ignore these hardware features in favor of “diffing” the screen state and rewriting characters, which is computationally heavier and hostile to accessibility.
The “Stale Bot” excuse: A Case Study in Neglect
Google and the maintainers of gemini-cli pretend to care about accessibility. “Pretend” is the operative word here. If you look at the repository, critical accessibility regressions like Issue #3435 and Issue #11305 have been left to rot. There is no discussion, no roadmap, and no fix. Even worse is the fate of Issue #1553, which was supposed to track these accessibility failures. It didn't get solved; it got silenced. It was closed automatically by a bot with this generic dismissal: > Hello! As part of our effort to keep our backlog manageable and focus on the most active issues, we are tidying up older reports. It looks like this > issue hasn't been active for a while, so we are closing it for now.”
This is unacceptable. Closing an accessibility report because the maintainers haven't touched it in months is not “tidying up”; it is hiding evidence. It effectively says that if a bug is ignored long enough, it ceases to exist. It boosts the project's “Closed Issues” metric while leaving the actual software unusable for blind users.
Conclusion
If you are building for the terminal and care about accessibility, stop using declarative UI frameworks that treat the terminal like a canvas.
The “modern” TUI stack has optimized for the developer's ability to write React-like code at the expense of the machine's ability to render text efficiently.
If you cannot guarantee that your application allows the user to hide the cursor, or if you rely on aggressive redrawing to show spinners and timers, you are building an inaccessible tool.
For the blind user, a dumb, linear CLI stream is infinitely superior to a “smart” TUI that lags, spams, and scatters the cursor across the screen.
Well, I use gemini via the web. Dumb way to use it I guess, but it's better. I dislike the terminal, unless I am playing some game that requires that to be running, or installing apps via brew, I stay far away from that monster.
@tardis @FreakyFwoof @stormproductions @serrebi It works, up to a point. And I've done a whole lot this past month alone to really know what I'm talking about and not just say nonsense.
The web ui will work but when your code is now spanning multiple files -- over 40 in my case, it creates problem. Gemini on the web constantly loses track of which file does what, which leads to very stupid things like:
Me: your fix does not work, I still get the error about session not properly disconnecting Gemini: well of course it won't work, I've analyzed your code and it's still the older one, do this and it will work. Me: ... of course it doesn't work, and besides you're refering yourself to an old version of this code.
Yeah, I noticed that too. But I haven't used it much for coding, just for messing around with it. Why not just use claude? IT's better at coding shtuff.
@tardis @FreakyFwoof @stormproductions @serrebi Or rather it is too expensive for the limits you have. Gemini with google AI pro gives me 100 messages daily with the pro model. Claude is much, much lower than that.
I can't code either, which is why I talk to an AI and get it to do it. I play music, not program anything. Wish I could, but now at least I can fake it till I make it. Seems to be working though lol @tardis @xogium @Tamasg @stormproductions @serrebi
@tardis @FreakyFwoof @stormproductions @serrebi Friend of mine tried it and it was far from being impressive. You're supposed to get 40 messages every 5 hours with claude pro, but that limit can and will go down really really fast if you have a lot of context, which is only 100k tokens max.
@xogium Yep, I have switched away from GPT, and won't be using it at least till they stop making new models just to up that number and not really add anything interesting.
@stormproductions If they are still up after all the crap they go through right now, that is...
Trials for the death of several people that trusted gpt with things, claims that chatgpt can even be trusted with health advices now when it clearly can't -- I'm not joking! I saw an ad about it!
Then also the Musk case. Oh dear oh dear. Now I don't like Musk at all but truth is, he's apparently got 60% chance of winning his case against OpenAI. And he asked for a jaw dropping $134b iirc.
@xogium Wow. Didn't even hear about that - I don't really follow such news that much. Well... I'm a happy Gemini Pro user and I won't be switching too soon.
@xogium @stormproductions I do use Gemini, quite often to double-check code GPT wrote, because it will scan through it a lot faster. But (at least on web) it's horrible at pasting back code you gave it exactly 1 to 1 still, like it'll omit sections that it's pasting back of the code I gave it, and I'm yelling at it "give me full pieces! No omissions!" Haha. It's gotten better but GPT having an entire container inside to access and run Linux commands, not to mention I've uploaded a 30-40 MB zip file to it with my entire Gitub repo and Gemini just yells at me, "file is too large." LOL.
Pravidelné kombinování různých druhů fyzické aktivity může být nejlepší pro prodloužení délky života, ale souvislosti nejsou lineární, což naznačuje možný optimální prahový efekt www.news-medical.net/news/2026012...
Regularly doing a mix of different types of physical activity may be best for prolonging the lifespan, but the associations aren't linear, pointing to a possible optimal threshold effect, suggests research published in the open access journal BMJ Med…
God, Mac OS is really awful these days! A notification comes in, no VO focus on it, so hit VO+F1 twice, notifications menu has "go to Notification Center" and..."No items in Notification center!" Also, "Go to Notification Center" doesn't actually do anything! If something this basic doesn't work, it's no wonder VO is so bad. Case in point, every time my Mac restarts now, it comes up saying "Installer has no Windows." I have to use VO+Shift+Space to click the mouse because it's hanging out in the Login Window, while Vo's focus refuses to move from this blank Installer Window no matter what. Pressing VO+F1 just says, "Authentication required." Shit, maybe I just need to blow the OS away and start over.
I did that last week and it actually did fix most of those weird behaviours. Not that VoiceOver itself is any better in things like Web navigation or text documents, but it clearly got rid of some odd bugs
New: NV Speech Player finally gets a "pause mode" setting, like Eloquence has. This should not chew into lines with say-all, but rather does clause-based punctuation pause detection. And, it does appear to work. You can choose between "off," "short," and "long." This should solve a lot of complaints around words "merging together" after punctuation. I also began my first pass at abstracting parts of the driver out: _frontend.py and _dll_loader.py. Logging was also improved, I cleaned up over 40 try/catch exceptions, We're not trying to play Pokemon here. download: eurpod.com/synths/nvSpeechPlay… Oh! and for those wondering "where's classic pitch?" It's located in the Voice settings dialog, you'll find it as a checkbox.
thanks for the last corrections concerning m/n between vowel and consonant for pt-br. Now, I've just added a phoneme which I've called "brã", for Brazilian ã, which consisted of a single change: take ã and change cf4 from 3300 to 2600. then I added to pt-br.yaml: - from: "ã" to: "brã" Now, when I command it to speak a word like "região", the new phoneme appears to be used, but it's much less stressed and somewhat shortened, or at least de-emphasised. Can you give a look please?
@clv1 hmm. try and replace your packs folder at eurpod.com/synths/packs-ptUpda… - I'm experimenting on this one. What you’re hearing isn’t really “less stress,” it’s that the new vowel is darker, so it feels less punchy even if the timing didn’t change. In our latest pack we kept your approach, but made the change in a more balanced way so the synth doesn’t “half use” the old sound: we added a dedicated Brazilian ã variant and mapped pt-br to use it, making sure the vowel stays consistent across the voice filters. If you still want it to pop more, the best next tweak is a tiny loudness bump on that new ã variant rather than changing stress/timing.
@asael @clv1 haha not sure. Depends on how well base pairs are defined right now. You could always add it inside the driver, here we have the list of languages, if you formatted it the same way as that string and used HR for Croatian it would speak today but with perhaps missing sounds. Then we map those missing sounds to specific phoneme pairs in hr.yaml. Generally many sounds can be reused across languages but there's definitely edge cases especially with glides, fronted sounds, ETC.
@asael @clv1 it should be. I got one report of it not working on the latest alpha, but I pulled it to try and it still loaded, so I'm really not sure what could be going on. If you do get errors the debug logs help a lot because they have it set to max level in the alphas too.
oh, eurpod.com/synths/nvSpeechPlay… - the base one is always latest, I branch versioned files because people's Browser cache gives them the older version if I keep the file name the same.
@asael @clv1 oh that's weird! Wonder if it was when I moved the DLL functions into their own helpers. You could try: eurpod.com/synths/nvSpeechPlay… - if that one works, we know it's changes I made to cleaning up the driver that introduced them.
@asael OK, regarding separation between consonants, it appears to be the option cluster gap stop clusure, which is set to 26. I changed it to 126 to experiment. It has an effect in words like "multi", where l and t get separated, but no effect in syllables like fra, cra, tra.
also mess with stopClosureVowelFadeMs: 8 stopClosureClusterGapMs: 28 stopClosureClusterFadeMs: 5 If we need we can introduce a new engine rule to detect these cases, but worth seeing if anything existing helps improve it.
Thanks, but I've found a more inteligible solution, same as used by ESpeak AFAIC: clone schwa and disable the _isVowel parameter. then, there must be a rule where the synth inserts this schwa variant before the r when It's between consonant and vowel, Thus, the syllable "tra" for example shouldn't be trˈa, but tərˈa
Thanks, Tamas, but after experimenting more here, I've decided to change this vowel in another way, i.e. revert cf4 to 3300, and change cf1 from 650 to 550. Likewise, I did a similar thing to õ: change cf1 from 540 to 440, so it becomes equally nasal but a closer vowel. Please do these changes in the standard pack.
@clv1 ooh, this is exactly the type of feedback I like to see :) thank you! For sure will give these a go. I really do like this pack system because people don't need to update their entire driver and it makes communicating value changes easier, so this is awesome, made my day better :D
Now to another issue: I need the distinction to be more noticeable when there is an r between consonant and vowel, i.e. in syllables like bra, cra, fra, tra, the consonants probably need to be more separate from each other. I have added some settings to pt-br.yaml, namely those concerning stop gaps, but I don't know yet how to deal with them... What is the correct setting for this case?
Our parliament holds 101 members + bunch of officers in the parliament + a ministry which deals with finances. They recently passed a law, which, apparently by error, made all gambling websites except from paying taxes in Estonia. Loss in tens of millions of euros, which in Estonias view is a considerable sum. Now, to avoid this situation in the future, they are talking about implementing AI in law making and in the process. What could possibly go wrong.
Whatever the AI says is right. Yours is probably better than mine since your AI is the best AI the world has ever seen. Therefore you are the best person, with the best skills. Just an FYI, I'd like to purchase Estonia because why not? I could make Estonia a big data center. I need it because @BorrisInABox and @FreakyFwoof are all running AIS on their Macs which might be bigger than mine and yours combined. So gimme. Gimme gimme gimme ... urp.
My partner and I visited Norway and hiked (a surprisingly short hike) to a place called Litlefjellet -This was the view! Litlefjellet, Norwegian for ‘Little Mountain’, is the name of the center mountain. Plus, we found the iconic pond that perfectly reflects the whole scene. I loved it so much that I decided to make it into wood art! I hope you love it, too #woodart #artandairplanes #art #woodworking #wood #norway #mountains #hiking #sun
Loving the trend of all of these new free lightweight apps, both human and AI coded. Just in the past few months: * New lightweight Mastodon/BlueSky client * New Github viewer * New ebook reader * New RSS reader * a web-based typing tutor * And many more I'm missing. What other cool apps do we need to shoutout?
I made three NVDA addons. WikiSeek: NVDA+w, Type your query, searches wikipedia, comes up with a result in a text field. Cuts out all the guff. Downside it cuts out tables too, but you have the option of opening up the article in browser too. Configurable in settings as to whether it auto-opens, and which wikipedia language you want it to pull articles from.
dateParser: NVDA+E. Enter a passed or future date to find out how many days, weeks months or years ago it was either for curiocity, or to find out when a special date happened or will happen for you in the future. You can say things like: 2025-05-01 or 35 Mondays ago, or 19 Tue from now.
Minuter: NVDA+Shift+Backspace. Enable/disable beeps, spoken seconds or use the supplied sound files to go off every second, 5 seconds before the top of the next minute and at the top of the next minute. Your settings remain active even when the dialog is closed. All can be downloaded from here: onj.me/nvda
@serrebi Anything created by @midzi , including SARA - which is accessible radio automation software and some GUI tools for Shazam on Desktop: github.com/michaldziwisz
Working on a real beauty currently with the extensive help of gemini 3 pro and a sighted friend to help fix up the UI. It's not ready for release yet, but I reckon I'll post the link for when it is ready in your thread, if you'd like.
SonarIRC is the first 100% accessibility first irc client ever made for windows. No more of this horrible half hacked together accessibility support via scripts in mIRC or others of the same. It is also up to modern irc standards and has a ton of irc v3 capabilities supported. Compatibility also with all screen readers that uses UIA.
@pixelate I didn't want to put that out there since it's on a private web site but yes, it's new too. Also a new telegram client, but I haven't used it and don't know the URL. I'm told it's on a telegram group. Easygram?
My wife and I are fortunate to be in Iceland at the moment, and we managed to capture a timelapse of last night’s aurora. It was one of those moments in my life when I could hardly believe my own eyes. Enjoy! #aurora #iceland #timelapse
Our system has been all weird since our stressful weekend. I'd like it to not. I'm sure we'll be okay, it's just annoying that stressors continue to effect our headspace and system behavior for a couple days after. - Samara
Relatable. And I'm sorry. Here it's either everyone distressed, physical body pain, or daydreaming exessively, like, can't stop even if I want to. It'll all get normal, but I advice to usually checkin with everyone, to see if something is bothering one of you and to just provide each other comfort. Safety in numbers, and unity makes system stronger etc. I hope you figure out the headspace stuff soon. And you all get a little virtual hug.
@tardis We're not very touchy people actually. But yes I agree on the rest. Definitely there is a better caretaking network with all of us present. I think it's just lingering stress from all the weird stuff that happened on the weekend, I'd say it'll likely all clear by the end of the week, if even that long. We'll be okay I'm sure. - Samara
Not a touchy person here, either. Unless I know the person. If I don't, I freak out, then someone else takes over, because stressful, triggering, to be touched randomly by someone.
@tardis Yeah, that. People trying to do touch stuff without our initiating are just something we don't prefer. Especially if we've repeatedly told them to quit, and they keep doing it anyway. We regularly have to put up with that, very frustrating. - Kylie
If you run a mastodon instance, it's time to upgrade. There are some security issues fixed in the version released today: github.com/mastodon/mastodon/r…
Upgrade overview
This release contains upgrade notes that deviate from the norm:
ℹ️ Requires assets recompilation
For more information, view the complete release notes and scroll down to the upgrad...
Got this error when updating to latest version. File "platforms\mastodon\__init__.py", line 3, in <module> File "pyimod02_importers.py", line 457, in exec_module File "platforms\mastodon\account.py", line 8, in <module> ModuleNotFoundError: No module named 'cache'
Oh goodness, the internet is saying that the new Trek show that just came out is super cringey. Why oh why do they keep pulling this garbage? I'm all for DEI stuff, but can't it be done without making it so cringey that people just check out? Honestly, as a person who is both blind and gay, it's kind of insulting that something this aweful is supposed to help people like me feel more excepted and whatnot in society. #GetALife
The applyVoiceToFrame function loops through all frame fields (47) for each frame, which is heavy. Since only a handful of fields are modified by voices, precomputing the operations for each voice once could significantly boost performance. By mapping the voice operations to a precomputed list, I can reduce the 94 lookups per frame. That will help things out here a lot I think.
Tuta
in reply to Stomata 🥼 • • •Fuse
in reply to Tuta • • •Yes. The fact that the variations between clients and the browser interface are *tiny*, and are limited to platform specific areas, is one of Tuta's HUGE strengths.
Well done, Team Tuta!
@Stomata
Tuta
in reply to Fuse • • •