A while back I saw someone on here mension a setting you can set in #ClaudeCode that would make it work better with a screen reader. Can anyone remember what it is? I think it might have bin @ZBennoui maybe? But if not sorry for the mension.
They are going to put ads into ChatGPT, apparently.
They should go ahead and put all the ads in there. Just load that thing up with ads. Go on. Do it.
Force ads into all the gen AI stuff too. Involuntary product placements. Unavoidable Coke cans in your expertly prompted masterpiece. At least one verse of your Suno banger singing about New Persil Automatic.
Popups for the latest mobile game in your vibe-coded app.
Hello hello. Lots of updates to the phoneme editor today. It now supports: - as you type a new value for any fields, you will hear the voice speak the phoneme. - Voice settings dialog added to settings menu. Here, you can mess with all 47 parameters or change to one of the preset voices. Enjoy. eurpod.com/synths/NVSPPhonemeE…
@x0 you have to first go to the settings menu > choose the dll location options, then file menu > open pack to choose the root folder for it where \packs lives.
@x0 the good thing is that it's a one-time process :) It creates the .ini file with those stored so next time you re-open there's no setting them again.
@x0 for flexibility. If someone wanted to modify a pack located elsewhere or their Espeak is a differentbuild in another folder, could be useful. But yeah, might be able to set it to where at least it checks it first to see if they exist, small enough tweak to do it.
Also, could we have either a spinbox or sliders or something in the phoneme editor for some of the numeric values? So we could do PRAAT-style sweeping instead of typing, which doesn't overwrite.
@x0 We do have diphthongs but we build them as sequences (vowel + glide, or vowel + vowel) rather than dedicated "diphthong phoneme" entries. That’s common in formant synthesis.
In the phoneme editor at least, though, it sounds horrendous. I tweaked the U.S. R phoneme slightly and wrote the phrase "I ran over a rat in my car" to test it, and it couldn't even say the word 'I' correctly. And car sounded like it snapped. Very disjointed, no glides.
You should probably emit diphthong phonemes, unless eSpeak actually doesn't, and serialize those internally aas glide ends to the formant synth yeah? Or does IPA not have diphthongs as individual symbols. In U.S> English at least I think some of the diphthongs differ from the vowels used to compose them.
@x0 so IPA usually doesn’t have a dedicated symbol per diphthong. It’s written as a sequence of two vowel qualities (sometimes marking the second as a glide). eSpeak follows that pattern most of the time, so it won’t hand us "diphthong phonemes" directly. But there's probably a bit more engine rules that can improve there.
@x0 I think I can find a possible solution: adding an "autoTieVowelPairsEnabled" feature, but since languages like Finnish can have long vowels or diphthongs, it should be opt-in to prevent issues.
@x0 yeah, universal detection is complex, as hiatus vs diphthongs are language-specific. So we'd have to make that a setting and then phoneme overrides would be made for those in the language file.
I know not the same thing, but I would love to upgrade my mixer, but I do not want digital crap that I can't use. I just need one more, maybe two more stereo inputs, and at least one more stereo output. the control room on my current board is mono only , which makes me wonder what they were smoking to have something like that mono, but what ever. sadly I don't think i am going to find what I want. oh sure, I could go to the 16 channelk mixer that I have the 12 channel one of, but I don't think that is going to fix my stereo output things. it might get me another stereo input though
@JamminJerry My last mixer had zero stereo inputs, and I was perfectly fine with that. 16 mono that you could do whatever you want with. Pair them, or don't. Main plus four mono subgroups (paired for two stereo), 8 mono aux sends, and direct outs for all 16 channels if you needed those, plus inserts on all channels. Dynamics processing on each channel, bus, send and FX return. And I still had everything maxed out. These days, I just do everything in-the-box, which is easier on Mac OS than Windows, and only have one audio interface with four physical mono inputs and outputs, and it does all I need. Funny how things change.
yes it most certainly is. well right now, TeamTalk has its own soundcard, and so that it doesn't hear its self, it is what listens to the control room, but the control room is mono! grrrrr. no stereo for the people that hear me on TeamTalk.
I don't do such things lately, but I have a setup for that, all using virtual audio devices. I can basically send anything to anywhere, minus what I don't want to exist, with as few or as many channels as needed. And, because Mac OS/core audio, the latency isn't particularly noticeable in that context.
I tried virtual audio cable, and even set to real time, I just didn't like the delay, and it was just all sorts of complicated for me. hents why I got the mixer, and a few sound cards.
@JamminJerry There's nothing quite like Loopback/audio hijack for Windows. I definitely don't miss having a bunch of audio interfaces strung together. Makes for a dirty setup. Easy to route stuff though.
@JamminJerry I did that for a short time. Put an in-line resister on the input of the phone to get it down to mic level, stuck a side-chain compressor on the channel strip to hide the loopback coming from the phone. Eventually got a real phone patch, didn't keep that long, went to a SIP solution.
Near the end of when I was doing live stuff, I got a Behringer audio interface with 8 inputs and 10 outputs, all balanced TRS, which I used a couple of I/O pairs in combination with the on-board 32 in, 18 out interface on my mixer, mostly for mundaine stuff like taking phone calls and auxiliary audio patching, leaving the board to stream the high quality stuff since it didn't have to get converted back to analog internally. Now none of that stuff matters to me anymore, really.
I just wish on the current board I have you could tell it what to send to the USB C port. if I could that would be perfect, as I wouldn't have to use the damned control room. lol.
I didn't even do that. I just baught the eighth inch to two RCA wire, and then the adaptor for the eighth inch side to make it fit into a chordless, and hooked it into the board, and ran with it. no no resisters, or anything between the chordless and the mixer. I don't think it sounded too bad. I did turn the volume down on the chordless though because it distorted all to hell if I didn't, and then controled the volume threw the board.
@JamminJerry Yeah, that works in a pinch, but you'll get a pretty high SNR on the input unless you attenuate it on the other side. Been there, done that, fixed it real fast because I didn't like the results.
I made my first 2 #FT8 contacts today thanks to a new accessible companion app built to work in conjunction with #WSJTX and the #JAWS and #NVDA screen readers. We're making huge progress toward a solution we can provide to the entire #blind #AmateurRadio community. #HamRadio
I’m extremely happy this is being developed. I worked with QLog and it’s self voicing interface for about half an hour yesterday and realized very quickly that using it on a regular basis would drive me completely insane!
Já neříkám, že vždycky hned vím, jak to je. O to nejde. Jde o to, že se tomu v Čechách nevěnujeme dost. A pád Ameriky je něco tak bezprecedentního, že vlastně nechápu, jak to, že já jsem ten největší alarmista. A u toho se mě lidé ptají na moje mentální zdraví. To je fakt úlet. Stojíme na prahu zničení demokratického způsobu vládnutí a já mám být alarmista?
Já neříkám, že vždycky hned vím, jak to je. O to nejde. Jde o to, že se tomu v Čechách nevěnujeme dost. A pád Ameriky je něco tak bezprecedentního, že vlastně nechápu, jak to, že já jsem ten největ…
Someone asked me how a blind person can do phoneme editing without a spectrogram. That's valid. The next version of the phoneme editor will let you preview the sound in real-time as you edit it. That is, while changing the value, it will play for you the new one you set. This will allow for realtime changes and tuning. Eventually, I will add an option to synthesize through an AI service, but that will be long long down the road. For now, just hearing the new sound will allow people to have a better workflow.
.Hřib je zase bomby u Moravce 😀 Ten, kdo se především zabýval příčinami války, byl pan Turek, protože toho hajlujícího moštáře jste vzal na výlet do Kyjeva
Well, eurpod.com/synths/nvSpeechPlay… updated with Language pack data. I will be doing less frequent updates (maybe 1-2 per day at most) to allow for people to work with the language files and have time to do so. Now, languages like Bulgarian, Romanian have specific letter rules. Hungarian actually is starting to sound like a native speaker talking. This is a solid language pack change, and we fixed the British "broad" to sound correct, too.
I should grab this and give you feedback about bg. Is it a separate voice? Unrelated to sapi or win core? If so, then awesome. We can make it sound native too. ;) :)
@tardis yeah, for now NVDA only, although the Phoneme Editor is the first to let it run outside of NVDA. A SAPI driver is not out of the question, as we already have one for BestSpeech, building a SAPI wrapper isn't the worst challenge. So it'll happen just later down the road.
Der Service der DKB sorgt für mehr Barrierefreiheit im Sport
Ab sofort können alle Länderspiele der A-Nationalmannschaften über den kostenlosen Webplayer verfolgt werden. Die Übertragung beginnt jeweils 10 Minuten vor dem Spiel.
I'm getting a bunch of these lately. Usually two a day, a couple of minutes apart from each other, calling a bunch of people in my area code. FACETIME, Grouped, now, +212 1 00 02 51 7, +212 6 25 11 39 30 & 16 others , Missed FaceTime Call
I get those... But as texts. It's the first 6 digits of my phone number and the last 4 that change... There's like 20 "people" in the group text. I dunno.
next week is my development talk, I'm scared y'all. It's funny, dev talks make you think it's some kind of engineering presentation when it's completely different and nervwracking. Ugh.
Európsky parlament prijal dňa 18. decembra 2025 uznesenie o masových únosoch detí v Nigérii, a to aj z katolíckej školy sv. Márie v Papiri.
Pripomeňme si, že 21. novembra 2025 ozbrojenci zaútočili na katolícku školu sv. Márie v Papiri v štáte Niger a uniesli 303 študentov, 12 učiteľov a zamestnancov, čo bol jeden z najväčších únosov v dejinách Nigérie. Bezpečnostná situácia v Nigérii sa naďalej zhoršuje v dôsledku ozbrojených gangov a džihádistických skupín, ako sú Boko Haram a Islamský štát, čo viedlo vládu k vyhláseniu núdzového stavu v oblasti národnej bezpečnosti 26. novembra 2025.
Parlament dôrazne odsúdil masový únos v katolíckej škole sv. Márie a všetky útoky na deti, učiteľov a vzdelávacie či náboženské inštitúcie. Vyjadril solidaritu s obeťami, ich rodinami a postihnutými komunitami. Požadoval okamžité a bezpodmienečné prepustenie všetkých zostávajúcich unesených a vyzval nigérijskú vládu, aby zabezpečila ich bezpečný návrat a vykonala dôkladné a transparentné vyšetrovanie.
Nigérijské orgány sa vyzývajú, aby prijali konkrétne kroky na posilnenie ochrany škôl a poskytnúť alternatívne vzdelávacie podmienky s cieľom zabezpečiť kontinuitu vzdelávania pre dotknutých študentov. Zintenzívniť úsilie o zabezpečení prepustenia všetkých civilistov zadržiavaných ozbrojenými skupinami a o lepšiu ochranu škôl a zraniteľných komunít. Zabezpečiť nezávislé a účinné vyšetrovanie únosov a stíhať zodpovedné osoby a zabezpečiť ich súdne konanie v súlade s medzinárodnými záväzkami v oblasti ľudských práv a zásadami právneho štátu. Prijať opatrenia proti neistote a náboženskému, etnickému a medzikomunitnému násiliu v regióne.
New Audio Player. I want to introduce you to an audio player which is new to me for both Windows and Mac. It is called the Flux Entertainment Centre. This is a highly accessible application. I've been listening to our station Team-FM with it this morning and it is very stable and reliable. Global shortcut keys are available and it even has variable speed and pitch playback. The only thing which is missing is the ability to determine elapsed and remaining time in hours, minutes and seconds. For many people like me that is an essential requirement. I must know an elapsed time quickly. Apart from that, it's excellent. You do not need me to spell out its features because you can read about it here. ims-productions.com/flux/
Flux
Do you stream a lot of content on your computer but are tired of dealing with clunky and constantly changing web interfaces?
If so, Flux was made for you.
Flux brings your audio and media experie
@luiscarlosgonzalez those I don't use, and in Pixelfed, for example, I don't see particular use for the blind. It's like photo photo photo, nothing more, I think.
@menelion @luiscarlosgonzalez I'm personally more likely to pay for software that commits to a core feature set and executes it well. Telegram support belongs in an accessible chat app, not a social media client.
@menelion @Bri @jscholes Yeah I also. One app for Mastodon, Bluesky, Threads, Misskey, Hubzilla, Frendica, Bonfire, Glitch, Sharki, Akkoma, Ploroma, Lemmi, GoToSocial, TruthSocial... wait not that omit that..., Another for Telegram
@luiscarlosgonzalez @Bri @jscholes Luis, I don't believe the app needs to support such a plethora of one-day solutions that keep growing like mushrooms, sorry. If a particular solution has an interesting function (like formatting support in Pleroma, for example), it's nice to have. Other than that... Mastodon/BlueSky/Threads, it's enough.
@Bri @menelion @luiscarlosgonzalez One underlying problem is that in 2026, there isn't enough accessible software being produced for Windows. So when someone comes along and demonstrates a willingness and ability to produce more of it, people want to see that production go as far as possible before it disappears.
Unfortunately that tends to encourage projects that lack polish and direction, burnout or disinterest on the part of the developers, and so on. In that sense it's a bit of a self-fulfilling prophecy.
Meanwhile on iOS, a suggestion that Telegram support be added to Mona would be perceived as fairly ridiculous. Partially because while it may not exist right now, there is more of an expectation that somebody could end up developing a reasonably accessible Telegram client on that platform even without having accessibility as a core intent.
@jscholes @Bri @luiscarlosgonzalez You're right, James, and I'll tell you the reason, at least how I perceive it: no one wants to pay, or those are rare who want. I'm starting such an endeavor right now, and I'm absolutely not sure I'll get paid.
@menelion @jscholes @Bri Neither do I, when ever someone creates the definitive fediverse app to unite them in one as a native app, but if you say that's too much... You've syncopated me
Yes, but with reservations. I would do it as long as it's a one-time payment and there's no online account and/or online activation. I'm always concerned that the developers will eventually lose interest in such projects and shut down the activation servers. The software would then become useless. That's why I don't like the concept of e.g. TweeseCake, even though it's currently free.
Recommending #Signal is like giving a man a slightly less poisonous fish. Introducing them to federated, provider independent instant messaging with #XMPP is teaching them how to fish.
never used WA, SIGNAL with some people, feeling at home with #matrix (#fluffy on ubuntu touch and Linux), being in #conversations with nobody (just once in a while - I guess people feel strangely alone there although I like it!) and testing #skred with somebody.
Ich bin jetzt mal ganz ehrlich zu euch und mache mich unbeliebt:
Ich brauche nicht noch ein Video, wo ein Mensch durch Paris radelt und sagt: Wie toll ist das? Das könnten wir auch haben!
Wir brauchen Menschen wie die Bruntletts, die klar sagen: DAS zu schaffen, braucht Rückgrat, Mut und Menschenliebe. Danke!
Das Video geht heute raus an alle Newslettter-Abonnent*innen. Alle 14 Tage gibt es hier Input in meiner Academy. Übersicht hier: katja-diehl.de/interviews/
Schönes Beispiel, wie man es nicht machen sollte: Wenn Bauanträge "nach drei Monaten ohne Widerspruch automatisch als genehmigt gelten", dann kann ich halt einen schönen Behörden-DDOS machen. Ich stell dann einfach in der Urlaubszeit schön per Serienbrief 823 Bauanträge und in der Zeit kriegen die die alle nicht fristgerecht abgelehnt
Eine schnellere Genehmigung für den Hausbau, den größeren Carport oder ein Nachbarschaftsfest: Die CDU will laut Generalsekretär Linnemann automatische Fristlösungen auf den Weg bringen, um die deutsche Verwaltung zu beschleunigen.
«Independientemente de la ley electoral —que efectivamente penaliza la fragmentación—, existe también una voluntad de conformar y construir un frente amplio sobre unos mínimos programáticos en los que estamos de acuerdo: la defensa de los servicios públicos, reformas estructurales del Estado, descentralización y políticas sociales».
🎙 Ya puedes escuchar la entrevista a Antonio Maíllo en ‘La mañana fin de semana’ de COPE
Hi! I joined Mastodon a long time ago but finally got fed up with mainstream social media this winter, so I'm probably going to be hanging around here a lot more. Read the alt text for details.
I've been doing a daily doodle challenge from the start of 2026, hope I can keep it up & see you later!
A quick #FOSDEM guide from someone who's been attending since it was still called OSDEM:
1. Be mindful of other people. This is an extremely busy conference. 2. If a talk is in progress and you want to enter the room: don't. Wait until the next talk. 3. If you didn't read 2, try to enter as quietly as humanly possible, and ideally even quieter. Many of the rooms have terrible acoustics and you will annoy hundreds of people. 4. The hallway track is very busy. Take it outside when possible.
⚠️ We're 900 € short on funds for our stand at #FOSDEM
FOSDEM is the largest free and open source software conference in Europe. Attending FOSDEM allows us to promote #Luanti, bring the community together, and connect with other open source projects.
If you can, please help by giving to our non-profit fundraiser. Any contribution helps lighten the load
It looks like Iran has successfully built a national 'walled garden' to control internet connectivity - both halting outsiders from communicating with those in Iran who are protesting, but also halting Iranians' ability to access the world wide web's myriad resources.
How well this 'wall' can be maintained remains to be seen but authoritarian regimes around the world will be watching, that's for sure.
The Observer this morning has a report on the Iranian protests that claims information has come our via satellite phone technology... so that may be one way?
I'd like to know people's opinions on this. I enjoying following people who post in a variety of languages. I also know I have followers who appreciate alt text in images and do my best not to boost images that have no alt text.
What do you do when the image has alt text but is in another language? To boost it as I would like to would mean those using screen readers are disadvantaged, as far as I know. To boost and comment with English alt text would seem rude to the original poster who has already included alt text. (I'm only using English as an example there as it's the one I usually communicate in). Do screen readers have their own translation functions by any chance?
If you're someone who uses a screen reader what do you do when you come across alt text in a language you don't understand?
Screen readers often have translation add-ons, yes. I'd say boosting without comment is fine; unless you'd also feel obliged to translate the text on the too.
Vystudoval jsem střední průmyslovou školu, obor Informatika, zaměření programování. Co se týče skutečné informatiky, měli jsme předměty Programování, Databáze, Počítačové sítě, Technické vybavení počítačů, Operační systémy, Webové aplikace, Počítačová grafika. Školu jsem dostudoval v roce 2019. V mé práci toho využívám nejvíce z předmětů Operační systémy a Technické vybavení počítačů. Kolik učiva z těchto předmětů je užitečného v mé práci? Ne, takto to není vhodně formulovaná otázka. Co v práci dělám a co jsme se neučili? Správa domény ve smyslu uživatelů, počítačů v Active Directory. Modul DHCP? Nikoliv. Propojení s Microsoft 365? V té době už to existovalo, neučili jsme se to. Exchange? Ne. Procesy v IT? Ne. Jak řešit problémy s Windows, technické příkazy pro opravu systému? Ne. Z tématu kyberbezpečnosti jsme začali a skončili u složitosti hesla. Doménové politiky v prostředí Windows? Ne. Za to umím editovat grub, znám snad všechny příkazy bashe, umím zakládat uživatele v terminálu, není problém pro mě konfigurovat apache, sambu, atd. Vím, co je to fragmentace paměti, stránkování, preemptivní nebo nepreemptivní plánování, vím, co je to žurnál. Vím, jak se ukládají data do paměti počítače. Vím, jak pracuje procesor, co je to instrukční pointer, registry, instrukce. Vím, jak využít výkon grafické karty v programování přes technologii NVIDIA CUDA. Umím Assembler, C, C++, chápu princip objektového programování. PHP, a další jazyky, na které si už ani nevzpomenu. A Novell, no. Všechno jsou to věci, u kterých jsem rád, že je znám, ale velkou většinu věcí, kterou dělám v práci, mě ve škole nenaučili. Naučil jsem se je sám, protože mě ten obor baví. Praxe českého školství je skutečně taková, že sice máte maturitu z technického oboru nebo titul, ale potom nastoupíte do práce ve stejném oboru a připadáte si hloupí, protože věci, které se v praxi používají, vás nikdo cizí nenaučil. Prostě to tak je.
@archos test do kamene, pamatuju si jeden pracovní pohovor (teda spíše rozhovor s recruiterem), někdy uprostřed jsem mu řekl, koukám tady na požadavky co máte a jste si vědom ze nemám VŠ? Úplně automaticky mi odpověděl, to neni problém, když máte více než 15 let praxe. Popravdě řečeno se mě na VŠ nikdo nikdy poslední dekádu a půl nezeptal. Až na jednu výjimku, a to byla jedna nejmenovaná česká konzultantská firma v ČR, která ani nebyla schopna mi nabídnout podmínky které už mam cca dekádu (Estonsko, Gibraltar - výše platu, bonusy, dny dovolené, zdravotní pojištění, etc…) a ještě mi řekla, ať se vrátím tedy vrátím do svého alternativního světa. Tak jsem ji slušnou češtinou upozornil, že to je ona kdo mě kontaktuje a poslal do prdele i celou firmou at už me nikdy nekontaktuje.
I installed #Signal and #Conversations_im on a clean install of #GrapheneOS on my Pixel 4a and measured the battery impact. The results are shocking!
Both messengers had only one contact: my regular phone.
I used my regular phone to send messages to the Pixel 4a (which was not used for anything else over the course of the experiment).
I always sent the same message via Signal and #XMPP (mixing up which app went first). In total I sent ~32 messages in intervals of 10mins to a few hours.
MollySocket allows getting Signal notifications via UnifiedPush. - GitHub - mollyim/mollysocket: MollySocket allows getting Signal notifications via UnifiedPush.
wo wir gerade beim Thema #UnifiedPush sind. Wenn conversations offline ist - z.b. beim Flugmodus - verschwinden die push Nachrichten. Sie kommen nicht verspätet sondern gar nicht.
Könnte man die nicht zum Zielserver ins dortige MAM schicken? oder wenigstens eine "verpasste Benachrichtigung" Mitteilung?
@radasbona Die angedachte Lösung ist das der "Rewrite Proxy" / "Push Server" (also up.conversations.im) die Nachricht nochmal sendet. Das ist besser weil es auf WebPush-Ebene sowas wie TTL und topics (Deduplizierung) gibt und der Push Server das besser verwalten kann wenn er versteht was da abgeht.
Der Push Server weiß auf Protokollebene auch schon wann der client wieder online ist. Aber das resend Ding ist noch nicht implementiert.
@ApostateEnglishman I'm trying to phrase my statements in a way in which they don't sound like an advert, but I'm developing and Android instant messaging client called #Conversations_im which uses a federated instant messaging network called Jabber or XMPP.
However the problem is that most people don't even think about switching from WA to Signal - and NEVER heard about conversations #XMPP
#Schools could / MUST do more! Instead they are pushing students into digital dependencies. School leaders avoid the hassle to think, they appreciate conformity.
A hard lesson learned: don’t hold people to the standards to which you hold yourself. It’s unfair because:
1. They probably don’t know what the hell your expectations are
2. If they’re not your employees, you have no reason to hold them to most standards (aside from the typical societal ones such as being decent, reasonable human beings)
3. There’s a pretty good chance the standards you hold for yourself are ridiculous, anyway *points at myself*
Random idea I had out of complete boredom, a bot that boosts publicized polls from anywhere it's federated with, of course not DMs or followers only or anything like that. No, I do not know if this is ethical or viable or anything similar, I just came up with it out of boredom and doing a search for "poll" in my mastodon client. Lol
Not at all. Here's in fact, the transcript from my last NVDA addon coding session, because if it helps others, it's worth sharing. chatgpt.com/share/696b6950-763…
Zach Bennoui
in reply to johann • • •