The XMPP Standards Foundation (XSF) is exited to announce the 28th XMPP Summit taking place in Brussels, Belgium next year - just before FOSDEM 2026. The XSF invites everyone interested in development of the XMPP protocol to attend, and discuss all …
Turn on subtitles, lyrics are displayed.[English,Indonesian,Tagalog,Portuguese］Album 「one-room survival」DL & STREAMING→https://lnk.to/oneroomsurvivalセラニポージのセ...
this is an incredible analog in more ways than one.
From how it’s mined, to how it was marketed as a wonder material, to how it is used as cheap filler, to how it’s impossible to get rid of safely, to how it’s deeply damaging to people’s health and wellbeing.
It's like we took the worst aspects of environmental harm brought to us by cryptocoins and instead of making fake internet money evaporate, we made people huff it like leaded gas fumes in an enclosed garage.
I predict that in a generation or two, if science as we know it is still being done then, researchers will find a link between AI use and early dementia.
Eloquence “Sandy/Bobby” also talk differently: micro pitch movement, timing, and energy. You can get 80% of the timbre without it, but the last 20% “personhood” often comes from prosody. We will never get that 20%. I'm crushed. This project will always be, "it's missing that last mile."
Apparently, we are supposed to get between five and 8 inches of snow here in Central North Carolina over the weekend. Problem is, I have places to be and things to do, and this place shuts down at the slightest provocation. It's only just now kind of sort of opening up again after the last storm. Never mind, I guess.
Edit: Got the answer I wanted and made my decision. Thanks.
Question for #blind people who use refreshable #Braille displays routinely: Does anyone run brltty on Windows? NVDA has the option to use brltty, but NVDA's default plug-and-play drivers don't use it. JAWS doesn't use brltty. Narrator does. So does anyone actually use Narrator's Braille feature, or otherwise use brltty on Windows?
Trying to gauge the usefulness and reliability of brltty on Windows specifically.
I used NVDA’s BRLTTY support a few times for testing my BRLAPI server for the Monarch before the Monarch got a Braille terminal, but otherwise I always use NVDA’s native Braille display drivers. I did test out BRLTTY on the terminal as well, but I found that it did not support the new Windows terminal, or at least the only console window displayed was the window BRLTTY was running in. That could have been a configuration issue on my end though. Also NVDA’s BRLTTY driver has very limited support for Braille commands.
Sigh. When you change the sample rate, the driver recreates the SpeechPlayer instance (because it needs a new sample rate), but it doesn't reapply the voicing tone to the new instance. Fixing it next update. Sorry y'all. This tilt thing took forever to solve so I rushed it a bit.
@clv1 hmm. most IPA can be left unquoted in YAML, and quoting is mostly about avoiding YAML’s "special meanings." You should quote an IPA string if any of these are true: It contains # (YAML comment marker) e.g. a# must be quoted or everything after # is treated as a comment. It contains : (colon + space.) YAML treats that as a key/value separator. It starts with characters YAML uses for structure, like -, ?, :, [, ], {, }, &, *, !, |, > (This one depends on context, but quoting avoids surprises.)
I can change these to all be quoted though in that file for consistency, probably not a bad idea. Sometimes I've hand-edited and maybe left out a quote on some of them so it's good to make that the same.
Thanks for the explanation. The thing is that the phoneme editor leaves many of them unquoted. I found it by comparing pt-br.yaml in my current installation, which I've edited, versus pt-br.yaml in your last add-on package.
What is your favorite tilt setting per voice? Let me know, and we can tweak it to be a default, if people think "adam" is too boomy, we can now change this. Will consider replies by end of day.
I learned recently that Didier Spaier, maintainer of the blind-friendly Slint distribution based on Slackware, passed away. May he rest in peace, and may his loved ones find comfort in the memory of all the good he did for the B/VI community.
The speechPlayer.dll now exports additional functions for real-time control of the voice source characteristics. These allow callers to adjust the "voicing tone" — the spectral shape and character of the glottal pulse — without modifying individual frames. • speechPlayer_setVoicingTone(handle, VoicingTone*) — Apply a new voicing tone configuration. • speechPlayer_getVoicingTone(handle, VoicingTone*) — Retrieve the current voicing tone settings. • speechPlayer_hasVoicingToneSupport(handle) — Check if the DLL supports voicing tone control (for backward compatibility). VoicingTone struct: voicingPeakPos double 0.85–0.95 Glottal pulse peak position. Lower values create a more pressed/tense voice quality; higher values voicedPreEmphA double 0.0–0.97 Pre-emphasis filter coefficient. Higher values boost high frequencies before formant filtering. voicedPreEmphMix double 0.0–1.0 Mix between original and pre-emphasized signal. 0.0 = no pre-emphasis, 1.0 = full pre-emphasis. highShelfGainDb -12 to +12 High-shelf EQ gain in dB. Positive values brighten the output; negative values darken it. highShelfFcHz double 500–8000 High-shelf corner frequency in Hz. Controls where the shelf boost/cut begins. highShelfQ double 0.3–2.0 High-shelf Q factor. Higher values create a more resonant shelf transition. voicedTiltDbPerOct double -24 to +24 Spectral tilt in dB/octave. Negative values create a brighter sound (less natural roll-off); positive values create a darker, more muffled sound.
We just tried out our collaborative text-editor Reflection on a smartphone using @postmarketOS and are a bit blown away by how cool that was. All it took was to install it via the Flathub store.
Is software ever done? When I started writing this I assumed the answer was yes, but by the time I finished, I'm not sure anything is ever done (not just software)
I posted a graph on LinkedIn. It showed that of the 10 million open source projects tracked by ecosyste.ms, more than half haven’t been updated in two years.
NVSpeech Player with phoneme editor and NVDA Addon version 1.8.5 Update from this link, or the updater in the add-on. I must run to work stuff. No more speechplayer until the evening. This release adds proper tilt control to the NVDA driver. No changes to phoneme editor, no time before work. Only changes to the speech player exports, and the driver, and packs to improve the R sound. github.com/tgeczy/NVSpeechPlay…
This is the first time I'm using this add-on with NVDA as one of its voices. I like the 44100 Hz sample rate! However, at least in this release I'm noticing that with the US English default voice, the k sound actually sounds like the t sound, either individually or as part of words. For instance, comma sounds like taama, and combo sounds like taambo. Any chance of improving this?
@amir yeah that's interesting, it's almost like the edge of the "K" gets too sharpened into a "T" at that high of a sample rate. It sounds more fine at 16000 or 22050 even, but it sort of, over-emphasizes certain consonats a bit more turning them into other letters.
Maybe I'm mad or something - LOL, but it also sounds like someone who suffers from a disease called Cleft Lip and Palate. I sometimes have difficulty understanding what it says.
@amir I think I can add a frication lowpass only when SR is high, that'll definitely help itabove 16000. I think it sounds best at either 11025 or 16000, then things just lose shape beyond it, but the wave generator isn't tuning that for different sample rates yet.
Absolutely right. Looking forward to new releases. I tried 44100 mostly because unlike, say, Eloquence, the voices don't sound full. I was thinking that I could get that lacking so-called fullness with higher sample rates.
@amir I know, me too. I'm in the same boat as you. Then I realized how shit formant synthesizers are at spreading out the band equally in voices when they sample higher. That's why most of them like Eloquence actually keep the synth's internal sample rate at 11025, then just upsample that instead, rather than it outputting at that raw sample rate. It's a little nuance.
@amir I could have done the upsample trick with Audioop in the driver, it might sound a bit cleaner but also create aliasing (unless we upsampled the 16000 hZ version to it, maybe slightly better.) So in true reality with these formant synths upsampling is always cleaner, but telling the engine to sample at a frequency results in more burstyness and voice character becomes more "whispery" as well.
Yo, so I've finally checked this out, and wanna say great work on this man. It really has potential, happy to see there is also german in the languages, I'm still trying to think of good words to describe whhat I'd love to be improved with the voices, but let me try around with it a bit more first. Well done, don't burn yourself out or down talk your work. If you need a break, take it, noone wwill judge you. But nice progress.
"Hello, We're Firefox, The Only Browser That Hasn't Hit Itself In The Dick With A Hammer. For years now, folks use us because of our un-hammered dick. Now, you may be wondering why today we've brought this hammer and pulled out our dick.
Mark Surman, president of Mozilla Foundation, speaks at The Wall Street Journal’s Future of Everything Festival in New York City, U.S., May 22, 2024. Andrew Kelly | Reuters Tech Mozilla is building…
The XMPP Standards Foundation (XSF) is exited to announce the 28th XMPP Summit taking place in Brussels, Belgium next year - just before FOSDEM 2026. The XSF invites everyone interested in development of the XMPP protocol to attend, and discuss all …
Wow! This powerful new Bruce Springsteen song just released all about the horrific events in Minneapolis, definitely worth listening to. It already has over 1.7 million views and it's only been up for 12 hours! Wow! youtu.be/wWKSoxG1K7w?si=o_EWrO…
Lyrics:Through the winter’s ice and coldDown Nicollet AvenueA city aflame fought fire and ice‘Neath an occupier’s bootsKing Trump’s private army from the DHS...
Next up will be a VPN ban. Many tech-y people will see that and think “lol yeah well that’s not going to stop *me* from using a VPN”
A VPN ban isn’t really meant stop you from using one. It means when they catch you doing so, they’ll use the fact you’re using this harmless technology itself as a *pretense* to lock you up without needing to do any “hard work” (i.e. an investigator’s job) like actually confirming whether you committed a real crime.
Don’t think you won’t be impacted just because you know how to outsmart an ISP filter! This is not a plan to protect children or stop you from consuming adult media. It is a ploy to eventually eliminate ALL freedom of expression and free access to information in the UK.
And the same goes for Chat Control and encrypted messengers, btw
Many #UK Users Soon Won't Be Able to Access #Pornhub
https://www.404media.co/uk-pornhub-blocked-age-verification-vpn/
#nsfw #porn #privacy #OnlineSafetyAct #AgeVerification #IdentityVerification
I think some are still missing the point. I’m not saying they’re going to lock up every single person who uses a VPN or watches porn.
Your government’s strategy is to make “crimes” like this as common and easy to commit as possible.
This way, if you ever do something completely legal which they *actually* don’t like, *which can be completely unrelated to VPNs or bypassing age verification,* they can pull up your internet history and easily prosecute you.
Organize a legitimate protest, and the government’s swift response will be to label you a child sex offender because they found you using evil VPN technology.
This is how free speech dies, not from a law that says “free speech is dead now,” but from draconian, authoritarian laws like this that give the government “plausible” reasons to go after anyone, which can’t easily be fought against.
This has always been the playbook to target innocents. In the USA, drug possession laws being created as a pretense for cops to raid and dismantle minority communities, for example. And now this is coming to the digital realm.
Európsky parlament schválil kľúčové rozhodnutie, ktoré umožňuje posilnenú spoluprácu medzi 24 členskými štátmi EÚ pri vytvorení finančného nástroja, a teda pôžičky pre Ukrajinu vo výške až 90 miliárd eur na roky 2026-2027. Do tejto iniciatívy sa Slovensko nepripojilo.
Už v predchádzajúci deň sa Európsky parlament dohodol na použití naliehavého legislatívneho postupu (urgent procedure) pri balíku opatrení týkajúcich sa finančnej pomoci Ukrajine. Tento postup má výrazne urýchliť schvaľovanie pôžičky, ktorú Ukrajina potrebuje na krytie rozpočtových potrieb aj obranných výdavkov. Posilnená spolupráca je mechanizmus zakotvený v zmluvách EÚ, ktorý umožňuje skupine členských štátov postupovať spoločne v prípade, že sa nepodarí dosiahnuť jednomyseľnú dohodu všetkých 27 krajín. Základným princípom tohto nástroja je, že finančnú zodpovednosť nesú len zapojené štáty.
V prípade tejto pôžičky sa ukázalo, že Česko, Maďarsko a Slovensko nesúhlasia so zapojením do finančného nástroja. Aby nedošlo k zablokovaniu celej iniciatívy, zvyšných 24 členských štátov sa rozhodlo využiť práve mechanizmus posilnenej spolupráce.
Celková suma až 90 miliárd eur je rozdelená podľa účelu použitia. Približne 30 miliárd eur má smerovať na rozpočtovú podporu, ktorá má stabilizovať verejné financie Ukrajiny a pomôcť pokryť rozpočtový deficit. Zvyšných približne 60 miliárd eur je určených na posilnenie obranných kapacít Ukrajiny. Financie majú umožniť rozsiahle investície do obranného priemyslu a jeho postupnú integráciu do európskej obrannej priemyselnej základne. Kľúčovou novinkou tejto pôžičky je spôsob jej splácania. Ide o tzv. limited recourse loan, teda pôžičku, ktorú má Ukrajina splácať až v momente, keď získa od Ruska reparácie za škody spôsobené vojnou.
Amazon have reported "hundreds of thousands" of pictures of child sexual abuse material found in shared AI training data... but is refusing to tell regulators which data sets.
If you're using generative AI tools, there's a pretty good chance you're generating imagery with child porn training data behind the scenes. bloomberg.com/news/features/20…
The XMPP Standards Foundation (XSF) is exited to announce the 28th XMPP Summit taking place in Brussels, Belgium next year - just before FOSDEM 2026. The XSF invites everyone interested in development of the XMPP protocol to attend, and discuss all …
Rachel Gilmore has found 9 Canadian companies profiting off ICE contracts, -- all but one of whom also have contracts with the Canadian government ... so our taxes at work for ICE. #cdnpoli
For those who decline to visit substack: Garda, Hootsuite, JSI Telecom, Roshel, Thomson Reuters, Jim Pattison, Magnet Forensics, Icor Technology, and CGI Inc. Only Roshel doesn't have a Canada gov contract. Note ATCO Frontec is not one of them. rachelgilmore.substack.com/p/a…
Yes, GardaWorld is a Canadian company based in Montreal. Chief of Staff to PM Brian Mulroney, special advisor to PM Harper, and Canada's Ambassador to the US in the 90s, Derek Burney was CEO of CAE and Chairman of GardaWorld's International Advisory Board from 2008 to 2021.
Going out after work today. Haven’t been outside since the massive dumping of snow on Sunday. Really looking forward to getting out of the house for the night. Apparently our sidewalks still haven’t been cleared so we’ll have to practice my snow climbing skills from living in Winnipeg and Colorado.
Seeing some company-internal post about AI trainings and AI strategy kickoff makes me want to reply with “I really couldn’t care any less about this. I hope it goes nowhere. I can’t wait for the AI bubble to burst.” I don't dare, so instead I'll share some thoughts here.
My much more negative stance makes me not want to interact with any AI, not so much because it has its flaws, but because the whole ideology and practical effects are fundamentally bad. For once, I don’t want my freedom as a programmer to be enclosed by Microsoft and Google, becoming completely dependent on subscriptions to their tools. That’s a direction they’re going for. I think we need to actively fight this and not just wait for it to fail in some way.
The XMPP Standards Foundation (XSF) is exited to announce the 28th XMPP Summit taking place in Brussels, Belgium next year - just before FOSDEM 2026. The XSF invites everyone interested in development of the XMPP protocol to attend, and discuss all …
Indonesia’s Communications and Digital Ministry opened a job vacancy and invited the public to apply by dropping their resumes and other required documents into a publicly accessible Google Drive. Turns out you can access and download all of the documents submitted by everyone who applied.
Someone pointed it out in an Instagram video which went viral and soon after he’s had multiple and sustained phishing and hacking attempts of his social media and email accounts.
A na základe čoho môže nezainteresovaný človek vedieť, či niekto reálne má alebo nemá ADHD (prípadne hocičo iné)? Skôr by som povedal, že je v móde spochybňovať diagnózy iných ľudí.
Určite nejaké obmedzenia pri ADHD či iných diagnózach sú, veď sám mám isté obmedzenia kvôli môjmu autizmu. Treba ale rozlišovať, či to je pre daného človeka naozaj obmedzenie alebo sa iba skrýva za diagnózu. To však zainteresovaná osoba iba z pozorovania nemôže veľmi vedieť, pretože nikto nevie čo ten človek prežíva, okrem neho samotného.
Každopádne si myslím, že ľudia by mali byť najprv tolerantní a chápaví, a nie hneď na prvú zľahčovať/spochybňovať diagnózu (a tým nemyslím nikoho konkrétneho, píšem vo všeobecnosti, keďže aj pri autizme ľudia veľmi radi hovoria vetu "nevyzeráš ako autista").
@Ralfeek A s tou "neschopnosťou vychovávať deti>deti majú ADHD" by som bol opatrný. Veľa autistických detí vyzerá ako keby boli nevychované, rovnako to je určite aj pri ADHD, ale o nevychovanosť tam vôbec nejde.
Je úplne normálne a prirodzené, že takýchto diagnóz pribúda čoraz viac, keďže v minulosti sa to neriešilo, kvôli čomu to teraz vyzerá, že je toho veľa, keď sa ľudia v dospelosti diagnostikujú, a taktiež sa ľudia množia, a keďže ADHD a aj autizmus je dedičný, tak je tiež logické, že ľudí s takýmito diagnózami bude každý rok viac a viac.
Určite to ale nie je móda, a ak to má niekto oficiálne diagnostikované, tak by som to určite nespochybňoval, pretože takáto diagnóza sa nedá získať len tak, a neviem kto normálny by vyhodil peniaze na diagnostiku iba preto, aby sa potom na to mohol vyhovárať.
Man, you're doing some incredible work in so many ways in so short a time. Honestly, I believe you *will* make an Eloquence replacement, the challenge is going to be reverse engineering the best phonimizer ever made.
• Negative tilt (bright) → fades toward derivative (has the highs to boost) • Zero/positive tilt (warm/dark) → stays on flow (warm body) Cleaner gain structure: • srcDeriv = dFlow * radiationDerivGain (derivative gets gain) • Flow stays as-is (no extra gain needed) Simpler mixing: Just (1.0 - radiationMix) * flow + (radiationMix * srcDeriv) This makes more sense because: • At Tilt 0: Pure flow = warm, full body ✓ • At Tilt -10 (bright): Mix toward derivative, then high-shelf boost = sparkly highs ✓ • At Tilt +10 (dark): Pure flow + low-pass = warm darkening without mud ✓
Already in Brussels for FOSDEM? Or just around? Join us today for OSGeo.be’s New Year drink at **Cercle des Voyageurs**—kickoff at 18:30! Everyone is welcome to celebrate open geo, new projects, and good company. Let’s toast to 2026 together! 🥂
“Once men turned their thinking over to machines in the hope that this would set them free. But that only permitted other men with machines to enslave them.”
shkspr.mobi/blog/2026/01/are-t… One of the dreams of Web 2.0 was that website would speak unto website. An "Application Programming Interface" (API) would give programmatic access to structured data, allowing services to seamlessly integrate content from each other. Users would be able to quickly grab data from multiple sources and use them for their own purposes. No registration or API keys, no tedious EULAs or meetings. Just pure synergy!
Something something … enshittification … blah blah … zero interest rate phenomenon … yadda yadda our incredible journey …
But back in the land of rationality, I've had a lots of experiences running APIs and helping people who run them. The closure and lockdown of APIs usually comes down to one or more of the following.
APIs cost money to run. Yes, even the static ones have a non-zero cost. That's fine if you're prepared to endless subsidise them - but it is hard to justify if there's no return on investment. Anyway, who is using all this bandwidth? Which leads on to:
Lack of analytics. Yes, I know tracking is the devil, but it is hard to build a service if you don't know who is using it. Sure, you can see traffic, but you can't tell if it is useful to the end consumer, or what value you can share. There's no way to communicate with an anonymous consumer. Which, of course, takes us to the next barrier:
Communication is key. If you need to change your API, there's no way to tell users that a change is coming. That might be the announcement of a deprecation, an outage, or an enhancement. You can try smuggling error messages into your responses and hoping someone notices a failing service somewhere - but it's much easier to email everyone who has an API key. And you know what else keys are good for?
Stopping abuse. It'd be nice if everyone played nice online; but some people are raging arseholes. Being able to throttle bad actors (figuratively or literally) is a desirable feature. On a resource constrained service, you sometimes have to put rules in place.
Still, if you know of any good open APIs which don't require registration, and that you think will survive until 2036, please drop a link in the comments. #api #coding
SuspiciousDuck
in reply to Schrödiɲger's Enby • • •