Já se ho zatím držím i kvůli tomu, že ho můj tým podporuje ve Fedoře a RHELu. Dokud ty věci půjdou vypnout, jsem s tím ještě OK. Ale mám v záloze připravené Vivaldi. UX mi nepřijde tak dobré jako Firefox, ale hodnotami máme shodu prakticky 100 %. Duch staré dobré Opery. Možná je načase se vrátit domů. Operu jsem používal mezi lety 2004 a 2014.
Seit einem Update ist es unmöglich, unter Windows mit Screenreader in PDF-Dateien Formulare auszufüllen. Wie ich vermute, wird die Behebung des Problems Monate dauern, weil es ja nur um so'n paar Blinde geht.
For some reason, and I can't possibly think why this is, I can never remember whether I'm meant to be using `useradd` or `adduser` to create a new user in linux.
I'm kind of just skimming and will do a more in depth covering of this but uh... why does Madam Pomfrey sound like a dude? It sounds like they had one of the producers or coaches read her lines and he raised his voice up higher and they thought, "welp, we can just leave that in there." #HPFullCast
The credits mention a proper name... well, obviously this could have slept through, but multiple times in multiple books? I think its the actual takes they are using.
Long post about the most boring part of the HP audio drama
Sensitive content
The credits for the Harry Potter full cast audio are off the chain. They are 13 minutes long! Now obviously it is a full cast, so you would expect there to be a lot of actors, but this is not most of the credits.
There's the roles you'd expect like casting directors, vocal coaches, recording engineers, dialog editors, music composers and performers, foley editors, mix engineers, runners, and of course the production interns. Then we get into the questionable positions.
First, everyone is a supervisor. There's the continuity supervisor, music supervisors, sound design supervisor, script supervisor, creative edit supervisor (what even is that?) and on and on. But also, the roles are subdivided to the point of madness. You thought dialog editting was just one guy? Hah hah hah. It is a whole team. There are senior dialog editors, dialog editors, and additional dialog editing. The "foley" section starts with the name of a company, then appears to list every person at that company: project producer, creative consultant, sound editors, lead artists, supporting artists, recordists...
Oh, did I make it seem like sound mixing was just a couple people? That is a whole other company with engineers and sound designers. Then there are audio producers, audio post production technicians, assistents, directors, audio proofers, production interns, assistants, coordinators, post production and production directors, then finally executive producers and commissioning editor. Then an additional round of production and post production directors, studio directors, a final round of executive producers, then at long last, JK Rowling.
@miki It does feel that way, yes. "This one recorded the on location audio for the scene where the Dursleys go to the zoo. This guy recorded his gardener snake in the back yard. This lady over here supervised them. These people over there assisted them, by hanging the microphones."
Don't get me wrong, I'm all for everyone involved getting credit, but there are hundreds of names here for not even 9 hours of audio, which is just insane.
Long post about the most boring part of the HP audio drama
Sensitive content
Obviously most of those names will be in all seven of them so it's more like, IDK, 70 hours or whatever, but that just makes me wonder even more how much the production cost.
I'm not usually one to tell people not to use a service, but Discord and Bluesky are both inaccessible due to their CAPTCHA system. If you have a community and you want to make it fully inclusive, please don't use them
Discord's version of HCaptcha has a text only option for the challenge, unless that's changed recently. Still absolutely hate HCaptcha as a company though, they're scummy and do not care about the disability community. Really wish Discord would start using Recaptcha instead, but I understand why some websites can't or won't adopt that.
Not as far as I know, although since it's text a wscreen reader can just read what you need to type in. I know HCaptcha finally added an audio option recently, but not sure if that's implemented everywhere or not. I haven't come across a website using HCaptcha for quite a while now.
@ZBennoui Alright, but then where's the challenge... You know, they probably don't even realize screen readers exist and so haven't thought of it, lol.
Well that text option isn't implemented everywhere, websites have to specifically enable it. You're not typing exactly what's displayed, they like ask you questions that you respond yes or no to. I think it's designed so that an AI model would get confused if presented with the same information, but a human should be able to figure it out pretty easily.
Wednesday I'll be streaming Pigments: "Draw The Rest of the Owl," or How & Why to finish your synth sounds. Subscribe to my YT channel and turn on notifications so you don't miss it, or just watch this space.
https://www.youtube.
Gemini came to my Google Home devices today, and now everything is better! It's all so much better! I don't know how anyone could think it is anything but better! Listen for yourself! #Tech #AI #Gemini #Google
Ok this is insane, how did they not test such a basic feature before deploying this new AI model? Not even sure this is AI related at all, just seems like the device doesn't know what to do with that specific command. Also wouldn't take too much stock in the "five inches" thing, pretty sure that's just a TTS error, although again, testing Google, testing.
@Scott Oh no, there was an attempt before this where I was not calm. So not calm that I tried it a second time and recorded it so I could complain to the whole Internet
@FreakyFwoof For now you can turn the volume up by moving your head closer, and down by burying the bastard thing in pillows. Just make sure you save one to shout into.
Listen to how little of a delay there is between the bloop and the response with that "use a remote" error message, versus in all other scenarios. This is definitely a canned string, the command was clearly understood and executed, something just went wrong somewhere.
@miki The thing that went wrong is I got Gemini. Whether the issue is to do with the LLM or not, it was not ready to ship as the assistant for controlling home devices
@miki Yep, my guess is it got stuck trying to interact with a specific device, I would get this now and again with the assistant on my Nest speakers. Including the name of the device you are trying to change, even if it's the one you're speaking with, usually fixed it
It stands to reason that it would be, considering they've had what, two years to perfect the technology? This is of course not counting the decade plus of time they spent on deep mind research before Gemini came out, making it the pinacle of AI technology. OpenAI? More like OpenBYE! Amirite?
it's truly amazing what LLMs can achieve. we now know it's possible to produce an html5 parsing library with nothing but the full source code of an existing html5 parsing library, all the source code of all other open source libraries ever, a meticulously maintained and extremely comprehensive test suite written by somebody else, 5 different models, a megawatt-hour of energy, a swimming pool full of water, and a month of spare time of an extremely senior engineer
Reading the RSS 2.0 specification and found this: "The purpose of the <textInput> element is something of a mystery. You can use it to specify a search engine box. Or to allow a reader to provide feedback. Most aggregators ignore it."
This, right here, back in 2002, was when the web went wrong. We don't know what it's for or why it exists, and everybody ignores it, but it's part of the standard anyway!
Also, bonus fun: RSS 0.9 claims to be RDF but isn't, RSS 1.0 really is RDF but is incompatible with RSS 0.9, RSS 2.0 is incompatible with all of the above, and Atom is the modern W3C format that is incompatible with everything else and nobody uses. And everyone wonders why normal people don't use RSS.
I learned all this because I wondered why my static website generator wants to produce feed.rss, feed.rdf, and feed.atom.
Hmm, so many people I know have lost their minds trusting the AI and not thinking things through with some critical thinking this week. (Essentially: "The AI said to shove a fork into the electric outlet! I'll try that!" ) #AI #LLMs
Watch the official "Will You Dance?" video: https://youtu.be/flbce8yaBaoThe Bird and the Bee's new album "Recreational Love" is out now! Download it on iTun...
I've never used the Arduino IDE before, but I'm about to do some hardware beta testing that will require regularly updating firmware using it. I'm more excited about this than I probably should be.
ACM is now showing an AI “summary” of a recent paper of mine on the DL instead of the abstract. As an author, I have not granted ACM the right to process my papers in this way, and will not. They should either roll back this (mis)feature or remove my papers from the DL.
It has come to my attention that the artwork for the original mozilla.org "dinosaur" logo is not widely available online. So, here it is.
As I explained in some detail in my 2016 article "They Live and the secret history of the Mozilla logo", I commissioned this artwork from Shepard Fairey to use as the branding of the newly-founded mozilla.org and our open source release of the Netscape source code, which eventually became Firefox. This happened in March 1998... jwz.org/b/yk0j
Meta released their Segment Anything Model for Audio that lets you separate audio with text prompts. It could be guitar, speech, bird sound, etc. #ML #AIai.meta.com/samaudio/
@Scott Ah, that's pretty common practice. They just want to get agreement from users for not to abuse and track who's using it. Even they do that for Llama models. Once you download it you can run HF models without internet.
The repository provides code for running inference with the Meta Segment Anything Audio Model (SAM-Audio), links for downloading the trained model checkpoints, and example notebooks that show how t...
If you're Canadian, and you support this decision, it's worth sending Skate Canada an email saying so. I'm sure they're about to hear from people opposing it, and it would be good to make sure they’re hearing from people who think they did the right thing as well. skatecanada.ca/about/contact-u…
Having a lot of fun getting back into flight simulation. For my Christmas present to myself this year, I bought a honeycomb bravo throttle and switch panel. Lots of switches buttons and levers that you can assign to various things on something like a 737.
#AndroidAppRain today brings you 26 updated and 1 added apps:
* FireWall Blocks: a firewall blocking internet access via Shizuku or VPN 🛡️
At our RB builder today, we had to fight with a load of failed RBs. Half we were able to fix, the other half not (mostly due to build timestamps and "dirty build"). So a call to the devs to please take care, see e.g. our corresponding hints: izzyondroid.org/docs/reproduci…
Need to mute a tab that's playing unwanted audio in Safari on macOS? You don't need to try to locate a button on the website or in Safari's toolbar, just choose either "mute this tab" or "mute other tabs" from the Window menu. You can additionally map keyboard shortcuts to these commands in System Settings > Keyboard, so you can mute tabs with just a keystroke. #TipTuesday
Why downloading #VMWare Workstation is such a pain in the back? Register on Broadcom, then go to some obscure download page, then the link is grayed out… what the hell!
@tspivey I managed to find an answer. You have to *visit* the Terms and Conditions link. Afterwards you check the boxes that become available, and you have the HTTPS Download link.
Finally got that dumb router to use my Pi Hole as DNS. I'm quite interested what sort of stuff I'll see there in the next days, mainly from IOT devices. And wether I have messed anything up/anyone will complain or noone will even notice, that would be perfect.
J'ai besoin d'aide pour trouver du travail rapidement sous peine de me retrouver sans logement.
Je suis un développeur PHP senior avec presque 10 ans d'expérience dans le dev web, je suis aussi pertinent en back qu'en fullstack (html, css, js et ses frameworks front).
Je cherche sur Montpellier ou en 100% remote.
Je passe mon CV en MP si besoin.
A défaut de pouvoir m'aider, un boost serait super utile, merci !
In other depressing IoT news, iRobot, the maker of the popular Roomba vacuums, files for bankruptcy and sells itself to Chinese company. Hello cameras and mics in bajillions of homes.
For a bit of fun, we're releasing a single just before the end of the year. Some of you might recall I posted a sketch of a piece called In My Head back in July. Well, I actually finished it! :) It's unapologetically poppy and electronic, with backing vocals distantly reminiscent of a boy band. 😂 Apple Music: music.apple.com/au/album/in-my… Spotify: open.spotify.com/album/3Bwx5hn… YouTube: youtube.com/watch?v=oq5WWN77dL…
Nice, love the subtle glitchy edits in there, really cool. Slightly unrelated, but what do you do for artwork? I wanna put more stuff out, but I'm totally blind so any photo editing stuff is gonna be kind of difficult. I've got like one single out on streaming, and for that I just used Distro Kid's AI art generator, but knowing what I know now about how those models are trained, I feel kind of icky about using them.
@ZBennoui Thanks! I didn't actually realise Distrokid had an AI art generator. I used Bing Image Generator for this one, plus Aira to help me put text on it, but apparently, the artwork is really weird and creepy, which is precisely what I didn't want. Like you, I don't really want to use AI art generators, but the services force you to do album art. In the past, I've had some sighted help, so one of our albums has an actual photo on it, but sighted help is not always available to me. In short, same struggle here.
Yeah honestly I wouldn't feel so bad if I knew the training data was collected properly and with consent, but at the same time, I'm not really sure what else I'm supposed to do. I'm happy to pay artists to design art, but I'm not making money from this at all so it's not worth it right now. I've thought about just taking pictures of myself standing next to some gear or something, and hoping that looks good, but then I guess getting the right dimensions would also be an issue and I'd have no idea how to even edit the photos.
@ZBennoui Agree on all of that. The temptation to just have a blank photo (if that would even be accepted) or just text with nothing else is immense. Or maybe just braille font to mess with people.
Yeah just having like the song title as braille would be really funny, I feel like people would be confused and click play out of sheer curiosity, could be a good way to get more listeners lol.
@ZBennoui Do you know if DistroPic puts the artist and album title as text in the image (as tends to be the case for album covers) or does it just do the image and you still need sighted help to get the text onto the image? @vick21
@ZBennoui Nope, distrokid never punished me for it, so I assume this is not a requirement. your title/artist info will be displayed along with the picture. I often generate the picture that will represent something about the tune I am posting. Of course, I verify with the sighted person before putting this in production! I guess you could quadruple-check with AI and if the description is consistent enough, that might work!
Scott
in reply to Dgar • • •