As is too often the case, it doesn't follow what I consider to be a fundamental OCR rule: the input document type should also be a possible output document type. So, PDF in should mean that composited PDF output is possible. This is necessary for groundtruthing.
Might be but aren't. They're inevitably someone I've never heard of from no recognizable organization. If they have credentials, they are invisible to me.
I have permanent prompts in Gemini settings to tell it to never include videos in its answers. Never ever for any reason. Yet of course it always does. Even if I trusted any of the video authors or material - and I don't know them so how can I trust them? - I still don't watch a video that could be text I could read in one-tenth of the time. Text is superior to video 99% of the time in my experience.
FWIW, for recently released books (<5 years) with more than a few hundred reviews, I find a 4.2 or above rating on Goodreads to be a fairly accurate indicator of whether a book Wil be a pleasurable read. The types of books mentioned in TFA have classic saddle curves - high on both ends, low in the middle. These controversial books may offer high engagement in a classroom or book club environment but for my time and money, the wide variety of possible negatives that cause the low-end of the saddle are sufficient to remove a book from my consideration no matter how high the other end of the saddle. Time is short, books are many, and although I am a very fast reader, I'm going to focus on reading writing that hits more positives rather than one that is split. The ends of a saddle-shaped vote curve can be discarded leaving, still, a middling rating.
Generally, this kind of writing, AI-aided or not, is characteristic of beginner writers. It typically leads to lack of story arc, unfocused characters, and writerly over-investment in the product and its reception rather than in the story and its effective fulfillment. It's also just draft one. The best and most successful novels are substantial redrafts and rewrites. I encourage you to drop AI, get a book on writing (perhaps Stephen King's), and now begin again, this time only using the parts of the story that work and never reusing something from the first draft because you like the way you worded it.
I use the third-party app Harmonic to browse HN and it has a dark mode. It's quite nice, actually. Though I do think your tone is unusually inappropriate for a feature request.
I used it and it was good. The thing about the compression for me was that because it wasn't for broadcast or sharing it didn't have to be as rigorous or detailed. Plus, it's a desktop. Stuff doesn't change as often as often as in a movie.
Intrigued. Still looking, though for robust multilingual end-to-end PDF to PDF archival OCR for consumer hardware that can replace OCRmyPDF without excessive customization for each run.
After trying a dozen or more paid and open source, I keep going back to MacWhisper. The dictation feature is in advanced beta but works well. The only thing I want it doesn't have is to have different models chosen for different tasks at the same time: one model for each drop folder, a different one for dictation and then another general-purpose one for drag-and-drop. I have the memory for it and MacWhisper can flush a model after a certain amount of unuse time anyway.
"I was surprised to learn that words with multiple definitions are defined in chronological order of first usage, and not, as I imagined, some kind of “importance” of definitions."
It depends on the type of dictionary. Merriam-Webster Unabridged (and most unabridged and historical dictionaries; you can tell a historical dictionary by the citations/quotations) do it chronologically. Merriam-Webster Collegiate (and most daily use mainstream dictionaries) sort senses (meanings) by likelihood of use frequency, or importance. As you can see, saying "Merriam-Webster" and meaning one dictionary is incorrect, as they produce more than one dictionary. Source: am lexicographer, have worked with Kory Stamper.
Might be a bot looking for a submit form to abuse that will post content to a forum. The HTML or CSS ID on an email field may look the same to a bot no matter the form. So it will submit what matches its goals even though the text body content field does not match.
Seems premature. My scanner software, SnapScan, still regularly updated, requires Rosetta. Abbyy FineReaser, the best Mac OCR, requires Rosetta. Although they may be related, as the SnaScan software does OCR with the FineReader engine.