They’re Coming For Librivox

This is something that happened a while ago, and let's go for the most generous interpretation: Microsoft decided to take a small fraction of their huge piles of money and do something good with it by letting Project Gutenberg, the free ebook people who've been going since the 70s(!), make free audiobooks using some of their fancier AI text-to-speech stuff. Obviously, if you look at audiobooks through the prism of accessibility it is a lot better to have a bunch of free ones of classics, even if there is still the vibe that you're listening to Microsoft Sam's great-great-grandchild. It might be a bit weird that it reads out the pages with all the publisher credits and stuff, like the Garth Marenghi audiobook did for a joke, but presumably they can trim that or something. The thing is, though, there's already an audiobook service for Gutenberg books. It's called LibriVox and it's one of the weirdest things on the internet.

Audiobooks—the normal kind you would've got on tapes in the library when you were a kid, but which are now available largely from the Amazon cabal—can be uneven sometimes. I remember my brother and I listening to the Song of Ice and Fire books as audiobooks when we were teenagers. Because they're read by Roy Dotrice, who was already in his 70s when the first book was published, if you listen to them one after another you notice that he forgets which voices or accents he's given to characters between books—or, as the books get longer, within them . This is not a dig on Big Roy, as goodness knows I couldn't hold all that stuff in my head either; he was 88 for A Dance With Dragons, it was 50 hours long and had probably >200 characters. Most books with reasonable page counts, though, are a bit more consistent. There are often little quirks—almost any book where a male narrator has to voice female characters will see them busting out some pretty egregious "girl voice"—but it's mostly pretty radio-professional.

LibriVox, by contrast, is absolutely bonkers. There are often different people doing every chapter, so you get wild variation in mic quality, room noise, volume, voice, pronunciation, accent, pretty much everything. It's not professional, and frankly it's sometimes not even pleasant—try listening to, e.g. the first few sections of Capital, where the first section switches between three men who seem to be competing with each other to see how much they can sound like they're recording from the inside of a bin. But I dunno, I do like the fact that you can tell that a real actual person was involved at some stage of the process? Maybe this is just me being charmed by their haphazard nature; but a frequent Memhaz topic is bemoaning the slow removal of humanity from all kind of things. One definition of professionalism is doing stuff less like a person would and more like a computer would. Now that computers can do more and more stuff, I find the sheen of professionalism appeals to me less and less in a lot of things.