#orca did the trick. The text was (as guessed by @benjaminpaikjones) being read as s, ave.
There was no distinction between kbd and paragraph, though.
Testing #ChromeVox on #Windows might be the next to-do.
Alright. #ChromeVox on Windows works out of the box. Using the kbd element actually broke it! A list with 4 sub items was being read as a list with 2 items, because the first letter in the third sub-item was a kbd.
Then, while reading text, there was no announce whatsoever about the `o` key being a keyboard input. It would pause, read o, announce "list item" to say now the list item was being read again in continuation before taking the pause for `o`.
To me, it sounds like this is a waste. I'm definitely gonna have to just spell it out in the blog post instead of hoping semantic HTML does the trick.
@benjaminpaikjones @celia +1. More specifically, how does a screen reader read that? (Pretty sure it does read kbd elements.)
@dubiousdisc @benjaminpaikjones Yeah, that's what I was hoping someone would know. 😅
@celia @dubiousdisc I'm guessing it says something like "O Pen", which isn't totally incomprehensible 😂
But the only way to know would be to test it out with one.
@celia Yes, even <em>, <strong> and <s> are usually not announced by screen readers. You can (and should) use semantic HTML, but you shouldn't rely on it being announced.
https://www.tempertemper.net/blog/be-careful-with-strikethrough
Not a good look for Google (to be fair, when is it ever...)
I installed #ChromeVox on #Chromium and there are no voices available out of the box.
To get voices, apparently you need to compile the browser yourself? https://chromium.googlesource.com/chromium/src/+/lkgr/docs/accessibility/chromevox_on_desktop_linux.md