iOS 17 speech features.
"Accessibility" has always been an example of Apple's tendency to choose words carefully and based on their positive, generic meaning. Not "features for the disabled," for example. This isn't spin: it's appropriate because they often aren't features to help people with particular challenges, but features to help everyone at different times. I've used the Mac's system-wide zoom feature, perhaps useful for the far-sighted, regularly for decades, just to quickly inspect text, details, or pixels.
So, many "accessibility" features are just cool, useful features. There are lots of them, and over the years, they've always felt smoothly integrated and not "grafted on."
The simplified iPad app interfaces ("Assistive Access") certainly seem cool, and a reminder of iOS's deceptively complex maturity, calling for this looping-back to something like "iOS 1.0" again. But the coolest-sounding one of all, in both senses, has to be a pair of features called "Personal Voice" and "Live Speech", where you supposedly read 15 minutes of text for an automatically-generated synthetic version of your voice, which can then recite typed text.
It seems similar to Acapela's "My-own-voice" product, though it looks like that one requires some validation by a human staffperson once the recordings are submitted. You can sign up for that service, send in your vocal samples, hear the result, and then buy the voice for your use in the future. Acapela's system voices have been pretty close to state-of-the-art for well over a decade, so I'm guessing they're still a contender in this space, and I'm interested to hear a comparison between Acapela's and Apple's versions of a voice. No doubt these will not be the only two options.
"Live Speech," as Apple calls the text-to-speech half of their version, seems like a system-level feature that's surfaced during things like FaceTime calls, and works alongside the ability to pre-select common phrases to "speak" without having to type them. Stephen Hawking's chair-mounted conversational method comes to mind, of course, and the world came to know his robotic "voice" as something that marked world-class software sophistication. I can only guess at the expense and effort that went into that. This decade, something far more sophisticated will be available to billions. No question I'll be trying it.