My guess is they can make more money off a subscription, and it's easier to develop and deploy a web application with a subscription model.

Even if they did, they couldn't publicly admit to it. To do so would be admitting copyright violations. The 'new' version does not seem to be sanctioned by any inheritors, and it deliberately capitalises on a name without a defender.

Whew boy -- the reminder of the Gell-Mann amnesia effect is powerful on this one.

I can't imagine it would not be more than a few minutes of downloading the image from archive.org and running it on any VM - I would be a bit surprised if this had any kind of sophisticated VM countermeasures in it.

Meanwhile, I can believe there is some 4 million line code of something (maybe it was written in assembly!) - in any case a horrible unmaintainable mess that nobody at this software company wants to deal with anymore. Rather just rewrite it in some web based shit like half of this site does for a living.

Also on actually reading TFA - it's rife with errors - at no point in history was the DECTalk the only game in town for speech synthesis - let alone 1999! - and the idea that it was the only one with controllable pitch is hogwash as anyone with a early 90s Mac can attest.

MacinTalk 3 (what you probably think of when you say early 90s Mac), i.e. voices like Fred, is technology closely related to DECTalk, and in fact the audio sample in the article sounds eerily close to a Fred-turned-orthodox.

But the article's description of how Siri synthesizes speech is grossly inaccurate.

Well, sure it's related (and MacinTalk is still builtin to modern macOS and iOS - you can use them for the system TTS voice). But DECTalk and MacinTalk weren't the only speech synthesizers available in 1999 by a long shot.

I had to chuckle about the genius of being able to program C and Lua as if typing English - I mean that's pretty much any reasonably proficient dev.