Text To Speech

[notjustjay]

One of the cool things about the Mac OS is that it has come with text-to-speech technology built right into the operating system, since System 7 in 1993.

The problem is, it hasn't changed a bit since then. The voices sound exactly the same now as they used to sound coming out of an LC III in my old high school.

There are a few companies that make some AMAZING sounding speech synthesis systems, like AT&T and Neospeech. On Windows, there are a lot of programs that default to using the Microsoft speech agent, but you can buy plugin packs from these companies to enhance the voices.

Any ideas if Apple ever plans to pursue this? I would LOVE to be able to plug Neospeech into my Mac, or any equivalent "next-generation" speech synthesis package.

Because, c'mon Apple, things have progressed in the past 12 years.

 

[RacerX]

One of the cool things about the Mac OS is that it has come with text-to-speech technology built right into the operating system, since System 7 in 1993.

The problem is, it hasn't changed a bit since then...

Because, c'mon Apple, things have progressed in the past 12 years.

Apple Text to Speech technology has change quite a bit since then.

First is that it is system wide, usable in every application (via a key combination set in the Speech preferences). In System 7 through 9.2.2, only applications that built it in could use it or you needed a third party app to add the functionality to apps.

Second, with Mac OS X it gain a major increase in vocabulary/pronunciation over previous versions. In the old Mac OS it would try to sound out words it didn't know, and there were a lot of them. In Mac OS X it can deal with most words without difficulty.

In fact, in Mac OS X it'll say "Mac OS 10" where it reads Mac OS X.

Since I have three Macs running Mac OS 8.x and two running Mac OS X, I get to here the differences almost daily. Things have progressed, even if you haven't noticed.

 

[notjustjay]

Cool. I didn't notice them. Still, those relate to operating system code (integrating it with the OS) and a mapping between a dictionary and the phonemes... I'd still like to see a new speech engine based (like AT&T and Neospeech) on real samples, for example. The downside is that these technologies require a lot of processing power and 500-600 megs of hard drive space for the sample database. Still, there should at least be some kind of option to download an advanced speech pack.

Here's hoping. In the meantime, my household caller-ID-annnounce system is going back to a Windows machine, where I can use Neospeech in a $30 add-on pack to the application I'm using (Homeseer).