Does anyone know a way to make new Text-to-speech voices? I imagine there is NO utility to make them or hack into the existent ones. Am I wrong?

If you listen carefully to the different voices available with TTS, you'll notice that not only the voice itself changes, but also the way it reads text. (try for instance with an all cap word, some voices will spell the word while some others will read it. Some voices read abbreviations while others try to pronounce the abbreviations as they're written, and so on) And voices are associated to a specific language. So voice files aren't only a set of parameters on the tone of the voice.

This leads me to believe that TTS voices aren't only data files, but also "executable" files, which contain instruction, and thus those instructions must be coded in a way or another (and concerning speech synthesis, it must be the hard way ;�)

So according to me there can't be a way to create your own voice other than using Apple's TTS SDK and coding in C.

Arnauld

Sorry for the delay in response.

Is there a Text-to-Speech SDK? How can I put my hands on it?

Apple's SDKs can be found at:
http://developer.apple.com/sdk/

You might want to wait until Apple releases the next TTS and SR SDK 2.0, so you could start writting new voices...

It might be ready by the time Apple releases Sonata. (System 8.7 or 9.0.)

Good luck...

[img]about:qtlogo[/img]

------------------

[jwild]

I hear from the mac guy who works with me that mac will be adding new vioces, but i personnally have not see any thing about it.

[This message has been edited by jwild (edited 06-25-99).]

I know it is possible to at least modify the voices because Microsoft Word 98 lets you do it. Go to the Tools menu and select "Pick Voice..." It lets you modify the rate, pitch, and modulation of any installed voice. Unfortunately, you can't export them from Word as far as I know.

BTW, changing the modulation has a major effect on how text is spoken. Perhaps that accounts for the variation you noticed among the different voices?

--argonaut

[KBerg]

The Mac Text-To-Speech system has three different engines, and used to be for computers of different power levels. The pro engine sounded the nicest (Voices like Agnes, Victoria and Bruce, as well as the High quality versions of those voices) the II engine, which has lower memory requirements and a broader variety of voices and finally the I engine which was for '020 machines. Each of these engines pronounces certain things differently, and as far as I can tell pronunciation is consistent across all voices that use a particular engine.

Hope this helps somewhat.
Karl.