>I was on a train station the other day and noticed the announcement referred to in the initial posting. Essentially the announcement appears to be made by stringing together words and phrases recorded by a real person (I don't think it is an entirely simulated voice).

Speech synthesis is done by concatenating phonemes one after another. I read somewhere that if phonemes are blindly concatenated, the resulting sound will not be natural, so, some modifications are required at the end of one phoneme and the beginning of the following one.

I think that railway announcements etc. will have the combinations of different variable values recorded and depending on the requirement, one of them will be used with the standard announcement. Otherwise, as <jmh> said, one would hear the standard parts of the announcement in one voice and the rest (time, platform no., etc.) in another.

Regards,
Manoj

Bangalore India
12°58' N, 77°39' E

http://www.geocities.com/kummini/

Bangalore India
12°58' N, 77°39' E

http://www.geocities.com/kummini/index.html