Novelty TTS Voices and calling TTS programmatically.

[Linux for blind general discussion]

Okay, so I have two questions that really aren't related except they
both deal with TTS and were born out of the same idea.

The first is whether there exist any, for lack of a better term,
novelty voices. Things like pirate, cowboy, comically exaggerated
foreign or regional accents, voices designed to sound like old school
sci-fi robots instead of just being relics from the early days of
tTTS, etc. Preferably free-to-use and compatible with espeak-ng or
another TTS engine that's readily available for Linux.

The second is whether there's a way to invoke a TTS engine from within
a program. I'm a old hat at writing C++ terminal applications that
read from the keyboard or a text file and write to the screen or a
text file, but what if, instead of just writing output to the screen,
I wanted to invoke a TTS engine and have it speak the output in a
voice of my choosing? Or in other words, how do I make terminal apps
self-voicing? I know the basics of invoking espeak-ng from the command
line, putting a string in quotes to have it read directly, using thee
-v option to set the voice and the -f option to read from a file, and
I know enough bash scripting to do simple conditional logic, but being
able to invoke it from within a C++ program would offer greater
flexibility, and while I know the system function can be used to
execute external programs, I've always gotten the impression that
should be a last resort).

What sparked these questions was the idea of a talking pirate skull
that speaks the roll of a d20 and adds an appropriate quip and
thinking about how to make it more flexible without having to record
voice clips for every spoken phrase... though I confess to not knowing
how to make my programs play audio files either.