To generate speech locally on a pi for some languages you could have a look at pico2wave (disclaimer: im the maintainer of that node). Just follow the instructions to install the libraries and install the node from the palette.
Do you have a speaker connected to your pi? To play the generated audio set the pico2wave node to output to a file than connect an exec node to it with append msg.payload selected and aplay as the command.
I then git cloned the repository so that I had a new folder
and then cd to pico and ran the compile instructions from there
Unfortunately, the last of the instructions on the GitHub site say make install
but this didn't work for me
neither did sudo make install
(The sudo one appears to do the job but I still got an illegal instruction error if i ran pico2wave)
However, there is a working pico2wave in /home/pi/picotts/pico
At this point, my Linux skills ran out so to make the node use the new compiled version I manually edited /home/pi/.node-red/node-modules/node-red-contrib-pico2wave/pico2wave.js
and replaced line 65 node.pico2waveCommand = "pico2wave -w " + node.filePath + " -l " + node.language;