This simply uses the eSpeak cli program to generate wav audio data from text strings. You can then use the raw binary data to output to a file, send as a http download, get as a base64-encoded data URI string, or whatever else.
$ npm install espeak
var espeak = ;// optionally set the path to the `espeak` cli program if it's not in your PATH//espeak.cmd = '/usr/bin/espeak';espeak;// optionally add custom cli arguments for things such as pitch, speed, wordgap, etc.espeak;
There are some related modules, although there seems to be some scattered forks and it wasn't clear to the author which was updated and/or compatible with Node.js/npm at the time of writing this.