MaryTTS is a free and open-source Text-To-Speech synthesizer made by the DFKI. More information about the system itself can be found on their website. The GitHub repository can be found here. Computronics uses a slightly modified version of the standard runtime jar so that it works in a Minecraft Forge environment.
Due to the size of the system, MaryTTS is not included in Computronics by default. To be able to use the Speech Box in Computronics, you need to install MaryTTS on the server; no processing happens on the client side so clients do not necessarily need it unless you would like to use the Speech Box on a singleplayer instance.
Computronics requires three files to be installed in a directory called 'marytts/' inside your game instance's directory (on the same level where your 'mods/' directory is): The TTS system itself, a language and a corresponding voice. Due to dependency issues inside a Minecraft instance, Computronics requires a custom version of the normal runtime jar file you can get on the website. You can get it here:
Furthermore, you need to choose a language and a corresponding voice; download those two files and put them into the directory; voice files usually start with “voice-” and language files usually start with “marytts-lang-” in their file name. Various languages (including English) are available in Mary's official download zip file (in its 'lib/' subdirectory), but others can be found on the Internet as well. Once you have chosen the language you would like the speech box to speak, you need to choose a voice speaking that language. A female US English voice is included in Mary's main zip file, but other English voices as well as voices for other languages can be found on the MaryTTS GitHub organization. Be aware that some of them may be outdated and not all might work. Also note that some downloads provide the language or voice as a zip file which contains the actual jar file; in that case, the jar file has to be extracted and placed in the directory to load properly.
Once the server has those three files available in the 'marytts/' directory, the speech box should be able to speak the installed language with the installed voice. The server log should mention a successful initialization.