I am using the MOVI Arduino shield for speech recognition and speech synthesis. MOVI is cloudless and therefore privacy friendly, no Internet or PC connection required. In the little setup below I am testing I2C communication between two Arduino Mega's. A char array will be created by typing some text into the serial monitor. The text will be transmitted to the second Arduino, and MOVI is then doing the text-to-speech synthesis. I still need to experiment with different loud speakers, amplifiers and resonating bodies to get a really beefy sound.