Okay, here you go, I drew it up for you. The microphone low pass filter cuts out everything below 250Hz; the AGC monitors and adjusts the incoming signal amplitude so it always stays within the range of the ADC (it can adjust the gain from 0 to 42dB, in 6dB steps); the antialiasing filter cuts out all the high frequency clickity-clack from the analog switches; and the ADC converts the resulting audio to an 8bit serial data stream. The remaining logic inputs are for you to decide how best to address (from a micro or equiv.), and software, well, that's altogether something else! I don't know where you're going to find that.
Also, regarding speech synthesis, the Lis'Ner 1000 used an SSI-263 phonetic speech chip in conjunction with the circuit below. I have not added it as it requires that particular chip: