Hi Kirk, Those were the days, weren't they? That fist system of mine was by Digital Group out of Denver, with an S-100 bus, and it was the system used for the column in Byte magazine called "Ciarcia's Circuit Cellar." But Digital Group went belly up, and Byte magazine would not allow any references to the system after that. Story of my life! I wasn't wealthy, I just got screwed a lot. My first speech software was table driven, not text-to-speech. I had a table of the 256 Ascii characters, with a phoneme string giving the name of each character. The plan was to just put the thing on line and let it spell stuff for me. But when I connected to the campus computer for the first time, it sent me so many rubout characters as part of the login sequence, and it took my stupid terminal so long to say "rubout" a million times, that I had timed out before I could log in. That taught me the importance of killing speech and reviewing the screen. Later I expanded the table to include multi character strings and the phonemes necessary to pronounce those strings, so it became a mix of spelling and whole word output. Pronunciation was perfect, but lots of stuff had to be spelled. I kept enhancing the table to cut down on the spelling, but it was a futile chase. Curious, when I showed the system to people, they would type a test phrase to see if my computer "knew" the words. And their phrases always contained the words fuck and shit. So I had to put fuck and shit into my pronunciation table. But what I did was associate a different phoneme string for those words, so when someone typed "go fuck yourself" my computer said impressively, "go intercourse yourself",. and similarly, "you are full of excrement" instead of what was actually typed. More fun than a barrel of monkeys. But then the TTS algorithms came along and I was left in the dust yet again. Chuck -- The Moon is Waning Gibbous (85% of Full) Only 10 kinds of people: those who do binary, and those who do not. But you can get a few downloads from http://www.mhcable.com/~chuckh