I used mostly the present tense to describe S.A.M., since I run it on an emulator, and feel the past tense would be somehow inappropriate.
S.A.M. (the Software Automatic Mouth) is a pure-software speech synthesizer for the Atari and Apple platforms, released in 1982 by Don't Ask, Inc. I have only used the Atari version, so not everything written here necessarily applies to the Apple version as well.
The synthesizer uses phoneme concatenation to create surprisingly convincing speech (well, a little above Speak & Spell level, and that's saying something for the early 80's) from phoneme strings such as "AY4 AEM AH KUMPYUW3TER", and includes a translation utility, called RECITER, which primitively bridges the English-to-S.A.M.-ish gap. S.A.M.'s phonemes are synthesized, rather than recorded, saving precious storage space; This, naturally, comes at the cost of CPU cycles, and so much so, that whenever S.A.M. speaks, it must black out the screen to save cycles.
S.A.M. was not the first speech synthesizer, but it was probably the first speech synthesizer that people could afford, especially since it ran on inexpensive 8-bit home computers. Its cheapness aside, though, S.A.M. offers features that had usually been seen only in astronomically-priced synthesizers when it was released; Particularly, it allows machine language and BASIC developers to include speech in their own programs through a simple interface.
An exciting new breakthrough:
{PICTURE OF A 5.25" FLOPPY DISK}
This disk can now make your computer speak!
It's the Software Automatic Mouth - S.A.M.
The brand new, all-software, high-quality speech synthesizer from DON'T ASK.
Incidentally, Sam is also the name of the SAPI voice which ships with Microsoft Windows (2000/ME and onward), sounds like crap, and would not have passed for a speech synthesizer even in S.A.M.'s day.