This program was originally just a hack of John Wasser's phoneme program, which converts english text to phonemes (see example below). It's since expanded into a generalized compiler for the Story Teller.
The speech portion of the Story Teller still uses a hacked up version of phoneme buried within it, for an antique phoneme-based speech chip, the General Instrument SPO256-AL2, available from BG Electronics and in the past, Radio Shack. It was "high tech" around 1978. The text-to-phoneme software is not mine, but written by John Wasser, last updated 1985, based upon U.S. Navy code from the 70's. I massively updated the text-to-phoneme tables; in 1980, when a 4MHz CPU with 64K (K not M) of memory was very high end, table size and complexity was a real issue.
Phoneme speech is a bankrupt paradigm, for any number of reasons, and the G.I. chip is terribly old. But it fit my needs (phoneme opcodes fit on a punched-tape word...). Text-to-phonemes is terribly attractive, until you actually think about it. It's a very sophomoric approach to computer communication, which makes it either useless and embarrassing, or a well from which to drink funny-tasting fluids with hallucinatory side-effects.
Here is T2A.ZIP, an archive containing the T2A program and all it's sources. It was compiled in Borland C. This is a "final" version, one I'm using myself. Hacks to the lookup table can never said to be truly "done". Look yourself, and understand how it works, and you'll see why.
If you run it in -P mode, t2a produces this fun stuff:
text: I saw the best minds of phonemes: AO AY s AO AO DT IY b EH s t m AY n d z AX AX v h text: my generation destroyed by phonemes: m AY j EH EH n ER EY P1 SH AX n d EH s t R1 OY D1 b AO AY text: madness , starving hysterical naked , phonemes: m AE d n EH s P5 s t AE r v IH NG h AA AY s t EH r IH k AX l n EY k D1 P5
I added the convenient spaces for readability; and inter-word spaces have their own "phoneme" for silence (actually four of them of different lengths), removed here for demonstration purposes. You can see one embedded in the word "generation", (P1; 10mS silence), as a way to psychologically decompose the sounds for easier understanding.
Here is another example conversion. The audio here really sucks, I got it to work much better than this, which is from late 1998, but it gives a feel for what the conversion is like. The sample text is from William Burroughs' NAKED LUNCH, the Meet Cafe paragraph, edited here.