Software Automated Mouth (S.A.M.) not working :(

Does the version of S.A.M. using the internal speaker work on your clone?

Also see here:

https://www.applefritter.com/comment/102542#comment-102542

"I think there were two versions of the software.

One version didn't require additional hardware, and one version required the mentioned card with a speaker."

~Two years ago I had the same problem.  A photo of the "S.A.M." sound card is here:

https://www.applefritter.com/content/help-needed-identify-strange-sound-card

after that sound card was identified, I also knew why my copy of S.A.M. was silent. Plugged in the card and it spoke !

(Unlike modern computers, it does not have a way to identify if the card is there. Not very foolproof = consumer friendly. Thanks to the fine work of a bunch of lawyers, nowadays corporations making consumer products must treat the consumers as possibly being complete imbeciles with an IQ of 60 or lower, but ironically, they still assume these moronic consumers can read ... what a waste of paper ! The movie "Idiocracy" show the right way: pictograms. IKEA does it already. Their assembly manuals are with no text, pictograms only. Perfect for illiterate people from all over the world).

One time in the future (if I get that time) I will make a disassembly of S.A.M. to see how it works. Back in the 1970s / early 1980s there were two market dominating, patented speech synthesis methods, the "Linear Predictive Coding" as used by Texas Instrument's TMC0280  (i.e. in their Speak & Spell toy) and the "Mozer Speech Synthesis" invented, developed and marketed by Forrest S. Mozer which used a super clever , if not brilliant, method to compress natural speech samples, which lead to a very efficient speech IC, the TSI14001, which was used in a calculator for the blind (I have one in my collection) and in most of the speaking "Chess Challenger" computers made by Fidelity (all in my collection, too).

I find Forrest Mozer's technique most amazing as it produces the best speech quality with the least amount of hardware. They had an improved version which was advertised in electronics magazines in the early to mid 1980s, which synthesized a naturally sounding female voice of excellent quality and still very low memory consumption. The ad had a 1-800 phone number where you could listen to that voice.  (Female voices are much more difficult to synthesize than male voices, but note the trick with the phone line ... sneaky).

I'd be curious which synthesis method S.A.M. did use. Maybe it was simple and stupid (KISS principle) and just sampled phonemes. There is a nasty background tone in S.A.M.'s speech which hints towards that. I had a similar artifact in all my own experiments to develop speech synthesis. This would go away with sophisticated digital filtering but at the time the microcomputers were too slow to do that in real time. It is known that Forrest Mozer prepared his speech synthesis data sets using FFT techniques, but offline, not in real time, so no digital or analog filter was needed in the TSI speech chips.

- Uncle Bernie

(P.S.: there also are collectors out there who collect these obnoxious talking clocks, a fad of the 1980s, most were based on National Semiconductors 'Digitalker' chip, which might involve either technique, I did not look into this yet. But I hate these clocks. Like cuckoo clocks, they should come with a shotgun as an accessory, to silence them forever ;-)