Software:Software Automatic Mouth

From HandWiki
Software Automatic Mouth
Atari cover for Software Automatic Mouth
Original author(s)Mark Barton[1]
Developer(s)Don’t Ask Software
Initial release1982; 44 years ago (1982)
PlatformApple II, Lisa, Atari 8-bit, Commodore 64.
TypeSpeech synthesis
File:C64 Software Automatic Mouth demo.flac
A demo of SAM on the C64

Software Automatic Mouth, or S.A.M. (sometimes abbreviated as SAM), is a speech synthesis program developed by Mark Barton and sold by Don't Ask Software. The program was released for the Atari 8-bit computers, Apple II, and Commodore 64. First appearing in the October 1982 issue of Antic magazine, it was one of the first commercial all-software voice-synthesis programs. S.A.M.'s main selling points were its software-only design, which required no additional hardware, and a relatively low cost of $59.95 for the Atari version.[2][3]

Don't Ask Software also sold PokerSAM, a poker game with speech.[4]

Technology

The Apple version uses an included expansion card which contains an 8-bit DAC.[5] The Atari version makes use of the embedded POKEY audio chip. The audible output is extremely distorted speech when graphic and text display is turned on.[6] The Commodore 64 makes use of the 64's embedded SID audio chip's 4-bit volume DAC.[7][8] When producing speech, the Commodore 64 version blanks the screen as the program accesses memory, although a "light" mode can be activated that leaves the screen active, producing a "gravelly" voice. Active sprites also cause a deterioration in the quality of voice output.[9]

Phonetic Mode

SAM accepts input in an ASCII-based phonetic notation, derived from the International Phonetic Alphabet, comprising roughly 50 phonemes. Each phoneme is written as a one– or two-letter code paired with a sample sound—for example, IY (as in "feet"), AE ("Sam"), or NX ("song"). The set is organized into vowels, diphthongs, and voiced and unvoiced consonants, alongside special symbols such as Q for a glottal stop and DX for a flap. Stress is indicated by appending a digit from 1–8 to a vowel.[10] A full word written in Phonetic Mode would look like this: WIHKIHPIY4DIYAX

Phonetic mode addresses some of the limitations of SAM's rule-based text-to-speech (RECITER) system, which mispronounces irregularly spelled English words and cannot reliably place stress; entering phonemes directly gives the user more precise control over both.[10]

Legacy

S.A.M. was used as the basis for the original MacInTalk speech synthesis software.[1]

References