| Abstract | [Introduction]
The history of attempts to construct reading machines for the blind may guide us in our attempts to devise speech-listening aids for the deaf. The goal of the early reading machine work was to construct an acoustic alphabet, that is, to find a set of discrete, discriminable acoustic patterns that might substitute for the visual alphabet. To devise such a set-of tones, chords, bursts of filtered noise, perhaps varying in amplitude or duration-was not difficult, and , if the patterns were presented one at a time in a comfortable test format, listeners readily learned to identify them. However, if patterns were presented in rapid sequence, listener performance dropped precipitously. In fact, no one has yet devised an acoustic alphabet more effective than the dots and dashes of Morse coed, with which highly skilled operators may reach a rate of 30-40 words a minute-roughly one fifth of the rate at which we typically follow spoken language. Not surprisingly, the search for an acoustic alphabet for reading machines has been largely abandoned in favor of synthetic (or compiled) speech. |
|---|
