It may be possible for a deaf person and a hearing person to converse over the telephone. The deaf user "speaks" by typing to a text-to-speech converter using a low-redundancy keyboard system. The hearing user speaks sentences one word at a time to a large-vocabulary, isolated-word-recognition system that displays a sentence lattice (a sequence of sets of likely matches for each word spoken). The deaf user then tries to find a sensible path through the sentence lattice. Successful implementation of such a system, under development at SRI International, requires adequate performance in text generation speed by deaf users, text-to-speech intelligibility, and word-at-a-time speaking by hearing users, as well as large-vocabulary speech recognition and disambiguation of sentence lattices by deaf users.

Subjects were exposed to alternate conditions of quiet and noise (computer line printer noise at 100 dB, A-weighted) while performing a demanding, rapidly paced, psychomotor task. Heart rate, peripheral pulse amplitude, and peripheral blood volume were monitored throughout four 8-min test sessions. There appeared to be no increase in physiological stress or in errors in task performance during the noise as compared to the quiet test segments. An attempt is made to explain possible causes of differences between the results obtained in this study and those reported for a earlier, somewhat similar, study by Cohen et al. [’’Noise Effects, Arousal, and Human Information Processing Task Difficulty and Performance,’’ Department of Psychology and Industrial Engineering, North Carolina State University, Raleigh, NC (1973)].

Experiments were conducted on the effects of various expected noise conditions upon autonomic system activity in men during rest and work (bicycle ergometer). Skin temperature and heart rate were not appreciably, if at all, affected by any of the noises. Wide-band, predominantly low-frequency noise, at a level of 92 dB, A-weighted, generally caused a decrease in pulse amplitude (indicative of constriction of peripheral blood vessels) during either work or rest, whereas an equally intense one-third octave band of random noise, center frequency of 3150 Hz, had no appreciable effect on pulse amplitude. Repeated exposures to the wide-band noise showed some adaptation or habituation of the pulse amplitude response during conditions of either work or rest. The results were the same for fast rise time (impulsive) bursts of wide-band noise as for bursts with slow, gradual onsets. Rapid interruptions of wide-band noise caused less of a decrease in average pulse amplitude than did uninterrupted noise. It is suggested that constriction of peripheral blood vessels in response to expected intense low-frequency or wide-band noises is more related to auditory, reflexive protective mechanisms than to autonomic system responses generally considered to be stressful to the organism.

The acoustic processing in the SRI speech-understanding system is described. This system relies heavily upon semantic and syntactic information for understanding the spoken utterance. This information predicts the presence of various words at particular places in the utterance and thereby guides the acoustic processor. The acoustic processor is implemented as a set of word-verification functions which establish the likelihood that a predicted word actually is present at a specified point in the utterance. This approach allows a straight-forward solution of problems of segmentation and within-word coarticulation, which have caused great difficulty in traditional attempts at continuous speech recognition. Typical word verification functions are presented, with illustrations of different methods of analyzing coarticulation effects. The possibility of incorporating these techniques in traditional systems is discussed. Finally, methods of extending these techniques to deal with coarticulation effects which occur across word boundaries are discussed.