236 



VOICE COMMUNICATION, I 



speech energy that a talker gets into his 

 consonants. Consonants, rather than vow- 

 els, serve as the major clues in identifying a 

 word. At the same time, since they nor- 

 mally contain much less energy than vowels, 

 consonants are the speech sounds most read- 

 ily masked by interfering noise. 



3. "Consonant precision" — the ideal pro- 

 nunciation for intelligibility in noise exhibits 

 a scrupulous, even exaggerated, articulation 

 of the consonants. 



4. "Noise penetrating quality" — subjec- 

 tively perceived as the possession of a clear, 

 ringing, vibrant, rather than a harsh, muf- 

 fled, dull voice quality. 



Aspects such as the duration of single 

 words, rate of speaking, steadiness of voice- 

 intensity and pace in speaking, the normal 

 pitch of a voice, and the average "shape" 

 of the voice spectrum of an individual talker, 

 were found to be of little or no importance 

 in determining intelligibility. Ratings made 

 by specialists on the basis of a standard 

 college speech interview (depending in con- 

 siderable part, of course, on aesthetic con- 

 siderations) were also found to be of very 

 limited validity for the peculiar situation of 

 a military talker who must make himself 

 heard in noise (10, IC-67, pp. 3-8). 



The ability to talk intelligibly has been 

 found to have little overlap with the ability 

 to listen in noise. Ideally, therefore, per- 

 sonnel for important communications posts 

 should be selected on the basis of both lis- 

 tening tests and talking tests. Talking tests 

 must be administered to one individual at 

 a time, by judges trained to rate a voice 

 according to its possession of the qualities 

 of sustained loudness, precise and powerful 

 articulation of consonants, and a vibrant, 

 penetrating vocal timbre. Reliable talker- 

 rating tests, consisting of a variety of speech 

 materials, and taking from one to three 

 minutes to administer, have been developed 

 for use by the Armed Forces (12, 13, 14). 

 The validity of such tests, when adminis- 

 tered by experienced personnel, has been 

 established against the criterion of the actual 



percentage of a man's speech correctly iden- 

 tified over an interphone system by a group 

 of listeners in a noisy environment (10, IC- 

 67). A special interphone for rating talkers 

 has also been developed which aids judges 

 in making valid estimates of intelligibility 

 in noise (11). This instrument permits sim- 

 ulated engine noise to be introduced into 

 the earphones of speaker and judges during 

 the speech interview, and includes a VU 

 meter by which the loudness of the talker's 

 voice may be measured with precision. 



The limited personnel available for as- 

 signment to a single submarine, and the 

 need for a telephone talker to possess other 

 than talking and listening skills alone, may 

 make it inadvisable to select operators too 

 rigidly on the basis of these tests. Never- 

 theless, it would seem wise to have a listen- 

 ing and talking score available for each man 

 in the Submarine Service, so that (1) no man 

 too incompetent will be chosen for a critical 

 communications post, and (2) any subma- 

 rine lacking a sufficient number of inherently 

 able talkers may recruit additional personnel 

 known in advance to possess the necessary 

 talents. 



III. The Training of Communications Oper- 

 ators 



Whatever their initial ability, the talking 

 and listening skills of all men can be im- 

 proved remarkably by practice and by ex- 

 pert instruction. In one experiment, for ex- 

 ample, a few minutes of instruction on the 

 proper positioning of the microphone and on 

 the optimal loudness of the voice resulted 

 in an average gain of almost 50 percent in 

 the number of messages a group of speakers 

 was able to make understood against a back- 

 ground of motor noise (10, IC-54, and IC- 

 60 pp. 6-12). Skill in listening also improves 

 with practice, although more slowly: in one 

 typical instance, fifteen hours of listening to 

 speech in noise effected a gain of 50 percent 

 in the number of messages one set of listen- 

 ers was able to identify correctly (5, pp. 

 17-19). Expertly supervised courses in 



