ij; 



CHAPTER I. \C--^ 



RECORDING SPEECH VIBRATIONS. 



The term "speech vibration" is used here to indicate the movement 

 of a particle of air a short distance in front of the mouth during speech. 

 The short distance must be so selected as not to exclude the vibratory 

 effects coming through the nose. The curve corresponding to a speech 

 vibration is termed a " speech curve." 



The vibration from the voice may be recorded by several methods. 

 The earliest method was that of Scott's phonautograph (1856), wherein 

 the air vibrations passed down a trumpet or tube to a thin, soft membrane ; 

 a light lever recorded its vibrations on a revolving cylinder coated with 

 smoke. The phonautograph did not turn the curve back into sound again. 



The phonautograph has been modified and improved in various ways 

 by a series of investigators, but in spite of all improvements I doubt that 

 it has at present any value for registering speech curves, although it is 

 of great use in recording the melody of the voice. 



The reasons for completely ruling out the phonautograph as a speech 

 recorder are the following: First, there is no guarantee whatever that 

 the apparatus records correctly. The curve can not be turned back into 

 sound so that the success of the record can be judged by the ear. Second, 

 experience in the development of talking machines has shown that crude 

 recorders of any kind fail to give good reproducing records. It is only 

 after some years of experience under control of the ear that the phono- 

 graph and the gramophone recorders have been developed to their present 

 stage of success by careful elaboration of every detail in the construc- 

 tion. It makes, for example, a difference whether the glass diaphragm 

 in a recorder is fastened to the box by dextrine or wax, whether the 

 recording point has a base that is circular, elliptical, or oval, etc. Not 

 one of these innumerable factors receives any attention in the phonau- 

 tograph. Where large vibrations are requisite, as in the gramophone 

 recorder, a hundred different glass diaphragms may be tried before one is 

 found that records sounds satisfactorily. The reason why the particular 

 one succeeds is not known. In the phonautograph a diaphragm of unknown 

 peculiarities is used with no possibility of testing the results. I do not 

 mean to say that the phonautograph records the vowel [a] with an [o]- 



13 



