ENERGY DISTRIBUTION IN SPEECH 



125 



tion. One of these curves is shown in Fig. 8. Curve A was constructed 

 by one of the writers in 1916 in an attempt to synthesize the energy 

 curve from the energy distributions of the vowel sounds, using the 

 vowel analyses of Dr. Dayton C. Miller. Curve C is the composite 

 "continuous speech" curve of Fig. 7. The vowel sounds analyzed by 



Fig. 8 — Energy Distribution: .4. Synthesized from Vowel Records of D. C. Miller 



(1916). B. Disconnected Speech Analysis of this Paper. C. Connected Speech 



Analysis of this Paper (from Fig. 7) 



Miller were intoned and the vowel sounds analyzed by us were spoken, 

 but Miller's work seemed to show that there was no essential differ- 

 ence between intoned and spoken vowel sounds. There is, however, a 

 very noticeable difference between Curve A and Curve C, the energy 

 in the fundamental tone of the speaker's voice coming out much 

 more strongly in Curve C. We should expect that our improved ap- 

 paratus would record the energy in the lower frequencies more cor- 

 rectly than the apparatus heretofore used but as we used different 

 test material (connected speech instead of disconnected syllables or 

 vowel sounds) it is not immediately evident which of these two factors 

 is responsible for the differences between the A and the C curves. 



In order to investigate this point more fully the testing routine for 

 all six speakers was repeated, using instead of the fifty-syllable sen- 

 tence, the fifty disconnected syllables of one of the standard articula- 

 tion testing lists, as used by Dr. Fletcher ii this laboratory. The 

 results for energy distribution are shown in Fig. 9, Curve A being 



