356 BELL SYSTEM TECHNICAL JOURNAL 



meaning of the words. Some of my friends in the musical profession 

 object to this statement of the situation but I think you will agree 

 that a singer's principal aim is to produce beautiful vowel quality and 

 to manipulate the melodic stream so as to produce emotional effects. 

 To do this, it is necessary in singing to lengthen the vowels and to 

 shorten and give less emphasis to the stop and fricative consonants. 

 It is for this reason that it is more difficult to understand song than 

 speech. 



Characteristic Pitch or Frequency Levels for the Vowels 



Now let us examine part of the speech wave of Fig. 1 in more detail. 

 Consider the vowel in the word "shoe." 



The fundamental cycle was repeated 170 times per second. It is 

 evident that the second harmonic is very much magnified until it is 

 nearly as intense as the fundamental. In Fig. 8 is shown another 



0.21 SEC. 



Fig. 8 — Oscillogram of vowel u. 



oscillogram of u intoned at 120 cycles per second. In this case the 

 3rd harmonic is magnified. An analysis of a number of u sounds 

 shows that components falling between 300 and 400 cycles per second 

 are always reinforced. This reinforcement is probably due to the 

 resonance characteristic of the mouth cavity. 



Similar characteristic low pitch regions exist for the vowels in the 

 words, put, tone, talk, ton and father. A characteristic high pitch 

 region also exists for these sounds but the intensity of the components 

 falling in it are much less. For the vowels in the words tap, ten, 

 pert, tape, tip and team there are two characteristic regions of rein- 

 forcement which are of approximately the same intensity and which 

 are independent of the fundamental pitch. This is illustrated in 

 Fig. 9, which gives a spectrum analysis of the vowel "e" pronounced 

 at the four pitches indicated. The characteristic regions are at 375 

 cycles per second and 2400 cycles per second corresponding to pitches 

 — 1.4 octaves below and + 1.3 octaves above the reference pitch. 



Experimental work ^ has indicated that for American speech the 

 characteristic pitch regions for the vowels and semi-vowels are those 

 shown in Fig. 10. For the first six vowels the components corre- 



2 "Speech and Hearing," Harvey Fletcher, pp. 58, 59. 



