352 BELL SYSTEM TECHNICAL JOURNAL 



essentially a melody. The melody of the sentence whose wave form 

 is shown in Fig. 1 may be illustrated graphically as indicated in Fig. 2. 

 In this figure the ordinates represent the pitch in octaves below or 

 above a tone having a frequency of one kilocycle per second; or if 

 the frequency / is measured in kilocycles, then the pitch P is given by 

 the equation 



P = \0g2f. (1) 



The abscissas represent the time in seconds. The lower curve gives 

 the changes in the pitch of the fundamental and represents the melody 

 as ordinarily understood in music. The middle two curves represent 

 the pitch positions of the strongest harmonics. The location of these 

 positions is determined by the resonant properties of the throat and 

 mouth cavities. These curves may be considered as secondary melodic 

 streams. The combination of these two secondary melodic streams is 

 interpreted by the senses as a sequence of spoken vowels rather than 

 as a series of pitch changes. The small number above each part of 

 the curve gives the number of the harmonic which is augmented by 

 the resonance of the mouth or throat. For the sound e in bench the 

 4th harmonic was the strongest at the beginning of the sound, but 

 the 5th came in strongest near its end. I have tried to indicate the 

 relative intensities of the harmonics as the sound proceeds by the rela- 

 tive thicknesses of the lines. An examination of the oscillogram shows 

 that the intensity of the harmonic always increases as its pitch becomes 

 nearer the characteristic pitch for the vowel being spoken. 



As indicated by the short lines at the top of the chart, there exists 

 at certain intervals high pitched components which are characteristic 

 of the fricative sounds. The unvoiced sounds t, k, f, z and sh, exist 

 only when the three melodic streams are stopped. The high pitched 

 components of the voiced sounds, j, th and b, are superimposed upon 

 the three melodic streams. 



Besides these four important streams of speech (Fig. 2), there are 

 a great many others with intensities which are in general much lower, 

 but when combined with the main streams they determine the kind 

 of voice, that is, whether it is smooth and musical or rough and 

 harsh. The main melodic stream for a woman's voice is between the 

 pitches — 1 and — 2 octaves while for a man's voice it is between 

 — 3 and — 2 octaves. The secondary melodic streams produced 

 while speaking the same sentence are approximately the same for man 

 and woman and of pitches shown in Fig. 2. 



In Fig. 3 is shown an oscillograph of the sentence "How are you?". 



