VISIBLE PATTERNS OF SOUND——POTTER 201 
the frequency of the harmonics varies with time. Patterns of the 
second type show only the broad frequency and time distribution of 
energy resulting from selective modulation. It is such modulation, 
produced by variations in the voice cavities accompanying the forma- 
tion of word sounds, that conveys the information in speech. In- 
cidentally, the vertical striations appearing in plate 1 (0) are pro- 
duced by beats between adjacent harmonics so that in the voiced parts 
their density or frequency of occurrence is a measure of pitch, in- 
creasing as the pitch increases. 
The patterns of plate 1 and others described later were made by an 
instrument that we have called the sound spectrograph. In this in- 
strument, the sound to be pictured is recorded initially on a loop of 
magnetic tape and played back repeatedly into a scanning filter, the 
pass band of which is moved slowly across the frequency spectrum. 
The scanning filter output is connected to a stylus that makes a trace 
upon a loop of electrically sensitive paper. Recording paper and mag- 
netic tape loops are moved in a fixed relation so that successive scan- 
ning cycles are recorded side by side, thus building up a frequency- 
time-intensity picture. 
If the words pictured in plate 1 were repeated at different times by 
the same speaker, the repetitions would look much alike unless a de- 
liberate attempt were made to change the voice. If the words were 
spoken by different individuals they would also have a similar ap- 
pearance, although the pattern shapes would vary with individual 
characteristics in much the same way that handwriting varies among 
individuals. To the extent that words sound alike they will also look 
alike in visible speech form, and to the extent that they sound dif- 
ferent they will look different. That is, of course, to be expected if 
the portrayal is accurate. 
The similarity in word patterns for various individuals is illus- 
trated by plate 2. Here are shown six enunciations of the word 
“speech” by as many speakers. As indicated, the upper three are 
female voices and the lower three male. The brief descriptions at 
the right are only intended to give an impression of the wide range 
of voice quality represented and should not be interpreted as meaning 
that all speakers with a “throaty voice” or a “Scotch-Irish accent” or 
an “English accent” would produce patterns that are just like those 
so identified. The important point in this illustration is the evidence 
that characteristic differences in pronunciation do not overshadow 
the similarities that enable one to recognize particular words. Such 
similarities illustrate the possibilities of these patterns for visual 
hearing. ; 
In the studies of sound patterns as applied to the problems of the 
deaf, the objective thus far has been to determine whether the pat- 
