./ i>y.\.iMic.tL STUPy OF Tin: roifiiL solwds 



2ii 



aiul rise lu-.irK to ni.ixiiiuiin anipliliiili'; >c('iiiiii a iiiiilillr prriiMl it) 

 which thf gi-iuTal amphtuiU' is iK-arly loiislant l)ul with varying 

 phase rt'lations l)t'twevii tlu- ditTiTeiU components and lasting about 

 0.17 second, and linally a periml of gradual deca\- lasting about 

 .Oy second in which all the coni[)onents disappi'ar. A typical record 

 so obtained is shown in {'ig. 2. 



A brief descriiuion of the inethud of inechanicalh anal>/,ing si:ch 

 a record is given in the appendix. The essential p)oint of the analysis 

 is that the whole record from start to finish is taken as the unit for 

 analysis and the data obtained are therefore the a\er.ige charac- 

 teristics of the sounds throughout their duration. 



It is usual to exhibit the properties of a vowel sound in a spectrum 

 diagram showing the amplitude of the component vibrations as a 

 function of their pitches or frequencies. For each vowel sound there 

 are, in addition to fundamental tones, certain characteristic regions 

 of resonance which may be at high or low frequencies. It would 

 be possible from the results of this analysis to present the sound 

 spectra of each vowel showing the relative amplitudes for the dif- 

 ferent frequencies as present in the original air vibration' but this 

 treatment has been modified to take into account the relative im- 

 portance of the various pitches in hearing. Using the data available 



' In previous publications (Phys. Rev. XIX, 1922, p. 228, Fig. 7, and Belt System 

 Technical Journal, \'o\. I, .\o. 1, p. 124,) data have been given showing the actual 

 distribution of energy in average sjx^ech. The tremendous concentration of energy 

 in the lower frequencies is somewhat misleading unless account is also taken of 

 the much reduced sensitivity of the ear in this region. 



