SPEECH POWER AND ITS MEASUREMENT 647 



of measuring the instantaneous pressures and velocities respectively? 

 First, as to acoustic ammeters: perhaps the two best known forms 

 are the Rayleigh disk and the hot-wire microphone. The disk gives 

 absolute values of U but its response is so slow that it can be used 

 only to measure the efifective values in comparatively sustained sound 

 waves lasting, say one second or longer, or for the ballistic integration 

 of shorter pulses. Hence it does not give a measure" of the instan- 

 taneous velocities for even the slowest audio-frequency vibration- 

 Furthermore, its use when exposed to the speaker in an open room, 

 is rendered difficult by its susceptibility to spurious air currents. 

 The only application of the disk to vocal power measurements which 

 has come to my notice, is one by Prof. Zernov,- published in 1908. 

 For sustained loud singing and shouting he found energy densities 

 ranging from 0.3 to 2.0 X lO^'* ergs/cm.^, at 2 meters distance from 

 the singer. Assuming uniform distribution over a hemisphere of 2 m. 

 radius, this gives a total power output of the voice of the order of 

 50,000 microwatts. At 2 m. distance reflections from the walls of the 

 room materially raise the energy density as compared with that in a 

 progressive wave. 



Another acoustic ammeter is the hot-wire microphone. The re- 

 sistance variations of the wire tend to follow the instantaneous values 

 of the air particle velocity in the sound wave, but the sensitivity is a 

 complicated function of the frequency, rapidly decreasing as the 

 latter increases. Hence it does not give true oscillograms of speech 

 sounds which in general include a frequency range of six or seven 

 octaves. 



Still another device in this general cla=;s is the glow-discharge micro- 

 phone. Its response is determined largely by the amplitude of the 

 air particle motion (E. Meyer, E.N.T., v. 6, 17-21, 1929). Hence, for 

 constant sound intensity the response is inversely proportional to the 

 frequency, roughly. No method for its absolute calibration has been 

 proposed other than comparison with a microphone of known per- 

 formance. The frequency response and the somewhat erratic be- 

 havior of the device in its present status, render it rather unsuitable 

 for speech power studies. 



Nearly all our information concerning speech power and the wave- 

 form of speech sounds has been obtained with acoustic voltmeters, 

 i.e., with devices responding to pressure in the sound wave. To a 

 first approximation, the ear belongs to this class. It includes the 

 resonators which Helmholtz used in his vowel studies. Generally, 

 the vital element in these devices is a diaphragm which vibrates with 



2 Aym. der Phys., 26, 94, 1908. 



