SPEECH POWER AND ITS MEASUREMENT 



655 



applicable to sounds other than speech, provided they are sustained 

 or can be repeated. Two illustrations are given. Fig. 7 is for a 

 piano composition. Characteristic in comparison with speech is the 

 relative smallness of high frequencies. Spectra of radically different 

 types of compositions were found to be rather similar. The instru- 

 ment, the room acoustics and the position of the microphone largely 

 determine the picture. In this case, the microphone was about 7 ft. 

 from the keyboard, and the reverberation time between 1.2 seconds 

 at 100 p.p.s. and 1.0 second at 4000 p.p.s. 



Fig. 8 gives the spectrum of street noise entering a laboratory 



■100 



100 1000 10000 



FREQUENCY -CYCLES PER SECOND 



Fig. 8 — -Average pressures per frequency interval of 1 cycle per second — -street 

 noise — -lOth floor Bell Telephone Laboratories facing east — -2.50-3.20 P.M., Oct. 10, 

 1928 — average total pressure = 0.75 bar. 



window, ten floors above the street. Street traffic and an elevated 

 railway are the chief contributors, the measurements being taken 

 during traffic peak hours. The relative poverty in high frequencies 

 is even more pronounced than for the piano. 



We shall now consider the type of apparatus intended primarily 

 for measurements when the speakers are not acoustically controllable. 

 Instead of averaging over a large number of words the measurement 

 is essentially that of mean power in syllables. Usually it covers the 

 whole spectrum rather than a particular frequency band. It has 

 been widely used for controlling amplification in radio broadcasting, 

 in phonograph and film recording of speech, and for rapid measurement 

 and electrical control of speech levels in telephone conversations. 

 A typical device of this sort is the "volume indicator" ^ shown in 



^ Essential features of this apparatus are shown by E. L. Nelson, U. S. Patent 



No. 1523827, filed 8/31/22. 



