630 BELL SYSTEM TECHNICAL JOURNAL 



and the air particle displacement, 



^=-r:j— ; / pdt (41.0 is a resistance factor) 



for a wave travelling in the positive direction. 



From the power intensity thus found at the transmitter we can 

 obtain an estimate of the power dev^eloped by the speaker. With 

 the transmitter surrounded by a plane reflecting surface so as to 

 give reflection for speech frequencies, the pressure is doubled and 

 the power intensity quadrupled ov^er the values they would have in 

 free air, hence the observed intensity is divided by 4. The usual 

 assumption is made that this same intensity is distributed over a 

 hemisphere whose center is at the speaker's lips. Hence the required 

 estimate of the speaker's power is obtained by multiphing the meas- 



ured power intensity at the transmitter by the factor — = 127. For 



the sake of convenience, these two values are always given together 

 in the accompanying tabulated results. 



Instantaneous and Mean Power 

 In dealing with the power in a syllable, the matter of greatest 

 interest is the maximum values attained by Pi and Pm throughout 

 the entire syllable. These maxima will be denoted by Pi and Pm, 

 respectively. Table I shows their approximate ranges in the case of 

 accented syllables. 



TABLE 1 

 Instanlutieous and Mean Power 

 Typical Maximum Values for an Accented Syllable 

 Speaker's Power Power Per Ciii.- 

 Microwatts at Transmitter 



£i 1000 to 2000 8 to 16 



Pm 60 to 120 0.5 to 10 



At this point it is worth while to consider an application of the fore- 

 going. A salient characteristic of speech waves is the generally high 

 ratio of peak value to mean square value (peak factor), as can be 

 inferred from Fig. 1. Failure to take this into account frequently 

 causes load distortion in speech transmitting amplifiers. It sometimes 

 happens that the effective output voltage or current has been meas- 

 ured, and the assumption of an equivalent sine wave (i.e., one having 

 the same effective value) is made; but this leads to a large error in 

 the estimate of the peak value. Thus with an insufficient allow- 

 ance made for the peak voltage impressed upon the grid of the tube, 

 there is the possibility of the grid becoming momentarily positive due 

 to insufficient negative bias or still worse, the plate may be over- 



