354 BELL SYSTEM TECHNICAL JOURNAL 



Automatic Measurement of Magnitudes of Received Speech and Noise 

 Even when the intensity of the wave reaching the transmitter of 

 the system tested is controlled, variations may occur momentarily in 

 the operation of certain types of carbon button transmitters. Such 

 variations affect an articulation test in several ways. The reproduced 

 speech may be momentarily greater or less than the average as to 

 magnitude, and possibly may even suffer momentary changes in fre- 

 quency composition. The receiving end transmitter may vary in the 

 amount of room noise which it picks up and conveys to the listener's 

 ear by the sidetone path. Both transmitters may occasionally con- 

 tribute burning noise. The combined effect of such variations is 

 sufficient to make it desirable to measure them in order to state 

 accurately the conditions prevailing in the tests. Also, when the 

 variations are large, it is sometimes desired to correct the test results 

 to an average level of speech and noise. 



Because of the rapidity with which the testing is carried out auto- 

 matic equipment is used to make the measurements. This has been 

 arranged to measure two quantities, the average magnitude of the 

 speech wave on the circuit and the average magnitude of the noise 

 reaching the receiver. The time at which the measurements take 

 place is under the control of a timing commutator which regulates all 

 of the automatic equipment. While the test sentence is called the 

 apparatus measures and records the speech magnitude. This ordi- 

 narily is measured at the receiver terminals, but if the noise magni- 

 tude is comparable with that of the speech wave the measurement may 

 be made at the input to the trunk. After the test sentence the appa- 

 ratus measures and records the magnitude of the noise at the receiver 

 terminals. At the conclusion of a list the average readings, through 

 interconnection with the data recording equipment, are automatically 

 typed with the analyzed data. 



In addition to recording the average magnitude of the noise through- 

 out a list of sentences it is frequently desirable to obtain data on the 

 distribution of noise magnitudes about the average. Such informa- 

 tion is valuable for explaining unusual variations in the recorded 

 average values. This distribution is obtained by a series of electro- 

 magnetically operated counters which count and record the number 

 of noise magnitudes occurring in each 2 db interval over a 30 db range. 

 No automatic means of making a typewritten record of these values 

 have been provided at the present time since they have been used only 

 as a check on the operation of the testing equipment. 



Room noise leakage under the caps of the observers' receivers is 

 also an important factor affecting articulation results. Since it may 



