EFFECTS OF PHASE DISTORTION 563 



lowering the cut-off of a low pass filter having negligible phase distor- 

 tion from 2400 to 2000 cycles causes an equal change in articulation. 

 Similar considerations show that the effective cut-off for twenty-five 

 5000 cycle low pass filters in tandem is of an order of 4700 cycles, 

 although the attenuation distortion of the twenty-five filters was essen- 

 tially the same as that of one filter. The effective transmitting range 

 of a network is a function therefore of both the phase and the attenua- 

 tion characteristics. 



In addition to decreasing the understandability of speech, phase 

 distortion introduces certain audible effects, as previously discussed, 

 which may be a source of considerable annoyance to the listener. 

 Their noticeableness depends upon the amount of delay distortion and 

 the frequency range in which it occurs. For speech, it was found that 

 one section of the all pass network when associated with a 2400 cycle 

 low pass filter, had sufficiently small delay distortion so as to be just 

 noticeable. This determination was made by alternately listening to 

 speech from the system under two conditions, one, the filter alone, two, 

 the filter with the all pass network. Judgments of which condition 

 contained the network were correct about 50 per cent of the time and 

 wrong about 50 per cent of the time for one section of the network. 

 The total delay distortion at the cut-off frequency In this case, i.e. 

 that due to the filter plus that due to one section of the network was 

 about .006 seconds. When three sections were used the distortion was 

 easily noticed. 



Similar tests with the 5000 cycle low pass filters indicated that some 

 number between five and ten filters in tandem would cause just 

 noticeable distortion and that the distortion was clearly noticeable for 

 20 filters in tandem. The amount of delay distortion at the cut-off 

 frequency for five filters is about .003 seconds, and for ten filters about 

 .006 seconds. 



The above figures depend somewhat upon the attenuation character- 

 istic as the cut-off' frequency is approached, small amounts of attenua- 

 tion reducing the noticeability of the effects. The figures also vary 

 somewhat with individuals depending upon their experience and hear- 

 ing characteristics. 



Tests on piano reproduction when single notes were struck or when 

 a passage of music was played indicated that the distortion caused 

 by twenty-five of the 5000 low-pass filters in tandem was not noticeable. 

 As in the case of speech, it would be expected that the noticeableness of 

 the distortion would depend upon the frequency range in which it 

 occurs. In general, the effects of delay distortion on music are very 

 much less noticeable than on speech which is probably due in part to 

 the more sustained character of music. 



