878 



NATURE 



[Decemu 



1923 



Loud-speaking Telephones. 



'I" HE Institution of Electrical Fn^inecrR and the 

 * Physical Society of L< * 'iiig 



on November 29 — Dr. Alex;i: -nt 



of bf*' tics, l)cjng in ti ui^tuss the 



proM K'ctcd with " I cii : for Wireless 



and I. ill. I i iirixjscs." The inttims .■■miHed extra- 

 ordinary interest owing to the popularity at the 

 present time of loud - speakers in connexion with 

 oroadcast reception, and several hundreds of members 

 were unable to obtain admission into the lecture 

 theatre of the Institution of Electrical Engineers. 



Prof. A. O. Kankine discussed the general principles 

 involved in the accurate reproduction of sound by 

 means of a loud-speaker. He pointed out that there 

 has been a sudden great public demand for a good 

 instrument, and that the solutions given have practi- 

 cally all been obtained by the methotl of trial and error. 

 Stated roughly, the problem is how best to secure 

 that sounds emitted in one place may be a sufficiently 

 faithful copy of sounds emitted in another. 



The difficulty of the problem lies in the fact that the 

 reproduced sounds must be of considerable intensity. 

 If we are content with feeble intensity in reproduction 

 there are already available sufficiently good loud- 

 speakers. It appears on theoretical grounds that 

 to procure reprcnluction absolutely perfect in the 

 physical sense — as distinct from the acoustical — is 

 not feasible owing to the variety of transformations 

 necessary in practice. 



There is first the amplification of the electrical 

 fluctuations ; in the second place, there is the 

 process whereby the current e.xcites corresponding 

 variations of air pressure ; and thirdly, there is the 

 treatment of the jicrial \ ibrations after they have 

 been created. So iai as the amplification of the 

 electrical waves is concerned, it is found that the more 

 thermionic amplifiers used the more difficult it is to 

 get exact reproduction. The second question, the 

 transformation of a portion of the electrical energy 

 into sound energy, is a very wide one. It may be 

 transformed by electromagnetic, electrostatic, or 

 thermal means, and each method provides a different 

 field for investigation. Lamb has stated that the 

 simple harmonic type of vibration has the pre- 

 eminent position in mechanics because it is the only 

 type which retains its character absolutely unchanged 

 when it is transmitted from one system to another. 

 We can conclude, therefore, that sounds cannot in 

 general be reproduced with perfect precision. All 

 that can be done is to avoid too great changes in 

 the character of the vibrations. Scientifically it is 

 convenient to dissect these vibrations into their 

 harmonic components. 



In aiming at loudness there is a temptation to 

 resort to resonance effects in order to secure it. For 

 example, in the majority of telephone diaphragms 

 there are natural frequencies within the frequencies 

 of the sounds used. The corresponding components 

 therefore get preferential treatment. This can be 

 remedied to a considerable extent by damping the 

 diaphragm, but unfortunately this reduces its general 

 sensitivity. An alternative plan is not to reduce 

 resonance but to confine it to values beyond the 

 upper limit of audibility, or at least as far in that 

 direction as practicable. Another plan is to choose 

 mechanisms of very low natural frequencies, but 

 there are theoretical reasons for considering this 

 method not so desirable. 



The method is used, however, in a device perfected 

 by Siemens and Halske. It consists of a strip of 

 thin metal foil suspended between the poles of an 

 electromagnet as in the Einthoven galvanometer. 



NO. 2824, VOL. 112] 



The plane of the foil is parallel to the ma^notic field, 

 and the incoming telephonic current f1 a 



the foil. This responds by mechanic 

 perpendicular to its y ' ■■ - nd is the < \ 



the ordinar>' teleph<> .im. Its 1 



natural i>eriod is two ;>*»,./.....-., and it is s>.»i.. i^ wjn^-i.iLc 

 without a horn. 



In Prof. Kankine's opinion, horns should, whenever 

 fx)8sible, be dispensed with owing to their rcscjnant 

 character. The ideal sound resonator wrmM ]k» 

 spheric;! ;>e and e.xcite<l in such a 



it impa: ihe neighbouring air svi 1 



fluctuations of pressure. For speech " 

 all room reflections should be damped 

 the sending and at the receiving stations, 

 number of listeners, however, appear to l 

 for echo effects. In his opinion, when loud 

 are used, echoes and reverberations should 1 

 ated at lea.st at one end. In broadcast of>era, where 

 transmission already unavoidably has this effect, 

 the listening room should be draped much in the 

 same way as the transmitting room usually is draped. 



Prof. C. L. Fortescue considered that, with • - ''•■ 



designed valves, no serious distortion was «i 

 amplifier. In the later stages of the ampl..:. ..i.w.,, 

 however, it is necessary to use valves having a con- 

 siderable power output. 



Mr. E. K. Sandeman gave a valuable demonstra- 

 tion of the relative importance of each frequency 

 region in the audible spectrum. By suitable wa\e 

 filters he cut off all the vibrations with frequencies 

 less than 500 transmitted to a loud-speaker. He 

 showed that the effect on the intelligibility of the 

 speech transmitted from another room was not 

 appreciable, but the " naturalness " of the speech 

 was notably impaired. When all frequencies greater 

 than 1700 were eliminated by filters, the speech 

 was scarcely intelligible. This might be considered as 

 the lowest limit for commercial sj>eech transmi-ssion. 

 He proved that the intelligibility was much the same 

 when all frequencies above 1500 were cut off as when 

 all frequencies below 1500 were cut off. Simple 

 and interesting methods of testing speech transmission 

 were given. 



Dr. W. H. Eccles compared the advent of broad- 

 casting news and speeches in the histor\- of the 

 w'orld to the advent of the printing press. WTiether 

 for good or ill, it had come to stay. Loud-sp>eiiktrs 

 could be used to broadcast political speeches to \ try 

 large audiences. He mentioned a case in America 

 where a speaker was plainly audible, by means of 

 these devices, to an audience of 700,000. 



Mr. G. A. Sutherland, who discussed " auditorium 

 acoustics and the loud-speaker," pointed out that 

 uniform loudness is associated in practice with tlie 

 absence of curved walls. Curved walls alwavs 

 produce main and subsidiary foci, and are . 

 to good acoustics. More satisfactory h' 

 likely to be obtained by distributing an duun . e 

 into a number of small rooms with a loud-speaker 

 in each than by attempting to accommodate thcin 

 all in a large hall. The presence of an audience is 

 very effective in reducing reverberation. A s.ire 

 indication that a room is suffering from excessi\e 

 reverberation is given when increasing the loudness 

 of the sounds increases the distortion. \\'hen a loud- 

 si)eaker is too rich in higher-pitched notes the presence 

 of a large audience has a corrective effect. 



Mr. S. G. Brown gave a successful reproduction, 

 by means of his " Frenophone," of a portion of an 

 opera that was being broadcasted by 2LO, the 

 London Broadcasting Station. This instrument has 



