Radio Broadcast (Nov 1926-Apr 1927)

MARCH, 1927 SPEECH CHARACTERISTICS IN BROADCASTING 485 around the corner for his unpostponable meal. I admire that man. He went the limit. All the broadcasters who have ever unbalanced an orchestra, neglected a gain control, or reduced the plate current of an amplifier to zero by overloading— they are all pikers compared to him. If 1 had the money, and that fellow could be purchased, I'd exhibit him in a cage at every broadcast station in the country, with a sign around his neck reading, "Greater Gall Hath No Man Than This." Abstract of Technical Article IV. The Nature of Language — A Resume of Recent Work on the Physics of Speech and Hearing, by R. L. Jones, Engineering Dept., Western Electric Co., Inc., Journal of the American Institute of Electrical Engineers, Vol. XLIII, No. 4, April, 1924. SINCE the organs of speech are substantially the same in different races, and are capable of emitting only a limited variety of sounds and pitches, the elements of speech show many similarities in different languages. The organs of speech include the lungs, which supply the motor element in the form of streams of air expelled through the vocal passages. This bellows action of the lungs, as far as speech is concerned, is a secondary function, the primary object being the interchange of oxygen and carbon dioxide, without which life cannot be supported. The breath supplied by the lungs passes between two muscular ledges whose tension and separation may be varied, permitting vibration over a range of frequencies— the vocal cords. The tongue and lips shut off or permit the breath to issue, and also have some influence in determining the resonance effects of the mouth, nose, and throat cavities. What we have, essentially, is a system of bellows, vibrators, valves, and resonance chambers, all adjustable with remarkable speed and precision, and controlled by reflex actions which become largely unconscious after speech has been learned. The sounds of speech, as represented by letters, fall into five classes: (a) Pure vowels, (b) Transitional vowels, (c) Semi-vowels, (d) Stop consonants, (e) Fricative consonants. In English, there are thirty-six letter sounds. The production of pure vowels involves vibration of the vocal cords in a manner characteristic with each speaker. There is a fundamental tone, somewhat lower for men than for women, with overtones. The mouth and throat cavities reinforce some of these harmonics, according to the position of the tongue and mouth. For example, the long u sound, as in "tool" is formed with the lips rounded, and the tongue drawn back so as to make the mouth a single cavity resonant at about 300 cycles. This single cavity is used for other sounds of u, o, and a, the resonance peak for broad a, as in "far" being around 1000 cycles, with the tongue no longer much raised, so that the effect of the throat cavity begins to be felt in a double resonance. This double peak becomes pronounced in the short a of "at," where the mouth and throat form connected cavities, with two re-inforced tones between 800 and 1 200 cycles. For the long e sound the resonance peaks are more widely separated, the frequencies being in the neighborhood of 300 and 2500 cycles. The tongue in this case is well forward, affording a large resonance chamber in the throat and back of the mouth, with a small cavity between the tongue and the lips for the higher frequency. Transition vowels or diphthongs are formed in passing from one vowel to another. For example, w is simply u plus a pure vowel. If one pro FIG. I nounces the long sound of u, followed by a pure vowel, such a word as "way" is the result, the w being a characteristic transition vowel. H is simply a forcible expulsion of breath through the glottis (the opening between the vocal cords) preceding a vowel. This letter is classified as a transitional vowel. L and r partake of the nature of both vowels and consonants, and receive a separate classification as semivowels. Stop consonants are characterized by the formation of a stop in some part of the mouth. The sound of the consonant "p" is made by exerting breath pressure against the closed lips, and suddenly parting them, releasing the air. The same motions, plus vocalization, produce "b." The unvoiced stop consonants are p, t, ch, and k. Those involving the vocal cords are b, d, j, and g. The nasal group, in which breath is released through the nose, comprises m, n, and ng. Fricative consonants, as distinguished from the stop group, utilize the rushing or hissing sound of breath passing through an outlet, involving the lips, tongue, teeth, or palate. These are the same organs of the mouth and throat used in producing stop consonants, so that the essential difference is the complete closing and subsequent release for the stops, and an incomplete closing for the fricatives. Similarly, vocal cord vibrations are present in some cases and absent in others. For example, the sound of "i" is obtained by expelling the breath through the outlet between the upper teeth and the lower lip, without vocalization. If the voice accompanies the same procedure, "v" is the result. Jones next describes briefly the equipment for physical analyses of speech. This portion of the paper, and the curve of the energy distribution of speech, or, as Jones entitles it, the "Acoustic Spectrum" of English, are taken from the Crandall-MacKenzie paper on "Analysis of the Energy Distribution in Speech," abstracted in the January Radio Broadcast. As was stated there, the vowel sounds carry most of the energy of speech. The reason for the 200-cycle maximum in the energy distribution curve is shown by some analyses of sung vowel sounds presented by Jones. These show maximum components, in every case, close to 200 cycles, for one particular speaker. The higher frequencies, up to 6000 cycles or more, are carried by the consonants, and, because their importance in determining intelligibility is by no means proportional to their 10.000 3 1000 0.001 £.0001 1 . tttnf* ^ \ \ r> -1 — 1 / p 0.0001 0.000.001 64 256 1024 4096 16, frequency; cycles per second Variation of Minimum Perception (Hearing and Feeling! in the Human Ear FIG. 2 weak energy, cause the most difficuly in high quality electrical reproduction. The paper continues with a description of the mechanism of the ear. The outer ear is essentially a collector of sound in the form of air waves, which impinge on the drum separating the outer from the middle ear. The middle ear contains a mechanical transmission chain of small bones which carry the sound vibrations to a membrane or oval window giving access to the inner ear. Some small muscles in the middle ear have the function of accommodating the mechanism for effective hearing of sounds of various intensities. The inner ear is a delicate and complicated system for converting acoustic vibrations into nerve currents. It is essentially a spiral shell of bone, the cochlea, filled with liquid, and containing the rod-like terminals of the auditory nerves, some 3000 in number, constituting the basilar membrane. These rods appear to respond selectively to vibrations of different frequencies, thus permitting the apperception of pitch. Besides the oval membrane through which the vibrations are received, the inner ear is provided with a round window or membrane which may be bulged out by the liquid in order to relieve excess pressure. Fig. 1 shows the general features of the auditory system as described above. This picture represents my own idea of the anatomy of the ear, and is not given by Jones. We do not know precisely how changes of intensity of sound are detected by the ear: it may be through proportionate agitation of the nerve terminals, or by the width of the band affected on each side of the point of selective vibration. The ear mechanism between the drum and the nerve terminals has a definite vibratory impedance, varying like any other such system with frequency and amplitude. At very high frequencies, say 20,000 cycles per second, the impedance may be so high that no perceptible energy reaches the inner ear. We may expect a varying sensitivity, accordingly, to sounds of varying pitches. This characteristic is shown in Fig. 2, which is part of Fig. 4 of Jones' paper. From this curve, showing just audible sounds at various pitches, we note that the maximum sensitivity of the human ear is at about 3000 cycles, and is fairly constant between about 500 to about 7000 cycles per second. The variation of sensitivity is great; a note of 30 cycles or 17,000 cycles requires a million times as much energy to become just audible than one of, say, 1000 cycles. The region most essential in speech is roughly that in which the ear is most sensitive, pointing to a progressive evolution of both organs to a common basis of operation, now sapiently utilized by the telephone companies in furnishing a 200-2000 cycle band for the conversation of their customers. Sounds may be felt by the ear as well as heard, if the intensity is sufficiently great (about 1000 dynes R. M. S.). The curve of sensitivity of feeling for sounds is roughly the reflection of the hearing curve shown in Fig. 2, it being concave downwards, with a maximum at about 1000 cycles, and dropping off to the limits of audition, which are about 20 cycles on the low end, and 20,000 on the high, for the average person of normal hearing. In other words, very low and extremely high pitched sounds are felt as easily as they are heard. While even the energy required for feeling is not great, that required for audibility is marvellously small. In the favorable region shown in Fig. 2, the minimum audible tone corresponds to a pressure change per square centimeter of about 0.001 dyne. This, Jones states, is about equivalent to the weight of a section of human hair one thousandth of an inch long, which is something like a third of its diameter! (To be continued)