Faber’s American Speaking Machine
About two years ago the newspapers announced with some pomp that a speaking machine had reached Paris, which far surpassed Vaucanson’s duck, and which would attract general attention. Unfortunately the invention was not in the first instance brought forward with any scientific authority, and was soon relegated to take a place among the curiosities exhibited by conjurors. In a country so essentially critical and sceptical as France, there are always those who profess incredulity, and who will even resist evidence, and it was asserted that the machine only spoke because its exhibitor was an able ventriloquist. This is an old assertion which has lately been made with reference to the phonograph. Some scientific papers echoed the absurdity, and the speaking machine was so discredited that it is now unnoticed, although it is a most ingenious and interesting conception. When will our country be cured of the error of denying everything without due examination?
Since we ourselves only judge of things after having seriously considered them, we think it just to vindicate the truth as to Mr. Faber’s machine, and this can only be done by an exact description of it.
As I said in the last chapter, there is a great difference between the production and the reproduction of a sound, and a machine like the phonograph, adapted for the reproduction of sound, may differ essentially from a machine which really speaks. In fact, the reproduction even of articulate sounds may be very simple, as soon as we possess the means of stereotyping the vibrations of air necessary to transmit these sounds; but in order to produce them, and especially to emit the complex vibrations which constitute speech, it is necessary to set in motion a number of special organs, fulfilling more or less exactly the functions of the larynx, the mouth, the tongue, the lips, and even the nose. For this reason, a speaking machine is necessarily very complicated, and this is precisely the case with the machine we are now considering. Such a machine is not now made for the first time, and the Academy has lately been reminded of a speaking head which was in the possession of the philosopher Albertus Magnus in the thirteenth century, and which was destroyed by St. Thomas Aquinas as a diabolical invention.
Mr. Faber’s speaking machine was exhibited two years ago at the Grand Hôtel, and may now be seen in the room adjoining M. Robert Houdin’s theatre, the same room in which Mr. Giffard exhibited the telephone. It consists of three distinct parts: 1st, of a large bellows worked by a pedal, which produces the currents of air necessary for the production of sounds, and to some extent acts as the lungs; 2nd, a vocal instrument, consisting of a larynx accompanied by diaphragms of various forms to modify the sounds, of a mouth with caoutchouc lips and tongue, and of a tube with an outlet somewhat resembling the nasal cavities; 3rd, of a system of jointed levers and of pedals, terminating in keys like those of a piano.
The most interesting part of the machinery, of which we represent the principle fig. 73, is the vocal apparatus, which involved the severest study of physics in order to succeed in the production of articulate sounds. It consists, first, of a rather thick caoutchouc tube, within which there is a kind of whistle L, as in a clarionet. The whistle consists of a small caoutchouc cylinder with a longitudinal slit, and before this is placed a very thin ivory plate lined with caoutchouc. This plate is fixed at one end to the cylinder, and deviates slightly from it at its free end, so as to permit the current of air projected from the bellows S to penetrate between the two parts, and to cause the vibrations in the ivory plate necessary for the production of a sound. The extremity of the caoutchouc cylinder is closed on this side, and is fitted to an iron rod t which comes out of the pipe, and is connected with a system of bars, corresponding to the keyboard of a piano, by which the force of sounds can be regulated. This force depends on the width of the opening between the tongue and the cylinder.
The whistle, which plays the part of the larynx, is necessarily placed opposite the opening of the bellows, and a sort of tourniquet M is fastened to the opening itself, which is able to move on certain conditions, so that it may produce the rolling sound of r. This is done by fastening before the opening a diaphragm in which there is a somewhat wide and long slit, and this slit can be almost closed by a little bar of the same size M, revolving on a transverse axis which supports it by its centre. In its normal condition, this little bar is kept in a slanting position by cords attached to the keyboard, and the air ejected by the bellows readily traverses the slit in order to reach the larynx; but two dampers are fastened to the rods which transmit movement, with which the cords just mentioned are also connected. On lowering the notes of the key-board, the passage of air is contracted, and the little plate begins to oscillate and to press against a band of leather, producing by its vibration an action similar to that produced by the cricket. This little tourniquet only begins to act when the dampers are lowered by a pedal worked by the hand; and this is also the case with the iron rod t, which modifies the acuteness of the sounds passing through the larynx.22
Below the larynx tube, which is only five centimètres in length, there is another pipe G, also of caoutchouc, which terminates in a spherical cavity connected with the outer air by a caoutchouc tube I, slightly raised, and closed by a valve, of which the movements are regulated by a pedal worked by the keyboard. When the valve is open, the sounds emitted through the larynx are somewhat nasal.23 The larynx communicates with the mouth through a square funnel-shaped pipe, to which six metallic diaphragms D are fastened; the diaphragms are placed in a vertical position behind each other, and have indentations on their lower end, which are intended to diminish more or less the orifice for the current of air, and to impede its passage with greater or less force. The diaphragms, which we represent separately fig. 74, are connected with the keyboard by jointed iron rods t, and, for the emission of most articulate sounds, several of the diaphragms are moved at the same moment and at different heights. We shall return to this subject.
The mouth consists of a caoutchouc cavity O, somewhat resembling the human mouth, and forming a continuation to the channel we have just described. The tongue C, likewise modelled on the human tongue, is placed within the mouth, and connected with two jointed rods t, t, fastened to its two opposite ends, so as to enable the tongue to raise its tip, or touch the palate, in obedience to the notes of the keyboard. The lower caoutchouc lip A can also be more or less closed, according to the action of the keyboard on its special rod. Finally, a circular metallic piece E, following the shape of the mouth, is placed above the upper lip, with a small opening in it to admit of the pronunciation of the letter f.
The keyboard has fourteen notes, of different lengths, producing the following letters when lowered, a, o, u, i, e, l, r, v, f, s, ch, b, d, g. The longest corresponds to g, and the shortest to a. There are two pedals below the g note and those of b and d, corresponding with the opening of the tube which produces nasal sounds, and to the rod which regulates the opening of the larynx, and this makes it possible to obtain p, t, and k from the notes b, d, g. The mechanical effects produced by lowering the different notes in succession are as follows:—
1. The a note moves the first five diaphragms.
2. o also moves these five diaphragms, but varies the pitch, and closes the mouth a little.
3. u does the same, only further closing the mouth.
4. t moves a single diaphragm, raises the tip of the tongue, and opens the mouth more widely.
5. e moves six diaphragms, throws the tongue further back, and opens the mouth still more.
6. l moves five diaphragms, sends the tongue against the palate, and further opens the mouth.
7. r moves six diaphragms and the tourniquet, lowers the tongue, and somewhat closes the mouth.
8. v moves five diaphragms, almost closes the mouth, and keeps the tongue down.
9. f lowers the circular appendix of the upper lip, and almost entirely closes the mouth.
10. s moves three diaphragms, half closes the mouth, and half raises the tongue.
11. ch moves three diaphragms, keeps the mouth half closed, and further lowers the tongue.
12. b moves five diaphragms, closes the mouth, and keeps the tongue completely down.
13. d moves six diaphragms, keeps the mouth three parts closed, and raises the tongue a little.
14. g moves five diaphragms, keeps the mouth three parts closed, and the tongue completely down.
m is produced by lowering note b and opening the valve of the pipe which gives nasal sounds.
n is obtained by lowering note d and opening the same valve.
h is obtained from note s by lowering the pedal which acts upon the larynx, and half closing it.
Since the other letters of the alphabet are compound sounds, they can be produced by combinations of the preceding letters.
Although the words pronounced by this machine are distinct, they are spoken in a uniform, drawling tone, which might, I should have thought, have excluded the idea of imposition. Some of them are indeed far from distinct, yet the results are not less remarkable; and when we consider the amount of study and experience which must have been applied to the combination of all these arrangements, it seems surprising that physicists have not given more attention to such an interesting machine.
As for the mechanical execution, it is impossible to admire too highly the simple and ingenious manner in which all the complicated movements of the different vocal organs have been connected with the keyboard, of which the mechanism has been so calculated as only to produce the precise action of the organ which is required for any given effect. For this purpose, the notes of the keyboard regularly increase in length, so as to produce at a single touch different mechanical effects on the rods which act upon the mechanism; and since most of the notes are required to react simultaneously on the whole mechanism, the rods which transmit the movement are fastened to a series of jointed levers which cross the notes of the keyboard at right angles. Pegs of different length are fastened to the notes at this junction, so as to produce the simultaneous action of the different organs of the machine.
The public will believe that the assertions of ventriloquism are unfounded when I add that I myself have made the machine speak.