Page 46 - Special Issue

Early Talking Automata
Figure 5. The mechanism of Faber’s “Euphonia,” illustrated by du Moncel (1882), showing the bellows (S), the vibrating reed (L), the trill mechanism (M), the moving blocks (D), the jaw (A and B), the tongue (C), the palate (E), the nasal cavity (G, H, and I), and the levers (t) connected to the keyboard; arrows indicate the direction of airflow. Reproduced from du Moncel (1882).
set up the keyboard mapping so that the keys would produce the following basic sounds: a, o, u, i, e, l, r, v, f, s, ch, b, d, and g, which we now recognize phonetically as subsets of vowels, liquids, fricatives, and plosives. One of the pedals controlled nasality and another controlled voicing and pitch, with the last pedal manipulating the bellows. By pressing different keys and pedals either together or in sequence while pushing air out of the bellows, any utterance that could be transcribed in terms of these basic articulations could be played on the keyboard and intonation could be added by continuously varying the pitch pedal. Spring loading in the key and rod mechanisms contributed inertia and damping, resulting in a smooth interpolation between adjacent sounds, approxi- mating coarticulation. When demonstrating his invention, Faber would ask for sentences from the audience in any lan- guage, which he would then play back on the machine after transcribing them phonetically in his head, reportedly with a German accent. In London in 1846, he even managed to get it to sing “God Save the Queen.”
A full description of the machine and its operation was pub- lished by du Moncel (1882), and the only known diagram of the mechanism is reproduced from that account in Figure 5. An illustration of the entire machine is shown in Figure 6, taken
146 | Acoustics Today | Suprminmge2r0201,9Special Issue Reprinted from volume 15, issue 2
from a contemporary newspaper report in Paris in 1877, exactly one hundred years after Mical.
Faber called his talking automaton the “Euphonia” and first presented it in 1841, giving public performances almost con- tinuously for many years until his death in 1864. Afterward, his invention was further improved and exhibited as the “Amazing Talking Machine” by his niece, Marie Trunka, and nephew-in- law, Samuel Husserl, until 1887. Fascinating accounts of the travels and travails of Faber’s mechanical speaking machine can be found in Altick (1978) and Lindsay (1997a).
Faber’s speaking machine is notable for many significant advances in the synthesis of human speech. For the first time, all sound production occurs by modulating the pas- sage of airflow through a single tube shape, just like in a real vocal tract, whereas previous models had to coor- dinate multiple separate mechanisms to create different classes of sound, breaking the analogy between model and physics. Also for the first time, synthesis is controlled by direct manipulation of the vocal tract area function, under physiological constraints inspired by the articulators. Com- pleting the gamut of musical analogies, this was the first keyboard instrument to successfully play speech, and this is scientifically significant because the keyboard mecha- nism encapsulated the need for coordination of multiple
Figure 6. Faber’s “Euphonia,” exhibited at the Grand Hôtel in Paris in 1877 by his niece and nephew-in-law. This is one of the few illustrations of the whole speaking machine, showing the bellows driven by a foot pedal, the box containing the mechanism shown in Figure 5, and the system of rods and levers linking the different articulators to the keyboard. Reproduced from L’Illustration (January 1877).