Page 47 - Special Issue

Figure 7. Techmer’s “articulatory score,” inspired by early speaking machines. The caption reads “A physiological transcription of simultaneous articulations by analogy with musical notes. The horizontal lines of notes express the place of articulation, whereas the vertical lines express the simultaneity of articulations, with additional signs denoting individual properties of the articulations and their variants.” Definitions of different signs are shown on the main chart. Two examples transcribing the Chinese word for “ear” (with a tone) and the Khoisan word for
“to love” (with a click) are given to the left of the schematic vocal tract, while a full articulatory score for the German phrase “Ist es wirklich wahr?” is given on the right. Reproduced from Techmer (1880).
vocal tract articulations, simultaneously and sequentially, to produce each single coherent unit of speech. This was also the first system for unrestricted real-time speech synthe- sis, allowing the performer to play any desired sequence of speech sounds based on concatenation of discrete phonetic units defined by the keys. It was the first system to systemat- ically incorporate the synthesis of intonation and performed the first song ever sung by a machine. All of these achieve- ments are quite remarkable for their time and represent the zenith of what was accomplished with mechanical synthesis until very recently.
Techmer’s “Articulatory Scores”
Faber’s “Euphonia” was exhibited across the whole of Europe, Russia, and America for more than 40 years, with hundreds of newspaper accounts of public demonstrations that often also refer back to Mical, Kempelen, and Kratzen- stein. Although these speaking machines are now almost forgotten, they were well-known to the major scientists of the day, not only as public spectacles for entertainment but also as genuine objects of scientific inquiry into the nature of human speech (Lindsay, 1997b).
One of those scientists, Friedrich Techmer, developed an entire theory of articulatory phonetics based on his fascina- tion with talking automata. In his “Habilitationschrift” from the University of Leipzig, published in 1880, he reviewed the last century of progress in speech physiology and pho- netics and attempted to synthesize many of the principles embodied in the speaking machines, with which he was
intimately familiar, into a practical method for phonetic transcription. Techmer (1880) proposed a symbolic alpha- bet of vocal tract constriction types as the fundamental units of speech. Revisiting the analogy between music and speech, he proposed that utterances could be represented in terms of such constrictions, organized according to an
“articulatory score” by analogy with a traditional score in music. Different lines of the score represent independently controllable articulators; single notes represent the acti- vation of specific articulators at particular points in time, whereas chords represent the simultaneous coordination of multiple articulator combinations to achieve particular constrictions. Critically, for the first time, the temporal as well as the spatial coordination of articulatory movements is explicitly represented, and timing becomes an intrinsic part of the representation of speech. The example he gave to illustrate his system, an analysis of the German phrase, “Ist es wirklich wahr?” is partly reproduced in Figure 7.
Techmer’s system can be viewed as the natural extension to speech of Engramelle’s earlier proposal for transcribing individual musical performances. It could potentially have been employed to combine Mical’s programmable cylinder mechanism with Faber’s vocal tract keyboard to yield a fully automated, programmable, mechanical speech synthesizer, realistically simulating the articulations and acoustics of a physical vocal tract to reproduce natural speech.
Unfortunately, popular and scientific interest in mechani- cal speech synthesis largely vanished with the appearance of
Spring 2020S, uSpmemciaerl I2s0s1u9e | Acoustics Today | 147 Reprinted from volume 15, issue 2