Tikalon Header

Bruce Baker and Minspeak

June 22, 2020

Children have always enjoyed toys that talk. The electronic talking toys of today are far in advance of the mechanical talking toys of my childhood, but the mechanical toys were still fun. I remember one pullstring toy that a younger brother of mine had. Pulling a string powered an internal mechanism like a gramophone to give phrases like "The cow goes, Moo," in which the Moo sound was an actual cow's mooing. The toy was much more fun when I discovered that sharply striking the toy after the word, "goes," produced a different animal sound; e.g., "The cow goes, Quack-Quack-Quack."

As I wrote in a recent article (A Voice from the Crypt, March 16, 2020), Charles Wheatstone (1802-1875), Inventor of the eponymous Wheatstone bridge, produced a bellows-operated "speaking machine" in 1837 that included a vocal tract, and also a tongue and lips. This was an enhanced version of the Wolfgang von Kempelen's speaking machine developed in the period 1770-1790. Kempelen is more famous for his mechanical Turk, a supposed chess-playing automaton that was actually a hoax.

A reproduction of Wolfgang von Kempelen's speaking machine

A reproduction of Wolfgang von Kempelen's speaking machine.

The bellows was activated by a weight attached through a pulley, a mechanism that's just barely visible in this image.

(Wikimedia Commons image by Fabian Brackhane)


Kempelen strived to simulate the complete physiology of the vocal tract by using a bellows as the lungs, a reed as the vocal cords, a simulated mouth, and a simulated throat and nasal cavity. One mouth assembly consisted of a wooden box with hinged shutters that acted as lips. Kempelen's speaking machine was capable of speaking recognizable phrases. Such speaking devices in which the mechanism of human speech is emulated perform what's now called articulatory synthesis.

An interesting variation of articulatory synthesis is to use parts of the human vocal physiology to create a unique manner of musical speech with a talk box. A typical talk box has a plastic tube that directs amplified sound from a musical instrument into the musician's mouth. The musician can then vocalize the instrument, and this vocalization is captured by the same microphone used to record singing. The talk box is typically used with an electric guitar, since the rich harmonic content of that instrument produces quality vocalizations. A good example of a talk box is in Peter Frampton's, "Do You Feel Like We Do."[1]

People who have had a laryngectomy have been able to speak again using a device called an electrolarynx. This device offers a substitute for the vocal cords to provide an audio excitation source for the pharyngeal cavity. The invention of the transistor allowed Bell Labs researchers to create a compact device for this purpose in 1959, as shown in the figure.[2-3]

Figure two from US Patent No. 3,072,745, 'Unitary artificial larynx,' by Harold L Barney, January 8, 1963.

Figure two from US Patent No. 3,072,745, "Unitary artificial larynx," by Harold L Barney, January 8, 1963. This circuit is a simple two transistor relaxation oscillator driving a one transistor class-D amplifier. The patent describes "An artificial larynx comprising, in combination, a pulse circuit for producing a train of pulses in the frequency range of approximately to 200 pulses per second, with a pulse duty factor of approximately five percent, an electroacoustic transducer responsive to said pulses for introducing a tone into the pharynx of the user..." (Via Google Patents.[3])


As the example of Stephen Hawking (1942-2018) shows, people with amyotrophic lateral sclerosis (ALS, commonly known as Lou Gehrig's disease, named after the celebrity baseball player of the New York Yankees who succumbed to it), are able to communicate effectively using a modern technology called augmentative and alternative communication (AAC). Developers of this technology have their own scientific journal, Augmentative and Alternative Communication. Hawking and such technology have been instilled in popular culture by television appearances of Hawking and his speech-generating device, notably in Star Trek: The Next Generation, The Big Bang Theory, and The Simpsons.

One of the pioneers of speech-generating devices, Bruce Baker (1943-2020), died on May 7, 2020, at age of 77. Baker created Minspeak™, which is likely the most widely used system for augmentative and alternative communication.[5] Although Minspeak™ is implemented by computer, Baker was not a computer scientist; rather, he was a linguist who was influenced by his knowledge of hieroglyphic languages to craft this particular system.[4] This approach was a significant departure from the word-based user interface for such devices in the 1980s.[5]

Middle Egyptian hieroglyphs for the adjective 'sweet' and 'elder'

Middle Egyptian hieroglyphs for the adjective "sweet" (left) and the symbol for "elder" (right). Sweet is rendered as a spelling of the word, "bnrj," while "elder" is represented by a symbol that looks like an old man. (Created using Inkscape from the unicode symbols U+130C0, U+13216, U+1308B, U+131CB, U+131DC, and U+13017.)


Baker devised Minspeak™ while working on a Ph.D. degree in linguistics at the Middlebury Language Center (Middlebury, Vermont) with a background in Greek and Latin.[5] He interviewed people with disabilities in searching for a dissertation topic, and he discovered that the AAC systems of that time were linguistically primitive.[5] He saw that such systems did not take advantage of the artificial intelligence techniques being used for language translation. The user interface for such systems were more like a typewriter, and they were difficult to use.[5] Baker's Minspeak™ system used hieroglyphics, rather than an alphabet, and the icons of his system could be sequenced in ways that simplified the expression of complex ideas.[4]

The first version of the Minspeak™ system, which had about 40 icons arrayed on a keyboard, was programmed by one of Baker's former studentss, Kenneth Smith, a mathematician and computer programmer. Minspeak™ was implemented on an AIM-65 computer with a Votrax SC01 voice synthesizer (The AIM 65 was my first laboratory computer, used to automate a liquid phase epitaxy system programmed using FORTH).[5] It was presented at a 1983 meeting of the American Speech-Language-Hearing Association.[5] After so many decades, Minspeak™ is still a commercial success, with a website at minspeak.com.

A portion of figure two from US Patent No. 4,661,916, 'System for method for producing synthetic plural word messages,' by Bruce R. Baker, Richard D. Creech, and Kenneth W. Smith, April 28, 1987.

A strange juxtaposition - God and gallbladder.

A portion of figure two from US Patent No. 4,661,916, "System for method for producing synthetic plural word messages," by Bruce R. Baker, Richard D. Creech, and Kenneth W. Smith, April 28, 1987.[6]

In that era without touch screens, actual keys were used, encoded with "... an alphabetical letter... a portion of the human anatomy and a proper name."

(Via Google Patents.[6])


References:

  1. Peter Frampton, "Do You Feel Like We Do," YouTube Video, July 31, 2018. The talk box effect starts at 7:25.
  2. H. L. Barney, F. E. Haworth, and H. K. Dunn, "An experimental transistorized artificial larynx," The Bell System Technical Journal, vol. 38, no. 6 (November, 1959), pp. 1337-1356, DOI: 10.1002/j.1538-7305.1959.tb01591.x.
  3. Harold L Barney, "Unitary artificial larynx" US Patent No. 3,072,745, January 8, 1963.
  4. Linnea Crowther, "Bruce Baker (1943-2020), invented Minspeak for nonverbal people's AAC speech devices, Legacy.com, May 7, 2020.
  5. Bruce R. Baker, "Minspeak™ History," Spring 1988, Minspeak™ Website.
  6. Bruce R. Baker, Richard D. Creech, and Kenneth W. Smith, "System for method for producing synthetic plural word messages," US Patent no. 4,661,916, April 28, 1987.
  7. Minspeak™ Website.