Description:

Speaker

Benjamin Lindquist, Northwestern University

Title

"On Phonemes and Feelings: A Short History of Text-to-Speech"

Abstract

After the Second World War, engineers automated the production of synthetic speech. Computers could now talk. But there was a problem: these rule-based voices sounded, well, mechanical. And it wasn’t clear why. Was simulating human speech too much for midcentury mainframes, or was the problem more profound? Slowly, researchers realized that stripping speech down to its “information-bearing” elements had counterintuitively undermined sonic communication. While historians of computing have long focused on discrete data as the driver of digital history, I forward a counter thesis. Starting in the 1960s, speech scientists challenged the dichotomy of signal versus noise, as they recognized that emotion, spontaneity, and what Roland Barthes has called the “grain of the voice” were inseparable from the semantic content of spoken language. These extra-linguistic—even aesthetic—qualities of speech were not obstacles to the dissemination of information. Instead, elements like human emotion provided a crucial conduit through which auditory “data” travels. In ther words, to effectively deliver data, computers first needed to emulate affect.

To this end, Professor Lindquist's talk recounts the work of computer scientist Noriko Umeda at Tokyo’s ElectroTechnical Laboratory (1960-68) and Bell Telephone Laboratories (1968-78). Focusing on Umeda’s attempts to add spontaneity, specificity, and emotion to artificial speech will narrow the gap between affect theory and the history of computing.

Biography

Before earning a Ph.D. from Princeton University, Benjamin Lindquist trained as a painter at the School of the Art Institute of Chicago (BFA) and Yale University School of Art (MFA). This background has shaped his current interest in the history of computing. Specifically, his work asks how tools and concepts drawn from the world of art influenced early computer simulations of human creativity and the mechanical synthesis of human attributes. At Northwestern, he will work on two book projects: a history of text-to-speech and another provisionally titled “The Irrational Computer.”