Vox ex Machina: A Cultural History of Talking Machines

by Sarah A. Bell

How today’s digital devices got their voices, and how we learned to listen to them.From early robots to toys like the iconic Speak & Spell to Apple’s Siri, Vox Ex Machina tells the fascinating story of how scientists and engineers developed voices for machines during the twentieth century. Sarah Bell chronicles the development of voice synthesis from buzzy electrical current and circuitry in analog components to the robotic sounds of early digital signal processing to today’s human sounding applications. Along the way, Bell also shows how the public responded to these technologies and asks whether talking machines are even good for us.Using a wide range of intriguing examples, Vox Ex Machina is embedded in a wider story about people—describing responses to voice synthesis technologies that often challenged prevailing ideas about computation and automation promoted by boosters of the Information Age. Bell helps explain why voice technologies came to sound and to operate in the way they do—influenced as they were by a combination of technical assumptions and limitations, the choices of the corporations that deploy them, and the habits that consumers developed over time.A beautifully written book that will appeal to anyone with a healthy skepticism toward Silicon Valley, Vox Ex Machina is an important and timely contribution to our cultural histories of information, computing, and media.

256 pages, Paperback

Published September 24, 2024

Reviews

[Karla Kitalong]

I learned so much, even related to my own hearing impairedness, which wasn't Sarah's goal. Some of the technicalities went over my head, but skimming those didn't detract from my understanding. I loved all the pop culture references and analyses.

[Linda · Rating 4 out of 5]

A lay reader may struggle with the lengthy, highly academic introduction to this book. There’s a good deal of technical content in the first three chapters, which set the stage for understanding the need for synthesized speech, exploring the linguistic aspects of how human vocalizations are formed, and the historical attempts to create machine-made vocalizations.

For me, perhaps because of my age, my interest was piqued when I reached Chapter 4; Speak & Spell, 1978. Speak & Spell was an educational toy developed by Texas Instruments to essentially relieve parents and teachers of the drudgery of repetitive learning drills in a way that allowed children to autonomously explore language. It was “the first widely available consumer electronics device to feature digital voice synthesis.” Its popularity demonstrated the possibility of widespread use of personal computers within the home and in the educational sphere. It also primed a generation of young users to feel comfortable in front of a digital screen and receptive to synthesized speech.

The following chapters explore synthesized speech and its real-life connection to human beings we have known, like Stephen Hawking and Roger Ebert. We learn how the human voice is critically representative of “the self.” And we learn how the marketing successes of synthesized speech in the gaming industry have propelled the technology forward to ever more realistic outcomes.

Bell also raises the red flags that trouble sociologists and historians. There is the very real concern that turning to digitized voices with such human traits will lead to isolation. But at the same time, can those voices alleviate the loneliness of the elderly and certain special needs individuals? How will the use of synthesized voices used in marketing propaganda impact consumers? As synthesized speech becomes more sophisticated, what are the risks of historical revisionism that is solidly anchored by the skillful ability to put words in the mouths of historical figures? And do we really need to have sophisticated talking machines replacing the labor of human beings who need the jobs that will be lost to further mechanization? As fascinating and exciting as we find our digital assistants on the phone and in the car and around the house, what repercussions may blind-side us in the future?

“To live effectively is to live with adequate information.” Norbert Wiener

[Eric · Rating 4 out of 5]

I’ve never been particularly interested in voice synthesis (or NLP), so I picked up what I assumed to be a narrow but informative history of the field. What I got instead was something much broader and more fascinating: a look at how computing’s role in culture has been perceived and imagined throughout the 20th and 21st centuries.

Since the earliest days of computing, (the dream of) machines that talk has reflected a broader anthropomorphism of computers, and has existed alongside anxiety about their growing role in education, the economy, and the home. There seems to be a strong consensus today that “something feels different” about the recent wave of LLM breakthroughs—yet the language we use to discuss the expected disruption, and the cognitive biases that lead us to underestimate scale while over-generalizing assumed intelligence, are truly cyclic. They are as old as computing itself.​​​​​​​​​​​​​​​​

Told through the lens of several key developments in computing, as well as through both academic and pop cultural references from the same periods, the book paints a fascinating picture of our complicated feelings towards the machines that augment us, but always seem to just fall short of truly replacing us.