The Long Quest to Make Machines Talk
As early as the 1700s, scientists built speaking machines that, through various combinations of reeds, bellows, and pipes, simulated the sounds of the human voice. Then, in 1939, Bell Labs debuted its “VODER” (Voice Operation DEmonstratoR), which simulated the physics of speech with electrical circuits—in a decidedly robotic tone.
Today, of course, we have the much more realistic voices of Siri, Cortana, and Google Now, which combine almost-human speaking capabilities with enough artificial intelligence to converse and answer our questions. But speech synthesis still isn’t entirely convincing—it lacks much of the emotion, melody, and creaky flaws of real human voices.
Brad Story, a professor of speech, language, and hearing sciences, walks us through the history of talking machines, and computer scientists Alan Black and Rupal Patel talk about making computerized voices more personal and engaging.
Brad Story is a professor in the Department of Speech, Language and Hearing Sciences at the University of Arizona in Tucson, Arizona.
Alan Black is a language technologies professor at Carnegie Mellon University in Pittsburgh, Pennsylvania.
Rupal Patel is founder of VocaliD, a company that builds personalized synthetic voices, and professor of communication sciences and disorders, and computer and information science, at Northeastern University in Boston, Massachusetts.