Stephen Hawking, the famous cosmologist and theoretical physicist, passed away last week at the age of 76. Like much of the world, I was fascinated by both his accomplishments and his iconic voice. Having spent my career in the Text-to-Speech field, I was lucky to have met Hawking once, working on his synthetic voice then and again years later, with Nuance.
The famous cosmologist and theoretical physicist Stephen Hawking passed away last week. Apart from his scientific contributions, he was also a role model for people living with a disability. “Concentrate on things your disability doesn’t prevent you doing well, and don’t regret the things it interferes with,” he said in a 2011 interview with the New York Times, “Don’t be disabled in spirit.” Millions of people became familiar with his synthetic, computer-generated voice, which he began using after losing the ability to speak in 1985.
The synthetic voice he used for more than 30 years was generated by a circuit board named CallText 5010, made by a company called Speech Plus, which is now part of the Nuance family. Hawking originally owned three copies of CallText, but one board broke after it fell to the ground. Concerned that the hardware would break or fail to work in the future, Intel, who had started providing him with a PC and technical support, wanted to replace his hardware synthesizer with a software version. They didn’t want to risk leaving the scientist without his voice again.
When I was a postdoc at the Oregon Graduate Institute working on speech synthesis, I was contacted to help with the project. In the following months I borrowed Hawking’s spare CallText board and recorded 2,000 speech sounds called diphones with it (synthesis by concatenation of diphones was the dominant text-to-speech technology in those days.) When Professor Hawking was in Oregon for a lecture a few weeks later, I presented the new voice to him at his hotel room in downtown Portland. I connected it to a loudspeaker so he could hear the sample sentences I had prepared from his lectures read aloud. After a few minutes of silence (during which Hawking was typing), came the reply: “I like it. But more importantly, will my wife like it?”
That same evening, I attended his public lecture. I remember feeling a personal connection as he was presenting, and a sense of privilege to have been a part of his story, no matter how small. Hawking ultimately continued to use his original circuit board synthesizer during public appearances. After all, it was that voice that the world had come to recognize as the iconic Stephen Hawking; the new implementation just didn’t sound quite the same. Hawking’s synthesized voice was as much a part of him as our natural voices are of us.
In late 2017, we revisited a project first discussed a few years early: The project entailed Nuance working with Professor Hawking and his team, agreeing to provide him with a version of the source code of his TTS voice, which we had stored in an archive. The goal was to transition Hawking’s system to a modern, software version while maintaining the authenticity of his original voice. Unfortunately, he passed before we could complete our work together.
Hawking once wrote on his website, “I try to lead as normal a life as possible.” Ironically, there wasn’t all that much normal about him. His wit was unmatched. He once acknowledged that although he had a PhD, “women should remain a mystery.” He was a brilliant physicist, a renowned cosmologist, a respected professor, and a prolific author. He won countless awards and held thirteen honorary degrees. Hawking’s professional success was matched only by the strength and depth of his personal relationships. “It would not be much of a universe if it wasn’t home to the people you love,” Hawking once said. He will be deeply missed by his family, friends, colleagues, and the countless people he inspired in the universe he helped explore. It is a gift to all of us that even when he could not speak, Stephen Hawking never lost his voice.
Johan Wouters is Director of Product Management for Text-To-Speech at Nuance. Earlier roles included Head of Research at SVOX, which was acquired by Nuance in 2011, and Director of TTS Language and Voice Development at Nuance. Johan obtained an MSc in Electrical Engineering from the University of Leuven in Belgium and a PhD in Computer Science from the Oregon Graduate Institute in the US. He is passionate about the speaking and singing voice in technology and arts. Some of his favorite podcasts are This American Life, a16z, and Interne Keuken (in Flemish).