Written by Chris Styles
You probably don’t know Perfect Paul, but you will definitely recognise his voice, although you will probably think of a very specific person when you hear it. Perfect Paul is a creation, a placeholder, better than nothing but no one’s first choice. Perfect Paul is an artificial voice, one of a set of siblings, for those who are unable to speak or vocalise, and arguably the most famous person to adopt one of these voices was Prof Stephen Hawking. Hawking’s distinctive artificial voice became a necessity after his Amyotrophic lateral sclerosis (ALS) slowly decreased the scientist’s motor function. Hawking started to slur his words throughout the ’70s and lost the use of his voice after surgery in 1985. Hawking was an early adopter of voice synthesis technology and with the help of human technologists, he was able to maintain some movement and the ability to speak throughout his life, allowing him a means to communicate the insights of the universe to the world, in part thanks to Perfect Paul.
This family of voices was created by the technology company DECtalk in 1984, and combined a voice synthesiser and text-to-speech technology to allow different types of interfaces to create words and speech. Hawking would originally use these systems with his hands and a clicker, but as his movements became more limited, the technology was able to assist, and eventually worked by detecting the muscle movements in his right cheek. Hawking later worked with IBM systems who were able to improve the system and increase his word output considerably throughout his life, but he stuck with Perfect Paul. Hawking went on to become, arguably, one of the most famous scientists of all time, he became an ambassador for science and the differently-abled, as well as a household celebrity, and his distinctive voice became part of his brand.
Hawking was able to adopt Perfect Paul, making it distinctly his. For many of us, our voice is the part of our literal identity, but for those who are unable to vocalise for a long time there was a very limited choice, in fact, only 8 to choose from in the DECtalk range. With time you may grow accustomed to the voice you chose, but it won’t grow and change with you. For many I imagine it must be hard to truly call it your own, to have to assign yourself a voice which you think best matches you. I can see some people perhaps liking the idea of their presence and personality being personified by “Huge Harry”, but would anyone want their lives tied to the voice of “Uppity Ursula”? Although from listening to samples of all these voices, none are really satisfactory for what you really need a voice to do (especially Whispering Wendy, she can go straight back to the nightmare she came from).
Well, there are now companies that are working towards providing humanistic and real-sounding voices that can closely resemble the natural cadence of the client. One company at the forefront of this technology is VocalID, whose CEO Dr. Rupal Patel walked into an assistive technology conference and heard men, women, children of all ages speaking with the same voice, Perfect Paul.
“We wouldn’t dream of fitting a little girl with the prosthetic limb of a grown man, why then was it okay to give her the same prosthetic voice as a grown man?”Dr. Rupal Patel, VocalID
That is where companies like VocalID step in; instead of just using a stock voice in their assistive technologies, they create something a little more human. By recording and analysing the qualities and characteristics of what vocalisations the client is able to make, they are able to compare this sample to a vast database of donated voices and determine which one would be the closest match. There is one last step of personalisation, the host and the donor voices are then combined to produce a close approximation of what the host would have sounded like. This can sometimes result in a slight degradation of the sound quality, but having a few pops and mumbled contestants seems like a small consolation for something unique to you, not a replacement voice, but a part of you.
The process of donating your voice is a fairly easy process, it may take a few hours of speaking, repeating a few hundred to a few thousand phrases and utterances, just enough to cover the key mouth sounds. These can then be patched together by VocalID software to create any string of words or phrases. So, should we all go out and become surrogates?
This is something I had thought about in the past. However, and this may seem like an oddly composed thought, but I have always had a strange relationship with my own voice. Growing up with speech and language difficulties and issues with my ears, for a long time I didn’t speak, and it was only after the age of 5 that I first started being able to vocalise my thoughts in any specific way. People say I have been catching up on lost time ever since, although I am not really sure I talk more than anyone else, it is definitely distinctive. I’m loud and I’m not really sure I sound like anyone else (forever people have liked to guess where my accent is from and the truth is always a slight disappointment), and for a long time I hated it. I know that this is common discomfort but it seemed like I actually had empirical evidence of how out of place I sounded, and for a long time I would not listen to myself. Although I have slowly gotten over these hangups, people still do say that I sound distinctive and I still slightly wince, but it’s fine. I have definitely developed my own voice. But after all that, am I willing to share it?
Well, the answer is yes, I was. I went to the VocalId website when I first learned of their existence a few years ago and signed the online consent form. After a few more tick boxes I came to the final stage before my submission, an audition. They required a sample of my voice. It was only a few paragraphs of text, but I remember not being overly pleased with the result listening back (but that was an inevitability), but I submitted the recording and I waited to hear back, and I never did. This was a few years ago, and perhaps the recording was not of good enough quality, or perhaps there was another reason, but I never did enquire any further. Perhaps I will try again soon.
I do have some thoughts about giving away an individual part of you, something that you are still using. In a practical sense I’m not sure I like the idea of a private company owning the rights to an almost unique dataset to you, in the same category as a multitude of other personal information, although your voice is not a common biometric, it could be in the future. This is the same reason I am apprehensive about these DNA ancestry websites. I do not particularly like the idea that I may become the product in an unscrupulous transaction, although it is arguable that we have long lost this war, the internet already knows too much. However, there is also a slightly selfish part of my soul that doesn’t like the idea of my voice being attributed to someone else, it’s mine, and I know that is stupid. I kind of feel the same way about my organs, but with them, I certainly won’t be using them when they are taken, and with my voice, I am sure I could get over it. I’m also pretty sure I’ll hear me coming a mile off.