Who do I think I'm fooling?
Misadventures in digital voice cloning.
A little idea took ahold of me a few years ago after I watched the Roger Ebert biographical documentary, Life Itself. As you probably already know, treatment for an aggressive cancer left the eminent film critic without the power of speech. Such were the limits of technology at the time, Ebert had to rely on stock voices on his MacBook whenever he needed to type out the things he intended to say aloud. Still, Ebert knew how to hammer out a beautiful sentence and so he found comfort in how his keyboard could continue to serve him, not just in the everyday world but also online, where his presence on comment sections and discussion forums was no different to anyone else.
I wonder how he would have taken to the digital cloning technology that is available today? One prominent example of AI voice cloning came to light recently when a voice cloning company created a new voice for Val Kilmer, who had gone through a similar illness to Ebert, though he was still able to speak, albeit in a noticeably raspy fashion. Kilmer compared his natural voice to a pirate in one interview. Still, the digitally created Kilmer voice went beyond uncanny, sounding indistinguishable from the voice that we know from his many movie appearances. Similarly, it has been revealed that the voices of the characters of Luke Skywalker and Darth Vader were created from digitally cloned voices of the actors James Earl Jones and Mark Hamill. It should be noted that nothing terrible had befallen the two actors short of time itself. Their voices had changed over the past forty-five years much like everything else had. In the case of Hamill, the voice didn't match the uncanniness of the deepfaked/ dee aged likeness of his onscreen character, but, for me at least, it still seemed to be lacking something. It seemed to lack all those chaotic, ineffable nuances that we can pick up from natural human speech.
So, as you have probably already inferred, these stories left me with a deep desire to clone my own voice. You have also probably inferred that the voice that is speaking to you right now, if you are listening to the narration for this post, is itself that very digital clone that I wished to create. Hurray for me, the immortal, digital Niall that will carry on waffling long after the meat Niall has gone the way of all inferior, carbon based life forms. Death to all humans, ha ha ha ha hahaha ha ha haaaaaaa!
I had a few reasons for cloning my voice, some were practical and others were more existential. I live in a cramped flat in South London with my wife and kids. It's noisy inside and outside and I don't really enjoy recording when other people are in the same room as me. I know that people enjoy the narration, even when it's not necessarily for accessibility purposes. So, a cloned Niall seemed like a practical solution to the technical and temporal obstacles that keep me from being the fine honed content machine that I secretly know I will never be.
But I also know that the desire to clone my voice only came about because of my exposure to stories of people who lost their ability to speak and reached out to digital technology to keep their voice alive. I was even slightly seduced by outsourcing different aspects of my work to an algorithm that might learn to write poems that are similar to my own and then read them out. The main use for this would be a kind of ideas board or dice role, something truly random to spark ideas for my next proper piece of writing. But I have to admit, I liked the idea of my creativity being able to carry on in some form after the end of my life. Likewise, if some misfortune rendered me unable to create and share my work, then there would be the opportunity for an automated version of my work to carry on.
If you are listening to the digital narration then you have already worked out that it's knee-deep in the dankest swamp of the uncanny valley. Those high profile digital voices needed a lot of hands-on airbrushing and fine tuning from the engineers and artists that worked on them, from the initial creation of the voice to the final nips and tucks to the finished audiotext. And even then, the voices couldn't recreate all the nuances and flourishes that we would expect from a trained, professional actor. But this voice? Well, I think my loved ones and acquaintances can rest easy if they suspect they might be defrauded by my digital doppelgänger. Admittedly, it sounds better than it did a few months ago, even recreating my glottal stops and effs in place of tee-aitches.
I’m sure that it would be the same breed of uncanniness with an algorithm that scraped through all of my available output and churned out its own versions, there would be some noticeable similarities but the majority of it simply wouldn’t convince. This is not because I am some special literary genius but rather because I am human. There are so many quirks and irregularities within our humanity that accompany the aspects of our personalities that are more predictable and therefore easier to imitate.
These ideas came to a head the other day when I wrote one of my clock card poems about whether my consciousness could be replicated digitally:
"It's not so much that I object to having a digital snapshot of my brain activity uploaded to a synthetic body on the occasion of my death - it’s more that I don't believe it would be me. It would be an act of digital taxidermy on a fragment of self that puffed out of existence the moment it became apparent. I confess that the prospect of my ghost being an idiot robot blundering about for eternity could be a bit of a hoot. Why spend all that money on robot people when we could have robot owls?"‘
I’m quite proud of the term “digital taxidermy”. I really think that’s what digital transhumanism is to me. '
I recently finished a wonderful cyberpunk narrative adventure game, Citizen Sleeper. The central character is a stowaway robot with the uploaded consciousness of a person who is sleeping in a cryochamber back on earth. However, this is no oligarch futurist’s method of escaping death, the sleeper on earth has agreed with a corporation to use their digital doppleganger for indentured servitude. These robots are known as "sleepers". The game begins with the robot waking up on a space station having escaped from its corporate overseers. Right from the off, the sleeper is rebelling against its oppressors, including the living human that they are derived from. I played through the game with a focus on that idea, that the sleeper was an entity unto itself and not a continuation of the life of an original human consciousness . I was happy to get an ending that rewarded me with the sleeper living from its own sense of selfhood and meaning rather than one it had inherited. You should definitely play this game, it really got its hooks into me in a good way.
This is pretty much the only reason I would, if the technology was ever available, upload a digital snapshot of my consciousness. Not as a means to immortality but to instead donate aspects of my mental life to another entity in the same way that I would donate my bodily organs to someone after my body grinds to a halt.
At the time that I became a baby atheist around the turn of the century, I certainly entertained ideas of extending my life beyond my death in ways that seemed scientifically feasible. It really didn’t take long for me to ditch the idea of a digital afterlife but I held on to more subtle versions. For instance, I was attracted to the idea that Douglas Hoffstadter conveys in his book, I Am A Strange Loop, about how we carry little copies of the people we love around in our heads, so that when our master copy winks out of existence, there are still smaller, lower fidelity versions of us living on within the minds of the people that knew us. Much like the proverb that someone only dies the moment that everybody in the world no longer remembers their name, I like the idea of my annihilation being a series of ripples, each a little weaker than the last as opposed to the sudden erasure that follows the final breath. I was always looking for the security that ideas of the afterlife gave me when I was raised as a Catholic as a child.
At some point, I started asking myself the honest question, “If I died right now, what qualities of this very moment would carry on and which would vanish forever?” For instance, and I apologise for being so morbidly specific, if I keeled over in the beer aisle of my local supermarket, then the supermarket would still be there, all the other shoppers would still be there. My trolley might tip onto it’s side with the wheels spinning but it would still be there. My body would still be there too, barely different to how it was the moment before I took my final breath. Even aspects of my identity and the stories of my life would continue in the remembrance of others. When I really ask myself what would truly be gone in that moment I become aware of the ineffable aliveness that I am feeling right now. Not my name or the story of the person I am but the very act of being conscious. When I witness this in this particular context, I feel incredibly alive. Notions of what I could be after this aliveness ends feel nonsensical and without any true value. This feeling never fails to manifest whenever I ask this question and I appreciate the concepts of the memento mori or Lorca’s Duende, why so many cultures seek to maintain an awareness of death within the core of their living.
I paid for this cloned voice, you know. I took out a subscription at the end of last year. However, I won't be renewing it and I am happy to consign my digital doppelganger to oblivion. The very thing that I fancied would outlive my own ability to speak is therefore just as ephemeral as my tonsils. Those that know my voice will hear it when they read my words. When it reaches a point where nobody remembers my voice, then the words might not hang on too long after that. Who knows? It really is no concern of mine.