The world as we know it has been shaken up in ways we can barely comprehend. The markets have collapsed, our liberty curtailed, and the shelves laid waste in shops worldwide. Airlines, retailers and restaurant groups are teetering on oblivion, and all physical sporting competition has been suspended. As I approach my 50th birthday, I cannot recall a period of greater uncertainty in my lifetime. The rules are being ripped up, and certainties are crumbling in the wake of Covid-19.
Throughout all this turmoil, another revolution gathers pace. The advance of Artificial Intelligence continues relentlessly. I confess until now I have watched AI developments with some detachment, but that has changed in the past few days. It's easy to regard technological developments as welcome progress - until it threatens your livelihood directly.
I blithely imagined that moving from one industry on life-support (radio and TV broadcasting), to a growing one - audiobooks and on-demand audio - would at least give me a few year's grace before technology would force me to adjust once more. I'm almost shaking as I write this - but that's no longer the case.
Despite the last-minute cancellation of the London Book Fair, London-based tech company, Deep Zen, http://deepzen.io/ managed to launch their new AI driven narration system, with three frighteningly realistic samples of their code at work. The company say that most of the main audiobook sales platforms are satisfied with the quality of the end results, and will stock productions featuring the AI driven performances of avatars such as Lauren and Gordon.
The results are impressive. Smooth and believable, at first listen they sound like recorded human narration. Listen closer, and the idiosyncrasies and dynamics that a real narrator can bring are missing; command of the moment, holding the pause or changing pitch and pacing to create impact on the listener are simply not there. In other words, the areas where a narrator uses their judgment to push the boundaries of performance are not there.
The clarity and accuracy of the AI read are impressive though, and I can imagine that many authors will willingly sacrifice texture and depth for speed of turnaround, and no doubt a cheaper production process.
One thing that is very noticeable is the lack of characterisation in the samples. One of the main areas I pride myself in is working hard to create convincing voices for every character in the novels I'm commissioned to narrate. It's a huge part of the art and very time consuming - and unless AIs start drawing on a bank of performance samples and data to cover every possible accent, age group, gender and attitude, I can't see this tech overtaking those of us who can conjure up characters for all scenarios.
For example, my partner Rebecca and I attended the Self-Publishing Show Live last week at London's Queen Elizabeth Hall. Hundreds of fellow independent authors, publishers and narrators gathered to reflect on collective successes, and peer into the future of the industry. It was great to share in the energy and optimism in the hall. During the lunch interval, I took Rebecca down to the neighbouring National Theatre, and the sublime bookshop there.
It's a treasure trove for those of us who love the English language and the power of performance, and has a fabulous range of books on the craft of performance. I came away groaning under the weight of books explaining how to approach accents, or how to stretch and train the voice. An AI can I'm sure absorb data from a myriad of sources, but can it approach the job with the passion that I feel I bring to it? Performance is a combination of things: talent, skill, intuition, experience, technique, and above all - human sensibility. We're masters of conveying emotion, when we allow ourselves to be.
Am I a luddite?
I don't believe so. I notice that Deep Zen are encouraging narrators to offer their talents to widen the range of synthesised performers available. Your AI avatar would generate royalties for the human you. That's tempting I confess, but I remain convinced that there's a way to go before the human performer is replaced by the AI version. At least I hope so, but who can be sure in these uncertain times?