bernleckie

And Sky, the UK’s biggest pay TV service with around 23 million customers, has been giving away free 6-month Apple TV+ trials and heavily promoting Ted Lasso and other shows of high UK interest (like Slow Horses) on the front page every viewer sees. 

Info: https://www.sky.com/pages/vip/Apple-TV-six-months

One one hand, Apple is right about the scale of a real problem. Millions of books don’t have audio versions. Making even a basic acceptable one by hand is time consuming, costing hundreds of dollars at minimum wage. You want better than minimum wage talent? Of course you do, so that would cost more. Would that cost ever be recouped by most authors or publishers? Sadly not. Publishing is a risky enough business anyway, scarcely able to reward most authors for the time and effort it takes to write good books - adding extra costs for accessibility simply won’t always happen. 

So there’s the potential to solve a real problem by lowering the cost of audiobook production. 

Have you heard how good AI voices can be now? If you’re thinking Siri is state of the art, you’re way off. What we commonly hear from our phones is at least two generations behind the best available. Neural processing has vastly improved the modelling and training of decent voices. Try them for yourself at sites like play.ht - you can walk through the options from cheaper to more premium, and then imagine what’s being worked on that isn’t yet accessible with a public API. 

I haven’t heard Apple’s genre-optimised voices, so can’t compare, but I wouldn’t complain about anything generated by these voices if it were the only audiobook available: https://play.ht/ultra-realistic-voices/

Here’s a business plan worth thinking about: at published rates, it would cost less than $600 a year to use those voices to generate audiobooks - less than the human cost of one or two at best. How about making dozens or hundreds with AI instead? They’re cheap enough to make it viable to have an audio version of every book, even if it only sells one copy. 

That’s the up side. The down side includes less work for narrators - except that there’ll always be demand for premium versions, and there’s no reason not to make more than one for a book that sells. 

The real down side might be our underlying anxiety about AI starting to do things we thought could only ever be done by humans. That’s a challenge to our senses of security and identity, and having the prospect over the horizon of bots writing and delivering half-decent creative content in overwhelming quantity sounds dystopian for sure.

But that’s where we are. If we don’t want the tech to develop, we should stop buying and using it. This seems like the wrong forum to suggest that!