Comment from Peter Saxon.
Almost six years ago, I wrote an article titled: 4 out of 5 jobs most likely to be lost to robots, are in radio.
That prediction was based on research from Oxford University that suggested that around 47 per cent of the workforce in the U.S. would be displaced by artificial intelligence or robots of some kind within the next 20 years.
American technology writer, Shelley Palmer, identified the top five jobs that he believed will be the first to go. Four of them exist in radio.
Middle Management; Salespeople (Ad Sales, Supplies, etc.); Report Writers, Journalists, Authors & Announcers; Accountants & Bookkeepers, all inhabit the corridors and studios of broadcasters.” Doctors were the fifth group on Palmer’s list – the only occupation not applicable to radio.
Attrition is already well underway in terms of middle management and accountant/bookkeepers. As with the banks that have shed local managers, cloud technology and online conferencing, has enabled networks to centralise and rationalise many of the tasks that middle management once performed.
As for on-air staff, the likelihood of announcers being replaced by AI anytime soon, seems farfetched. On the other hand, Journalists and Newsreaders… not so much.
Here’s two reasons why.
1. Providing a local 3-minute news bulletin at the top of the hour is proportionately an expensive proposition. For a small regional station, that 3 minutes can cost as much as the rest of the hour to produce.
2. Qualified journalists (along with broadcast technicians) are incredibly difficult to find and recruit – especially for regional markets.
These two factors combined tend to focus management minds on exploring alternatives. Enter the AI developers, keen to offer solutions.
Already, some online mastheads, like the venerable Washington Post, are releasing an increasing number of their articles with an audio option so users can listen to the news while driving or playing solitaire. Here’s a sample:
Obviously, the system needs some work, but so did sat-nav when it was first introduced as a $3,000 option on motor vehicles 25 years ago. I can tell you that in just the past nine months since I started listening to these audio articles, they’ve already improved significantly.
It is not hard to imagine that the kind of advanced text-to-speech technology used by Adthos could sample the sonorous voice and style of the multi-award-winning newsreader Steve Blanda, (below left) who nearing retirement, may agree to sell the rights to his voice for a tidy sum.
Couple that artificial voice with the growing number of AI copywriting tools such as ChatGPT and it’s easy to guess where all this might be going.
As this segment on last week’s Media Watch shows, and as ChatGPT’s website admits, it still has some work to do before it is industry ready. But with such a big potential market, these AI tools will be perfected sooner than you think.
The biggest hurdle for these AI copywriters is that they can’t gather breaking news that doesn’t already exist somewhere on the net, or has been fed to them by a service such as Reuters.
Regardless of how good AI gets, it will always struggle to beat humans for originality. Those newsreaders that can inject spontaneous personality, whose voice and style is instantly recognisable to the listener, who like, Gold104.3’s Patrina Jones, can step out of the newsroom and chat with the show’s host – in Pat’s case Christian O’Connell – those newsreaders will always find a home in the major competitive markets where “average” won’t be good enough.
However, come the day when AI can write and tell jokes that make humans laugh, for me, life as we know it, will be over.
Peter Saxon
If you listen to some videos on youtube, narration is performed by an artificial voice. The sound quality is more authentic than the demonstration by of an reading by a synthetic voice as demonstrated in ths article of a reading of the Washington Post.
To illustrate pay attention to the sound of this voice from a youtube video. Listen only to the first few seconds:
https://www.youtube.com/watch?v=9IhBmF8sl5M&t=9
It is the synthetic voice of Brian with a British accent. This website is a text-to-voice site that lets you type in text and save your work as an mp3 file.
https://readloud.net/english/british/1-male-voice-brian.html
What these synthetic voices don't have is the property to read text with feeling and expression.
Like Brian, the article mentions that it there is the potential to make a synthetic voice of Steve Blanda.
However, will the AI-based newsreader or voiceover be able to express all the nuances of voice expression and know when to express those feelings through the voice?
I cannot imagine that or an AI system can have the feeling of a voiceover artist like Mr Lofty Fulton's voice. Each advertisement featuring the artist is unique.
An AI-based voice will sound perfect but will be bland because of the lack of variation inbeing able to be fully expressive.
Thank you,
Anthony, of there is a lack of spices ifvthere's no variety, Belfield, in the land of the Wangal and Darug Peoples of the Eora Nation