A current Radio World ebook explores the use of artificial intelligence in radio broadcasting. This is an excerpt.
In 2018 WorldCast Systems released its first technology with artificial intelligence for FM transmitters, SmartFM, to optimize RF power based on the audio content, in real time. The company says benefits include lower power costs, reduced cooling, less maintenance, longer transmitter life and a smaller carbon footprint.
David Houzé is a product manager for WorldCast.
Radio World: What’s the most important recent or current trend in how artificial intelligence is being used in radio and audio?
David Houzé: We are still at the beginning of artificial intelligence in radio broadcasting but the trendiest use of AI is mainly in the studio and concerns the script generation for shows, news or weather reports.
In our daily life, most of us already use ChatGPT and other alternatives to simplify content creation based on ideas.
It is logical to reproduce this new use at the job. I do like the fact that radio organizations tend to automate time-consuming tasks to focus on their core activities.
We will see more and more AI in the studio with playlists being automatically generated based on music trends, speech transcriptions for articles, as well as script generation and so forth.
On the transmission side of things, which is where WorldCast steps into the picture, there are also AI innovations broadcasters can turn to.
RW: How does the AI work in your products?
Houzé: Technically speaking, the AI integrated into SmartFM has been trained with power correction based on psycho-acoustic models tested across listeners’ panels. These models have been regularly updated based on data collected on the field.
The technology is integrated into our Ecreso transmitters and is also available for other FM transmitter manufacturers looking to improve their transmitter TCO or total cost of ownership.
RW: The deployment of transmitters by Uplink Network in Germany is massive. But what’s the most important thing a radio technologist should know about it?
Houzé: The most important advantage is the energy saved with our transmitter.
Most of the FM transmitters on the market, including Ecreso transmitters, have an efficiency of 72% to 76%. With SmartFM technology, the energy consumption can be reduced by up to 40%.
As an example, the consumption for a 74% efficiency transmitter running at 10 kW is around 119 kWh per year. With SmartFM, the figure drops down to 70 kWh per year, resulting in energy cost reduction.
RW: Are there other notable installations of SmartFM?
Houzé: Apart from Uplink and the German market, SmartFM is being significantly deployed in Morocco, with some recent customers with 50+ transmitters from 1 kW to 5 kW.
Their electricity costs are being drastically reduced with the activation of SmartFM. Of course, there are also deployments in other parts of the world, but these are the two most notable.
RW: Who developed SmartFM? How did it come about?
Houzé: The idea of SmartFM was invented by Daniel Werbrouck, the former president of Audemat, now advisor at WorldCast Systems. He began by analyzing the propagation of audio on FM carriers, then studied the perfomance of FM receivers, transmitters and more globally the complete chain.
His in-depth knowledge of FM broadcast over the last 30 years enabled him to think about a disruptive approach to solve an unsolved problem: how to increase the transmitter efficiency while we have almost reached the limitations of hardware.
Then our R&D team at WorldCast transformed the idea into psychoacoustic and RF studies, then mathematical modelling.
RW: In what other areas of the radio technical airchain might AI be deployed in the future?
Houzé: One area of interest for WorldCast is the fault detection and automatic correction on any equipment of the chain.
Radio tends to limit its investment in maintenance and human resources. Our monitoring tools collect many key information parameters about the equipment’s health. It will be interesting to get, via artificial intelligence, automatic health reports with maintenance recommendations as well as automatic configuration adjustments based on the broadcasting condition.
Over the last decade, we have been gathering millions of pieces of data on our FM transmitter range, with more than 15,000 products deployed over the world.
All of this information has enabled us to identify points for reliability improvements but also create a first step in the development of a maintenance algorithm that can be embedded in the equipment itself as well as in the monitoring equipment. We hope to have more information to communicate around this in the future.
RW: How will the development of generative AI tools change how radio managers and employees do their jobs?
Houzé: I am not the best person to answer this as I mainly focus on the broadcast transmission side and not the generation of content.
However, generative AI is an evolution affecting what radio organizations do. I am sure it will facilitate the creation of articles, podcasts and program schedules, and AI will be an assistant to radio employees who will be able to focus on their core activity instead of administrative tasks such as reporting or generating texts.
Information about how SmartFM works is available in an article on the WorldCast website.