A current Radio World ebook explores the use of artificial intelligence in radio broadcasting. This is an excerpt.
Radio.Cloud is the developer of a cloud-native automation system and a certified partner of Amazon Web Services.
Account Director Andrew Scaglione joined the company in 2022; he is former director of affiliate operations for G Networks/Gen Media Partners and also worked in affiliate relations for Envision Networks/Sun Broadcast Group.
Radio World: What do you consider the most important trend in how artificial intelligence technology is being used in radio broadcasting and audio applications?
Andrew Scaglione: The ongoing development of large language models has been game-changing for artificial intelligence in the scope of radio production and broadcasting. In the simplest version, AI can be used for inspiration. It’s a starting point for talent to generate ideas for on-air shifts and more. When you dive deeper and combine LLMs with generative and cloned voice technology, you’ll see that AI voice tracking has improved greatly over the last 12 months.
RW: Describe the technical use of AI in your products or software.
Scaglione: With the industry-wide lack of on-air talent for night shifts and weekends, we incorporated artificial intelligence into the Radio.Cloud automation software in several different ways to solve this issue.
The most notable is our AI voice tracking feature Voicetrack.ai. This allows radio stations to pick a voice (cloned or AI-generated), choose topics like music, news, weather, etc., and save voice track presets in their hourly clocks. Based on these presets and our complex prompt engineering, we automatically generate AI voice tracks each day when the playlist is created. Because we’re adding more prompts every week, this technology can even be used to run a fully automated 24/7 station. Best of all, each voice track is set on the ramp/intro of the upcoming song.
RW: How widely will radio broadcasters deploy AI to replace live talent?
Scaglione: We do not see AI replacing live talent in radio. We view AI as a tremendous tool to help improve the day-to-day workflow of hosts, programmers and other radio station employees. And where AI’s true benefit comes into play is keeping local news and the station image front and center in off hours when no one is staffing the station in-person, such as night shifts, overnight shifts and weekend shifts.
RW: What responsibility does a broadcaster have to inform its listeners that AI-based technology is being used on the air?
Scaglione: Some AI technology has been in use for years in radio. For example, we use artificial intelligence to mix and segue between songs on our client’s playlists. In instances like this where AI is streamlining technical operations, I don’t believe it’s necessary to announce that to your listeners.
In generative AI/voice content instances, I think each broadcaster should deliver some kind of announcement that at least part of the program is produced using AI. Trusting your local media is more important than ever. Playing off AI voice tracks like humans is deceptive, but telling your audience this in advance gives them the information needed.
Somewhere in the middle of those two examples lies another great usage for AI: producing local tags or short local imaging reads for network or syndicated content. With where we are currently as an industry, even if there’s just a few 2–3 second reads that are produced with a cloned version of the host’s voice I think it’s vital to announce that to listeners.
RW: How will the development of generative AI tools change how radio managers and employees do their jobs?
Scaglione: This allows managers and employees to be more creative both in how they use AI and in how they choose to incorporate it into their on-air product. As AI develops, having “Ask ChatGPT” or “What Would AI Do” type segments for morning shows is a fun way to tell your listeners that you’re aware of AI and keeping up on trends. It’s also a great tool for generating ideas to then mold into your own for on-air content.
The other side to generative AI is using AI voice tracks and content in a specific daypart where no on-air talent is available or hard to attract. This is another tool in the toolbox for managers to decide where best to employ it. Ideally, to enhance the product and promote the current hosts.
RW: Is the potential impact of AI in radio being overstated?
Scaglione: Absolutely not. We’re so early in the AI game that I don’t think any of us can anticipate exactly what the industry will look like in five or 10 years. But we know that AI is only going to impact radio more. There are tools already available to streamline radio sales and marketing insights, tools to create and produce custom content with a few clicks, voice cloning tools and much more. The key to keep in mind is that while some of these tools are already great, they’ll only improve over time.
RW: Give an example of a broadcaster using your AI-based tools.
Scaglione: One of our German customers, Radio Galaxy, recently launched a completely AI-generated one-hour weekend chart show on 12 stations. All affiliates have their own localized version of the show including different music depending on what’s trending on the Spotify, Shazam and Apple Music charts for specific regions or cities. We take the main list of songs from the local music director or program director, then use AI to fill in the gaps each week depending on the charts. The AI host even does a recap of the top three songs at the end of the show in an automated hook promo format.
