Written material could give radio producers a route into new audio formats, from literary serials to narrated essays and educational programmes. But adapting a book or article is not simply a matter of converting words into speech: the project needs a clear audience, a reason to exist and an editorial approach suited to listening. Isabel Wiese, chief marketing officer at CaptionCast Media, argues that AI-assisted production could help broadcasters explore these possibilities, while leaving creative decisions with people.
CaptionCast’s service is designed to turn written content into produced audio, using synthetic voices alongside production processes, according to the company’s website. For broadcasters, that could mean considering a short series drawn from an author’s work or essays that complement a station’s cultural coverage. These remain ideas to test rather than guaranteed formats: their appeal would depend on the source material, the intended listeners and the care taken in adaptation.
The sound of the finished programme would matter more than whether a single generated line seemed convincing. Pronunciation, pace, consistency between voices and the relationship between speech and sound design all shape the listener’s experience. Wiese’s proposed approach keeps editors and producers involved in selecting texts, choosing voices, directing performances and checking the final mix. The European Broadcasting Union has similarly highlighted AI’s potential in broadcasting alongside the need for human oversight to protect accuracy and editorial integrity.
A small pilot could help stations judge whether the approach is worthwhile. Producers might begin with one cleared text and a defined format, then track preparation, direction, corrections and quality-control time. Listener responses would offer another measure of success, beyond the speed of production. The EBU Academy’s training on AI in live radio also places practical uses within a wider framework that includes ethical questions, transparency and internal policies.
Rights and audience trust need attention before production begins. The US Copyright Office explains that permission may be required to make or authorise a new version of a copyrighted work, while the application of fair use depends on the circumstances; its guidance is specific to US law. Broadcasters should therefore establish what permissions apply to the text and any voices or sound used, taking account of the relevant laws and licences. UNESCO’s World Radio Day guidance recommends telling audiences when a voice is AI-generated, through measures such as credits, announcements or programme descriptions.
For CaptionCast, the opportunity is to help broadcasters investigate more ways of bringing written work to audio, not to remove editorial judgement from the process. A limited trial could show whether a particular format serves listeners and fits a station’s identity, as well as how much work it demands. The wider industry discussion, including the European Broadcasting Union’s account of AI in public-service media, points to a balance: technology may support production, but people remain responsible for the choices and standards that shape what audiences hear.
Source Reference Map
Inspired by headline at: [1]
Sources by paragraph:
- Paragraph 1: [2]
- Paragraph 2: [3]
- Paragraph 3: [2], [7]
- Paragraph 4: [4]
- Paragraph 5: [5], [6]
- Paragraph 6: [7], [2]
Source: Noah Wire Services