Sanoma
AI-written, synthetic-voice local weather forecasts for radio
Finnish media group Sanoma airs 78 extra AI-voiced weather forecasts a day on two radio stations
Problem
Sanoma's radio stations had no localized weather forecasts for small regions. Producing them by hand was not practical.
Approach
Azure OpenAI's GPT-4o writes five-sentence forecasts from Finnish Meteorological Institute data. A synthetic voice built with Azure Custom Neural Voice, trained on about 750 utterances, reads them three times a day for 26 regions.
Results
Takeaway
Sanoma tested listener attitudes before launch. A follow-up survey found 70% said the synthetic voice suited news fairly or very well.
Categories
- Industry
- Media and entertainment
- Business function
- Production and qualityMarketing and sales
- Innovation type
- Product and service innovation
- AI technology
- Generative AI
- Adoption
- Vendor solution
- Evidence
- Vendor's case study
Sources
Source summaries
Microsoft explains how Sanoma generates and voices local forecasts for 26 regions on two radio stations. It reports 78 added forecasts a day and listener survey results.
Full details are in the original sources.