In France, in 2019, 18 million calls were made to fire and rescue services, and the average time to process an alert (answering and listening, analysis and decision, dispatch order) was 2 minutes and 18 seconds. Systel, a major telecommunications player for civil protection, has once again placed its trust in CATIE for a large-scale project to make use of this waiting time to reduce response times.
In an emergency, every second counts and can save a life. In this project, CATIE is responsible for automating the processing of calls to civil protection services and analyzing audio conversations so that help can be dispatched as quickly as possible.
The Systel teams provide CATIE with an anonymized database of calls previously recorded by the emergency services they work with. Once these calls are converted with speech-to-text technology, the Algorithms and Data team uses a Big Data approach to search the considerable volume of data collected and stored. It then defines algorithms to classify incidents by category and assign them an urgency score.
Based on the history of interventions over the past 5 years, an artificial intelligence model is trained to predict where incidents will occur in order to anticipate and optimize the positioning of resources.
This work by CATIE could bring major improvements to emergency call handling in the future.
During the call, key information is collected and automatically transcribed into the emergency software using “Macarena”, our speech recognition technology, simplifying the operator’s tasks through a decision support solution.
Likewise, in a crisis, when call centers are overloaded, the system could eventually prioritize calls before operators take over, based on the urgency scores determined upstream and its ability to detect the level of anxiety in the caller’s voice.
This AI-based emergency call automation solution comes with a visualization tool designed for rescue services.
Keywords:
- Civil protection
- Speech-to-text
- BigData
- Data visualisation
- Macarena
- VANIILA
- Prediction
- Incident management
- Artificial intelligence
More information here