In case of a call for help or an alarming noise, such as broken glass, quick action is required. However, full-time supervision is not always possible at home or in care. In industrial production or at individual workplaces, there may not be a colleague close enough to call help.
To make people in distress heard in these situations and to comply with applicable occupational health and safety regulations, experts from the Fraunhofer IDMT in Oldenburg have developed sound- and event-based recognizers that can also be integrated into existing emergency call and safety systems.
They recognise different keywords as well as different scenarios, for example when caregivers need assistance. For older people, call for help recognition offers an opportunity to maintain independence in the home environment for as long as possible. At the workplace, in smart cities or public infrastructures, it can enable security monitoring of larger areas or be a supplement to existing feedback systems (motionless detector, dead man's switch).
The call for help recognition also works reliably in noisy environments, for example next to a television or in production environments. With the focus on the lowest possible number of false detections and alarms, it creates real added value in everyday life and safety that can be relied on.
The call for help recognition can be combined with other Fraunhofer IDMT developments in smart sensors and devices. By integrating speech recognition, further dialogues can be connected. For the industrial workplace, call for help recognition can be combined with applications for voice control, documentation or condition monitoring in a communication device, such as a hearable or headset.