Some illnesses are more complicated than others, characterized by subtle features, by a slow change in the patient, and are therefore difficult to detect. To help with patient diagnosis and care, Comma Soft is building an app that allows users to log their health data throughout the day – quickly and easily. Voice technology is the key element. To make entries in their health diary, all users have to do is press a big red button and speak: “I’m not feeling so well today. I have a headache and my knee hurts again. I already took a pain pill at lunchtime.” Once entered, this can also be corrected, and the app asks for other agreed-upon data, such as “How’s your blood pressure today?” With little effort, continuous health monitoring can be done. Artificial intelligence is used to analyze the data entered, which is then forwarded to the patient’s doctor in a structured format – but only if the patient concerned releases and sends it. In addition, the app “warns” the user if something is conspicuous.
The health diary app
The health diary app focuses on cardiovascular diseases, so the target population group tends to be older and less accustomed to using computers and cell phones. Therefore, the ease of use of the app is a key issue. Now, feedback from the target group is necessary to assess whether this product would be accepted and is easy to understand or use. Working with two practicing cardiological physicians, Comma Soft plans to identify between 20 and 30 test subjects who will use the app for a defined period of time. The test phase will show whether the AI will correctly recognize all complex medical topics – even spoken in dialect with background noises, for example.
Of course, data protection plays a particularly important role here. Crucially, all data remains 100% with the user! It does not leave the user’s cell phone. Only the interface, in which spoken language is transformed into written language, takes place outside and must therefore be particularly trustworthy. There is no cloud or other central location for the data.
Voice assistance in the hotel: When’s breakfast?
Digital voice assistants have become a natural part of many people’s everyday lives, whether via smartphone, in the car or at home via smart speakers. Jochen Emig, a speech technology enthusiast and founder of ONSEI GmbH, has voice assistants in every room at home. When he was traveling around Germany to visit clients and spending a lot of time in hotels, he missed these practical bots. He found that it was far too complicated to call the front desk or search in the hotel booklet for simple questions like asking for the WiFi password.
The hotels he approached were open to a “digital concierge” – under strict data protection requirements because a hotel room is a home for one night, and all information about guests is very private. These requirements were a perfect fit for the Fraunhofer-led SPEAKER project. ONSEI already developed several prototypes to now equip a hotel with smart speakers.
The focus on the needs of guests and hotel staff is crucial: staff should be relieved and guests should be able to access hotel information and entertainment offers quickly and easily with the help of the digital assistant. Customers should have the best possible experience – with the highest data protection guarantee.
It goes without saying that the complex hotel processes are important: a guest may want to get up at six in the morning, but the next guest be unexpectedly roused from his sleep by the alarm. Starting this fall, customers will be able to use the system. And every hotel visitor will be free to choose whether to join in or switch off the voice assistant.
Live events with live subtitling
SAP had a slew of ideas for potential voice assistance use cases. Two applications were chosen for the SPEAKER project: speech input for chatbots (dialog systems that allow users to chat with a technical system) and live subtitling of selected SAP events. Live subtitles make a valuable contribution in the important areas of diversity and inclusion, as they help make content accessible to a wide audience. In tests that focused on speech recognition in German and English, both use cases produced good results.
Key arguments in favor of a German-made speech assistance platform are data sovereignty and data protection. SAP’s in-house technology was enhanced by linking it with the SPEAKER technology: the Fraunhofer platform meets the highest possible security standards.
SAP’s focus on speech recognition and voice assistance technologies has grown in recent years, and the SPEAKER use cases are relevant both for use in-house and, potentially, for SAP customers.
Here is the first article about the speaker project: Smart voice assistants conquer many industries – but mumbling doesn’t help – International Data Spaces
To learn more about the project, visit: https://www.speaker.fraunhofer.de/en/