by Volker Buddensiek
LISTEN is a new hands-free system for controlling smart home applications that is being developed as part of a new EU project. Partners from Germany, Greece and Italy are working together on a natural language interface to control Internet services as well as functions in intelligent homes.
The goal of the EU project sounds beneficial: Software systems to control temperature, lighting, media and communication will make living within one's own four walls more comfortable and affordable. Smart home systems will be particularly beneficial in the everyday lives of elderly people or people with disabilities. The European Union, however, thinks that controlling such systems is still much too complicated. For this reason, it has initiated a project to implement a simple interface: a person's voice.
Experts from the fields of language technology and signal processing are working together to make this a reality. They are developing a voice recognition system for smart homes that will also allow Internet applications such as search engines, e-mail and social networks to be controlled reliably and naturally using verbal commands. The two key components are a large recognition vocabulary and a suitable arrangement of microphones inside of a building.
"We are optimising our speech recognition system to allow it to control both Internet services as well as smart home functions," Dr Siegfried Kunzmann, director of R&D at EML said. "We will also support multiple languages." The project is currently focusing on English, Greek, Italian and German.
The developers' vision is to enable smart home residents to interact with the system using only their voice. Users will not need to speak directly into a microphone, headset or other electronic device for this. Instead, the developers are planning to implement an omnipresent voice detection system using a wireless network of acoustic sensors. Wherever people are inside of a building, the smart home system will be listening in on them in real time.
The four-year project kicked off on 1 June 2015. The project partners include EML European Media Laboratory GmbH from Heidelberg, which is providing the speech recognition software, the Language Processing and Pattern Recognition working group at RWTH Aachen, Cedat 85 from Italy, and the Greek Foundation for Research and Technology Hellas (FORTH), which is also coordinating the project.