Smart sound sensing
We develop and deploy smart sound and vibration monitoring solutions, tailored to your specific needs.
We offer a complete service, including translating your requirements into formal specifications, the design and realization of hardware and embedded software for data capture and sensor data management, the application of machine learning algorithms for data analysis and computational modelling, as well as the development of web-based front-ends for data visualization and presentation. As a spin-off company of Ghent University, we strive to introduce the latest scientific knowledge into our approach.
Next to the 1/3-octave band spectrum with a temporal resolution of 1/8s, our sensing solutions provide all common noise metrics (LAeq, Lden, percentile levels etc.) on-the-fly at custom time intervals, and allow to assess a wide range of advanced sound indicators, including psychoacoustical indicators (loudness, sharpness, roughness, tonality) or state-of-the-art sound metrics such as the Common Noise Index, with an arbitrary temporal resolution. We can even add your own designed indicator!
Our sensor nodes analyze sound in a way that is highly similar to how humans listen to sound. Through the use of artificial neural networks and state-of-the-art machine learning methods, they can detect and classify sounds according to their source, and in this way provide an idea of how the acoustic environment would be perceived. This can be used to evaluate existing soundscapes, but our algorithms can also be applied to simulated or auralized environments, such that soundscape design measures can be evaluated.
Our technology allows to detect, record, classify and signalize events occurring near the sensor nodes, such as gun shots, explosions or other impulsive events, as well as car/fire alarms, accidents, aggression or panic situations, through a continuous and in-depth spectro-temporal analysis of sound and/or vibration. This makes our smart sound monitoring technology well suited for security applications.
Our sensor nodes can act cooperatively, creating multi-device arrays. Through a combination of multi-sensor time/direction-of-arrival and cross-correlation analyses, the location of sound sources (impulsive, intermittent or stationary) can be estimated. We have a particular experience with the localization of low-frequency sound sources, for which typically large arrays are needed.
Our mobile sensor nodes are equiped with GPS and allow to map the soundscape of extensive areas with great spatial detail, an application that is particularly suited for urban areas, squares and parks. In combination with fixed sensor nodes, they can also be used to refine dynamic noise maps, through a model-based spatio-temporal interpolation technique developed at Ghent University.
Sound contains a wealth of information. We provide solutions to derive information on the local state of the environment on the basis of the analysis of sound/vibration combined with machine learning. Some examples are monitoring the proper functioning of machines based on the noise they generate, or the detection of pavement wear on the basis of tire-road noise (for the latter, see our Roads service).
