At ISRG, our machine-listening research involves the modelling of human auditory perception, and codifying these models using machine-learning approaches. This research is relevant in applications dependent on acquiring detailed, acoustically-informed analyses of complex sound-signals. One of the primary outputs of this research will be its deployment on handheld devices for mechanical-fault detection (acoustic diagnostics). Future Industry 4.0 implementations will target the development of intelligent, in-situ machine interfaces that will possess capabilities to accurately resolve auditory cues presented by human-users in acoustically-chaotic and reverberant environments.
When implemented on a large-scale network of sensors, our machine-listening research enables detailed sound-scene analysis and classification of urban and rural soundscapes. This technology functions as a backbone for developers and local authorities to design mobile/wearable apps that would benefit from detailed, accurate descriptors of objects and events based on acoustic-signature emissions.
Our perceptual models contribute to an emerging technological trend - hearables. These devices, which focus on providing auditory services and enhancements in a wearable form-factor, require a level of acoustic precision that can only be provided by in-depth computational auditory-scene analysis. Our research in this area also involves evaluating EEG data of users presented with auditory stimuli. Our future goals in this regard include collaboration with colleagues in neuroscience to link brain electrical activity with hearable control parameters.
One of our primary research activities is determining perceptual and cognitive constraints when users interact with immersive, multimodal computers. This work involves understanding cross-modal interactions between visual and auditory stimuli; working memory limitations; and the characteristics inherent in attention mechanisms and other cognitive processes. The primary output of this research is the development of frameworks that guide interface developers through the process of presenting multidimensional data to users utilising both visual and auditory sensory streams.
If you are interested in our research, or would like to talk about collaborating with us, please contact Dr Flaithri Neff by ObscureMyEmail.
Site developed using Rapidweaver: https://www.realmacsoftware.com/rapidweaver/
ISRG icon adapted from an SVG file created by Rapidweaver stacks developer, Big White Duck: https://www.bigwhiteduck.com
Machine Listening & Perception Background Image: photo by Felix Berger on Unsplash
Autonomous Sound-Scene Classification Background Image: photo by Jakob Owens on Unsplash
Active Listening & Hearable Design Background Image: photo by Mateo Abrahan on Unsplash
Immersive Multimodal Computing Background Image: photo by Samuel Zeller on Unsplash
Research Staff Background Image: photo by DXL on Unsplash
Postgrads Background Image: photo by Robert Penaloza on Unsplash
Partnerships Background Image: photo by Jaredd Craig on Unsplash
Industry Links Background Image: photo by Vincent Delegge on Unsplash
Unsplash images: https://unsplash.com