Other research projects

The Effect of Hearing Loss on Sound Texture Perception

Oliver Scheuregger, Research Assistant

Sound textures offer an ecologically valid means of probing the auditory system, whose statistics are shaped by auditory processes. Previous studies have already explored perception of sound textures in normal hearing listeners, however no work has investigated how hearing impaired listeners perceive textures. This project will be a direct extension of his master's project, in which Oliver measured sound texture identification and discrimination performance in normal, impaired and aged listeners. The goal is to investigate if sound textures may be used to better understand and diagnose differences between normal hearing and hearing impaired listeners, with the ultimate goal of aiding in the development of new compensation strategies.

Integrating the visual in AVIL

Kasper Duemose Lund, Research Assistant

Kasper is working on the integration of virtual reality in the Audio Visual Immersion Lab (AVIL). As of now, the AVIL does not have any permanent setup for providing visual stimulation to subjects participating in perceptual experiments. With this HTC Vive based virtual reality implementation, researchers have the framework for exposing subjects to 3D visual environments corresponding to the played back audio environments of the AVIL. A full synchronization of the existing audio engine and the new visual integration will be included. Audio-visual scenarios can therefore be presented in a highly controlled way. Conclusively, a protocol for constructing audio-visual perception experiments using this system will be developed.

Binaural speech segregation in noisy and reverberant environments using deep neural networks

Philippe Gonzalez, Research Assistant

The recent success of deep neural networks (DNNs) has substantially elevated the performance of speech enhancement systems. Despite this success, the major challenge of learning-based approaches is the limited generalization to unseen acoustic conditions. The goal of this project is to develop a binaural DNN-based speech enhancement system and to systematically investigate the general applicability to unseen room reverberation, unseen interfering noise sources and arbitrary target speaker locations.