A Biologically-Inspired Neural Network for Sound Lateralization
Perustieteiden korkeakoulu | Master's thesis
Unless otherwise stated, all rights belong to the author. You may download, display and print this publication for Your own personal use. Commercial use is prohibited.
Acoustics and Audio Technology
Master’s Programme in Computer, Communication and Information Sciences
48 + 0
AbstractModels of the auditory periphery have been used to study human perception of sound. These models contribute to the improvement of listening devices such as hearing aids, cochlear implants, headphones, and loudspeakers. Physical models that explain the mechanics of the auditory periphery have proven to be accurate and slow. Due to the computational complexity of physical models, the use of deep learning to create biologically-inspired and easily parallelizable models has gained traction. In this thesis, we build on the recent advances in biologically-inspired neural networks and previous work in auditory modeling to model sound source localization in the frontal horizontal plane. This thesis presents CoCoNetti, a biologically inspired neural network that leverages the strengths of both CoNNear and Count-Comparison models for azimuth prediction in the frontal horizontal plane. CoCoNetti achieves accurate azimuth prediction while following trends in human perception.
Thesis advisorLladó, Pedro
deep learning, auditory modeling, sound localization, spatial hearing