Speech processing using adaptive auditory receptive fields
Ashwin Bellur, Mounya Elhilali · Proceedings of the International Symposium on Auditory and Audiological Research · 2017
The auditory system exhibits a remarkable ability to adapt to its listening environment, driven both by sensory-based cues and goal-directed processes. Here, we focus on the role of attentional feedback in facilitating processing of speech sounds in presence of nonstationary noises. We examine a theoretical formulation for retuning of cortical-like receptive fields to enable robust detection of speech sounds in presence of interference. The framework employs modulation-tuned filters aimed at emulating tuning characteristics of neurons at the level of auditory cortex. This bank of filters is then modulated based on goal-directed feedback to enhance separability between the feature representation of speech and nonspeech sounds. We hypothesize that this retuning procedure results in an emphasis of unique speech and nonspeech modulations in a high-dimensional space. We discuss the implications of this retuning on the fidelity of encoding speech sounds in presence of seen and novel noise conditions, and discuss implications of such plasticity in facilitating listening in challenging acoustic environments, hence opening the door to adaptive and intelligent audio technology that can emulate the biological system.