An interactive user interface for selecting optimal listening conditions

Abstract

An audio-visual interface was developed to allow listeners to hear a range of different conditions and convey their preferences. Feasibility for selecting optimal listening conditions was assessed for eight normal-hearing listeners. The visual display was a 4-by-4 matrix with the x and y axes representing two different manipulations of the frequency-gain characteristic. For a speech-in-noise condition, the axes represented low-frequency gain and broadband gain. For a filtered-speech condition, the axes represented high-pass cutoff frequency and bandwidth. Subjects altered the amount of processing applied to the ongoing speech (or speech in noise) by moving a screen pointer from cell to cell. Selections were compared to percent-correct scores for lists constructed from the same speech items (nonsense syllables), and to articulation indices (AIs). It took approximately 2 min to select, and selected conditions typically had AIs within 0.10 of the highest AI in the matrix. A potential application is determining settings for nonlinear hearing aids. [Work supported by NIDCD.]