Development of an Online System Aimed at Improving Listening Comprehension for Elderly People by Emphasizing Consonants

Abstract

The difficulty in recognizing what others say gets more when they speak rapidly and/or with poor articulation skipping some of the consonants. Because such damaged speech requires additional cognitive load for recognizing words. Elderly people with hearing loss have more difficulty even for clear speech because they lack the capability of hearing the high-frequency component of speech signal. We conducted a preliminary experiment to see the efficacy of modifying acoustic features such as amplitude and temporal resolution together with boosting high-frequency component of consonant parts. The result demonstrated that emphasizing the acoustic features of the consonant parts of input speech significantly improved their audibility. Therefore, we have developed an online hearing-aid system using a high-speed computer that detects consonant parts from the input speech and emphasizes them based on the result of the preliminary experiment. Consonants are categorized depending on the acoustic characteristics, and the category determines the method of speech emphasis.