An automated procedure for detecting human frequency-following responses to voice pitch

Abstract

The frequency-following response (FFR) to voice pitch has been widely examined in research laboratories and has demonstrated its potential to be transformed into a useful tool for patients with hearing, speech, and language disorders in the clinic. During the past decade, many aspects of the FFR have been reported. The presence of such a response, however, still relies on subjective interpretation of the observer. Aside from a recent study reporting two algorithms for detecting such a response, there has been limited number of studies reporting the development of an automated procedure for FFR. The purpose of this study is (1) to develop an automated procedure that utilizes the statistical properties of the temporal and spectral energy distributions in the recorded waveforms and (2) to explore the effectiveness, accuracy, and efficiency of the automated procedure and compare them with those obtained from conventional algorithms and human judgments.