ABSTRACT
The objective of this study was to assess the effectiveness of a bone conduction (BC) communication interface in two-channel speech recognition tasks. Twenty-four listeners participated in a speech-on-speech masking task to assess the effectiveness of a two-channel BC listening device designed for speech perception in a two-channel communication system. The task involved listening to either one target sentence presented randomly to each ear (monaural condition) or two sentences (one target, one competing) presented simultaneously to separate ears (dichotic condition) via a BC headset or air conduction (AC) headphone set. Each participant listened to two trial sets during the study. In each trial set either twenty individual or twenty pairs of sentences were presented to listeners who were asked to respond by selecting the appropriate words from word lists presented on the 54computer screen. The speech signals were sentences from the Synchronized Sentence Set (S3) test which enables simultaneous presentation of one target (T) sentence and up to three competing (C) sentences during a test condition. Results of the study indicate that listeners using the BC headset perform significantly better in the monaural condition than in the dichotic condition; however, from an operational standpoint, listeners still had acceptable performance under both conditions. A comparison of the AC and BC systems under the same dichotic conditions indicated no significant differences in performance of the two types of systems. This implies that the BC system is as effective as the AC system in two-channel speech-on-speech masking tasks.
