Factors such as semantic and syntactic context, predictability, talker characteristics, background noise characteristics, and spatial differences between the target and maskers play major role in identifying speech in complex listening environments. Here, we present psychometric properties of two closed set speech corpus: (1) co-ordinate response measure (CRM) and (2) Boston University Corpus (BUC) obtained from a group of fifty young normal-hearing, listeners. CRM sentences are of the form “Ready CALLSIGN go to a COLOR NUMBER now” and BUC sentences are of the form “NAME VERB NUMBER ADJECTIVE NOUN.” Method of constant stimuli was used to obtain psychometric functions for the two corpora in the presence of speech and speech shaped noise. One half of the participants were presented the stimuli at a constant level (20 dB SL), while it was presented at comfortable listening levels for the half. The effect of the stimulus complexity, presentation level, and spatial configuration of the target and noise on the identification thresholds and the shape of the psychometric functions will be discussed.
Read full abstract