Path models showing the effects of musical expertise (0, nonmusician; 1, musician) and age (0, young; 1, older) on speech-in-noise perception threshold via auditory working memory (sum of forward and backward digit span) as the mediator under four conditions: (A) noise separation, (B) noise colocation, (C) speech separation, and (D) speech colocation. Dotted lines indicate insignificant paths. Standardized path coefficients are displayed on direct paths. *p < 0.05, **p < 0.01, ***p < 0.001.