Combined effects of face masks, acoustic environments, and speaker gender on speech recognition ability of university students in an auralised classroom

Abstract

The prolonged use of face masks owing to COVID-19 has raised concerns regarding their effects on speech recognition in classroom settings. Currently, the influence of gender corresponding to the effects of face masks on speech recognition has not been studied. This study investigated the combined effects of gender and face masks (non-masked, surgical, horizontally folded and vertically folded KF94 masks) on speech recognition with the participation of two professional voice actors (1 female and 1 male) and university students (36 females and 35 males) using an auralisation technique under realistic listening conditions that included the effects of both room reverberance (RTs of 0.6 s and 1.2 s) and background noise according to teaching style (SNR of 0 dB for team learning and SNR of 12 dB for lectures). The results of this study revealed that speech recognition was significantly affected by four factors - face mask, acoustic environment, speakers’ and listeners’ gender tested in this study with different effect sizes. The room acoustic environments had the greatest effect on speech recognition, followed by speaker gender, face mask, and listener gender. High signal-to-noise ratio (lectures) in the classroom improved speech recognition in all face masked conditions including a non-masked condition compared to the low signal-to-noise ratio (team learning). In the three face masked conditions, reduced reverberation time of 0.6 s did not significantly affect speech recognition.Consequently, this study suggests that appropriate signal-to-noise ratio should be prioritized over reverberation time, face mask type, or speaker gender in mask-wearing conditions to improve speech recognition in classrooms.