On the acoustical design of university classrooms: effects of room acoustics on behavioral and subjective indices of listening effort

The acoustic design of university classrooms, besides targeting a high speech intelligibility, should also ensure a minimal effort in the speech reception task. An effortful listening, which involves an increased amount of cognitive resources, may compromise students’ learning and academic achievements, especially so for non-native listeners. The aim of this study is investigating the effect of room acoustic properties on the speech reception performance of university students, with specific reference to the listening effort as measured by either behavioral or subjective metrics. To achieve this objective the acoustic simulation model of a university classroom with a volume of 198 m3 and acoustical treatment on a lateral wall was calibrated with field measurements of T30 and C50 in order to obtain realistic auralization. The room acoustic conditions were virtually modified by either changing the room shape or removing the acoustic treatment. The simulated binaural impulse responses were used for laboratory experiments with headphones. Consonant confusion tests (Diagnostic Rhyme Tests) in the Italian language were proposed to 21 students, aged 23 to 39 years and self-reporting normal hearing: 10 native speakers and 11 non-native (German) speakers. A speech-shaped stationary noise was used to mask the speech signal, and listening conditions with Speech Transmission Index values ranging between 0.42 and 0.52 were obtained. During the experiment, data on the number of words correctly recognized (speech intelligibility, IS), response times (RT) and subjective ratings of listening effort (LE) were collected. No differences were found between the compared acoustic conditions in terms of IS results. The effects of room acoustic properties were instead discriminated when the listening effort was considered, with the greatest number of significant differences observed with the behavioral metric of RT. A decrease in IS results was found in all conditions for non-native listeners; no differences were observed in the corresponding listening effort results.