Speech intelligibility tests with real and simulated classroom noise sources

Abstract

Speech intelligibility tests using the MRT stimuli were administered to 13 college age students in actual classroom spaces in the presence of noise sources identified in a survey of 41 elementary school classrooms in 7 schools. Twenty two different noise sources were identified during the survey of schools in a local school district. A-weighted, C-weighted, octave band and 1/3 octave band measurements of 182 individual noise sources were recorded in the sampled rooms. Noise sources were categorized as students speaking, HVAC equipment, other building equipment, water running in sinks, computers, lighting fixtures, A/V equipment, and miscellaneous sources. The noise source measurements provide a useful data base for future classroom acoustical studies. Five noise levels (35, 45, 55, 65, and 75 dBA) of 4 different types of noise were tested in the rooms. The speech intelligibility tests found significant effects (at the 0.01 level) from level, spectra, type, content, and annoyance caused by the different noise sources. A predictive model based on these variables had an R2 of 0.90 for speech intelligibility under the extremely diverse set of conditions tested.