Volume 106, Issue 4, October 1999
Index of content:
- ARCHITECTURAL ACOUSTICS 
106(1999); http://dx.doi.org/10.1121/1.427931View Description Hide Description
Acoustical measurements were performed in 30 randomly chosen, unoccupied classrooms at the University of British Columbia (UBC). Tests had previously been done in 46 unoccupied UBC classrooms, as well as in 10 of these when occupied by students. The results for the 10 classrooms were used to correct the “unoccupied” results to the half-occupied and fully occupied conditions. The objective of the work was to characterize the 30 classrooms, which were used in subsequent studies, to determine the acoustical quality of the UBC classroom stock and how this depends on the classroom design and the presence of students, and to elucidate characteristics of classroom acoustics relevant to optimal design. The results showed that the UBC classroom stock is of far from optimum acoustical quality when unoccupied, but is much better in the occupied condition. Generally, many classrooms have excessive reverberation and result in low speech levels, especially at the back of the rooms; in addition, they have excessively noisy ventilation systems.
106(1999); http://dx.doi.org/10.1121/1.427932View Description Hide Description
Speech intelligibility in rooms is influenced by room acousticseffects and by the signal-to-noise ratio (S/N) of the speech and ambient noise. Several measures such as useful-to-detrimental sound ratios and the speech transmission index predict the combined effects of both types of factors. These measures were evaluated relative to speech intelligibility test results obtained in simulated sound fields. The use of simulated sound fields made it possible to create the full range of combinations of room acoustics and S/N effects likely to be found in rooms for speech. The S/N aspect is shown to be much more important than room acousticseffects and new broadband useful-to-detrimental ratios were validated. Useful-to-detrimental ratios, speech transmission indexmeasures, and values of the articulation loss for consonants were all reasonably accurate predictors of speech intelligibility. Further improvements to these combined measures are suggested.