A method has been developed for determining typical long-term speech and background-noise levels during lectures. Lectures are recorded and the recordings digitized and processed to obtain sound-pressure-level frequency distributions to which three normal-distribution curves are fit. The maximum values of these curves are associated with long-term sound-pressure levels associated with speech, ventilation noise, and student-activity noise. Recordings made during 18 university lectures in 11 classrooms have been analyzed. Average (standard deviation) A-weighted levels for the various sound components were determined as follows: ventilation noise, 40.9 (3.9) dB; student-activity noise, 41.9 (4.0) dB; total background noise, 44.4 (3.5) dB; and received speech-signal, 50.8 (3.9) dB. The average (standard deviation) A-weighted speech-signal to background-noise ratio was 7.9 (3.1) dB. That of the instructor sound-power level was 64.5 (4.2) dB. Empirical models have been developed to predict the room-average A-weighted results using multivariable regression analysis. Further analysis in the 63- to 8000-Hz octave bands confirmed the spectra of ventilation noise and of speech, and determined the spectrum of student-activity noise and of the speech-signal to background-noise ratio.
[1]
T. Houtgast.
The effect of ambient noise on speech intelligibility in classrooms
,
1981
.
[2]
A Markides,et al.
Speech levels and speech-to-noise ratios.
,
1986,
British journal of audiology.
[3]
V Viljanen,et al.
Acoustic conditions for speech communication in classrooms.
,
1991,
Scandinavian audiology.
[4]
R C Seewald,et al.
Ear level recordings of the long-term average spectrum of speech.
,
1991,
Ear and hearing.
[5]
H. Dillon,et al.
An international comparison of long‐term average speech spectra
,
1994
.