Sound pressure level of running speech: percentile level statistics and equivalent continuous sound level

Sound pressure level estimates of running speech must accurately reflect everyday communication in an appropriate unit of measure. Several factors may influence the overall sound pressure level of speech under everyday circumstances, such as age of the listener and speaker, speaking distance, surrounding noise and room acoustics. It is very difficult to estimate how all these factors will combine to affect speech intensity in a given situation. Without that knowledge it is best to deal with measures that reflect running speech in a representative sample of talkers. As to measuring units, there is considerable variation in the way speech intensity is measured and quantified. Due to their fluctuating nature, sound pressure levels of running speech require some type of averaging over time. Equivalent continuous sound levels (Leq) and percentile levels (L(N)) by definition reflect temporal distribution of sound energy. The aim of this study was to collect time-weighted sound pressure level estimates of speech in order to obtain normative descriptive statistics. 400 volunteers between 7 and 95 years of age participated in the study, providing a sample representative of official local demographic data. Descriptive statistics of overall Leq and eleven L(N )measures are presented. Overall Leq is extrapolated to other speaker-listener distances and absorption coefficients. 3 x 2 ANOVA analyses are used to screen the data for significant differences among gender and age groups. Restrictions and advantages of time-weighting units for speech intensity level are discussed, as well as their relevance for audiometry and voice pathology.