Univ. of Ottawa, Ottawa, ON K1N 6N5, Canada
It is well known that the presence of noise produces an increase in vocal levels [E. Lombard, Ann. Maladies Oreille Larynx 37, 101--119 (1911); H. Lane and B. Tranel, J. Speech Hear. Res. 14, 677--709 (1971)]. Pick et al. [J. Acoust. Soc. Am. 85, 894--900 (1989)] suggested that through training the effect could either be enhanced or reduced but not completely eliminated. It is possible, nevertheless, that people with voice training react differently to that effect [W. D. Ward and E. M. Burns, J. Res. Singing 1 (2), 24--44 (1978)]. Two groups, one with voice training (n=23) and one without voice training (n=23) were recorded with and without masking noise. It was found that 39% of the masked speech samples of the trained group failed to produce the Lombard sign whereas only 9% of the untrained (n=23) did not increase vocal levels under masking noise. Voice training was significantly related to cases where the Lombard effect was absent. This contradicts Pick et al.'s conclusion that the Lombard sign is always present in masking noise.