Year: 2010
Type: dissertation
Abstract: When exposed to noise, speakers modify the way they speak, possibly in an effort to \nmaintain intelligible communication. These modifications are collectively referred to \nas the Lombard effect. The work described in this thesis compares speech production \nchanges induced by noise with various spectral and temporal characteristics, and \nexplores the perceptual consequence of these changes. The thesis consists of a series \nof experimental studies, which involve the analysis of speech corpora collected \nunder different noise conditions, with and without a communicative task. \nIntelligibility is also measured and predicted using a computer model. \nThe first study concerns the acoustic and phonetic consequences of N-talker \n“babble” noise on sentence production for a range of values of N from 1 (competing \ntalker) to “infinity” (speech-shaped noise). The effect of noise on speech production \nincreased with N and noise level, both of which act to increase the energetic masking \neffect of the noise. In a background of stationary noise, noise-induced speech was \nalways more intelligible than speech produced in quiet, and the gain in intelligibility \nincreased with N and noise level, suggesting that talkers modify their productions to \nameliorate energetic masking at the ears of the listener. \nThe effect of low- and high-pass filtered noise on speech production was also \nexamined to address the issue of whether speakers can compensate for energetic \nmasking by actively shifting their spectral energy to regions least affected by the \nnoise. Little evidence was found that speakers are able to modify their speech \nproduction to take advantage of those spectral regions clear of noise. \nTo evaluate the origin of the increased intelligibility of Lombard speech, the \nfundamental frequency and spectral tilt of speech produced in quiet were artificially \nmanipulated to match those of speech produced in speech-shaped noise. A perceptual \nevaluation showed that spectral flattening made a larger contribution to Lombard \nspeech intelligibility, but failed to find an influence of an increase in fundamental \nfrequency. A computational modeling study indicated that durational changes could \nalso play an important role in increasing intelligibility. These findings suggest that \nspeech modifications which reallocate energy in time and frequency to introduce more \n“glimpses” of clean speech in the presence of noise are able to contribute to speech \nintelligibility. \nAn analysis of the effect of noise on speech production requires material recorded \nwhile undertaking realistic tasks. The effect of a communication factor was explored \nusing conversational speech collected in the presence of maskers with differing \ndegrees of energetic and informational masking potential. The size of speech \nproduction changes was found to scale with the energetic masking potential of \nbackground noise, extending the findings with read speech to a communicative task. \nIn addition, relative to the non-communicative task, speakers exploited temporal \nplanning to reduce the amount of overlap with a modulated background noise, an \neffect which was stronger when the noise contained intelligible speech. \nIn conclusion, the strategies used by talkers to promote successful speech \ncommunication under various noise conditions reported in this thesis could enable \nspoken output applications such as dialogue systems to adapt to communicational \nenvironment.
Author: Youyi Lu
Language: English
Subfield: Experimental and Cognitive Psychology
Field: Psychology
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W1587989752