Contributions of temporal envelope and temporal fine structure to segregation of Lombard speech from background noise

Abstract

Lombard speech is known to make it easier for listeners to perceive speech signals in the presence of background noise. The aim of this study was to investigate how temporal characteristics such as the temporal envelope and temporal fine structure (TFS) in Lombard speech contribute to improve the segregation of Lombard speech from background noises. In the speech production task, we confirmed that Lombard speech had increased low-frequency amplitude modulations in the temporal envelope and increased frequency modulations in TFS compared to plain speech in frequency ranges covering first and second formants. In a speech perception task, ease of speech segregation was compared between Lombard and plain speech. We also compared ease of segregation of tone-vocoded speech created from Lombard and plain speech to examine the effect of eliminating TFS. We found that Lombard speech with both high amplitude and frequency modulations had improved speech segregation in both original and tone-vocoded speech conditions, but the difference in ease of speech segregation between Lombard and plain speech tended to be larger in original speech than tone-vocoded speech conditions. The present findings suggest that an enhanced envelope is the main contributor to improved segregation of Lombard speech, while increased frequency modulation plays a complementary role.