Abstract

Speech produced in noisy environments (Lombard speech) is characterized by a range of acoustic and phonetic changes. These changes stem from increased speaking effort which reflects communicative intent as well as decreased auditory feedback of the speaker’s own voice. An accurate understanding of real-world Lombard effects is important in hearing science for the development and assessment of signal processing strategies targeting realistic speech signals. While Lombard effects are well known from the literature, studies of Lombard speech have typically been based on relatively unnatural speaking tasks such as reading from a script and have been measured in simplified acoustic backgrounds such as stationary noise or constructed babble noise. Lombard speech produced under such unnatural conditions may differ significantly from speech produced in real-world settings. This study describes a novel method of eliciting natural conversational speech across five highly realistic everyday acoustic environments. Th...

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call