Lombard effect based speech analysis across noisy environments for voice communications with cochlear implant subjects

Changes in speech production including vocal effort based on auditory feedback are an important research domain for improved human communication. For example, in the presence of environmental noise, a speaker experiences the well-known phenomenon known as Lombard effect. Lombard effect has been studied for normal hearing listeners as well as for automatic speech/speaker recognition systems, but not for cochlear implant (CI) recipients. The objective of this study is to analyze the speech production of CI users with respect to environmental change. We observe and study this effect using mobile personal audio recordings from continuous single-session audio streams collected over an individual's daily life. Prior advancements in this domain include the “Prof-Life-Log” longitudinal study at UTDallas. Four CI speakers participated by producing read and spontaneous speech in six naturalistic noisy environments (e.g., office, car, outdoor, cafeteria, etc.). A number of speech production parameters (e.g., short-time log-energy, fundamental frequency, etc.) known to be sensitive to Lombard speech were measured for both communicative and non-communicative speech as a function of environment. Results indicate that variability in the speech production parameters were found in the upward direction with an increase in background noise level. Overall higher values in acoustic variables were observed in the inter-personal conversations related to the non-conversational speech.