Investigating speech processing in the inner ear through otoacoustic emissions related to running speech

• Applicant: Professor Dr. Tobias Reichenbach
• Subject Area: Human Cognitive and Systems Neuroscience ; Cognitive, Systems and Behavioural Neurobiology
• Term: since 2022
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 514955521

Understanding speech in noise is important for communication in many real-world environments. Investigating the underlying neural mechanisms is, however, complicated due to the complex structure of speech. However, novel statistical approaches have recently allowed for significant advances. As an example, they have identified a role of neural responses to the temporal fine structure of voiced speech. Voiced speech possesses indeed a particular spectral structure, with a fundamental frequency and higher harmonics, that is tracked by neural activity, in particular in subcortical areas. This neural activity can be regulated through corticofugal connections from the cerebral cortex and can contribute to speech-in-noise processing.Such top-down control could extend further to the inner ear that transduces the mechanical sound vibrations into electrical signals. In particular, the inner ear spatially segregates a complex sound into its distinct frequency components, and mechanically amplifies weak sound stimulation through an active feedback. A consequence of the active feedback are otoacoustic emissions, sounds generated and emitted by the cochlea. In particular, the cochlear amplification can produce combination tones through nonlinear distortion (distortion-product otoacoustic emissions or DPOAEs). However, DPOAEs have so far only been measured using pure tones.The goal of this proposal is to develop and use statistical approaches for investigating the processing of running speech in the cochlea. In a first step, we will develop the methodology to measure DPOAEs that are related to the temporal fine structure of voiced speech. We will thereby seek to elicit and measure multi-band emissions, as well as emissions recorded from the ipsi- and from the contralateral side to which speech is presented.In a second step we will develop a computational model that will allow to clarify how and where the speech-related DPOAEs are generated in the cochlea. In an iterative feedback loop, the results from the computational model will inform the development of the experimental methodology, the results of which will in turn inform and validate the modelling. The computational model will also be utilized to investigate how effective the filtering of background noise in the cochlea can be for enhancing a target voice.In a third step, we will employ the developed methodology to investigate whether the inner ear already contributes to selective attention to one of two competing speakers. Multi-band speech-related DPOAEs will thereby be measured while volunteers attend one of two speakers. Ipsi- and contralteral emissions will be analyzed separately, and the influence of the frequency band on the attentional modulation will be quantified.We will thereby establish speech-related DPOAEs as a novel tool for investigating speech processing in the inner ear. This will open up new applications, from basic research to clinical assessments of speech-in-noise impairments.

• DFG Programme: Research Grants