Seminars in Hearing Research at Purdue (SHRP)

Date: Thursday, September 2, 2021

Time: 10:30 – 11:20 am

Location: LYLE 1150

Title: Deep Neural Networks for Speech Enhancement in Cochlear Implants

Speaker: Agudemu Borjigin, PhD candidate, Weldon School of Biomedical Engineering

Abstract:

Despite excellent performance in quiet, cochlear implants (CIs) usually fail to restore normal levels of intelligibility in noisy environments. Current state-of-the-art signal processing strategies in CIs provide limited benefits in terms of noise reduction or masking release. Recent developments in the field of machine learning have produced deep neural network (DNN) models with impressive performance in both speech enhancement and separation tasks. With sponsorship from hearing implant manufacturer— MED-EL, this work was an exploratory attempt to evaluate the use of DNN models as front-end pre-processors for enhancing CI users’ speech understanding in noisy environments. This talk will focus on model architectures, dataset, workflow (tools and resources), objective evaluation results, and pilot data collected from CI subjects.

Zoom Info:

https://purdue-edu.zoom.us/j/93108158900?pwd=RDdTQ0Z4UE9Rb0JUenhjMG1SMkp2QT09

Meeting ID: 931 0815 8900

Passcode: 11501150

---

The working schedule for this academic year will be available here shortly:

https://purdue.edu/TPAN/hearing/shrp_schedule

The titles and abstracts of the talks will be added here:

https://purdue.edu/TPAN/hearing/shrp_abstracts