An Open-source Speech Processing Platform (OSP) for Research on Hearing Loss and Related Disorders

ABSTRACTHearing loss (HL) is an invisible disability affecting 360 million people worldwide that can have far- reaching consequences, including reduced quality of life, increased risk of falls, and cognitive decline, yet only one in three individuals with HL owns a hearing aid (HA). Despite a strong desire and need in the hearing loss community for quality, affordable hearing aids, the proprietary “black box” nature of commercial HAs is a major obstacle to achieving breakthroughs in audiology research, improving hearing aid technology, and translating it to use by hearing impaired individuals. Motived by the need for alternatives to commercial HA devices for audiology research, the Nadi team developed the Open-source Speech Processing Platform (OSP) for hearing aid research and development, which has the potential to bridge the gap between audiology research and commercialization of HL technologies, as well as promote psychophysical research beyond what is feasible today. OSP comprises multiple real-time master hearing aid algorithms (MHAs) and an embedded web server that hosts onboard apps (accessible by any browser-enabled device, such as a smartphone, tablet, or a PC) to monitor and control hearing aid operation. OSP hardware includes a wearable processor based on smartphone chipsets and behind-the-ear, receiver-in-the-canal (BTE-RIC) form factor ear-level assemblies.The feasibility of the OSP system was demonstrated in the lab and the field for both audiologist-fitting and self-fitting. The proposed direct Phase II project will validate the platform with clinical testing, the results of which will inform development of a manufacturing package for OSP hardware and a software support infrastructure for audiology research using OSP. The OSP software, including audiology research tools (Aim 1) and outcome assessment tools (Aim 2), will be validated by measuring aided speech perception outcomes in quiet and noisy environments for adults with HL. This will establish the validity of OSP for clinical research under realistic conditions. In addition, ecological momentary assessment (EMA) tools (Aim 3) will be validated using well established software test procedures for correctness of operation. The hardware revisions and Agile software development based on user feedback will result in a rich suite of algorithms, including multiple MHAs, apps for both researchers and listeners with HL to control MHAs, automated tools to assess outcomes, and tools to administer EMA surveys in the field. This project will be conducted by a strong interdisciplinary team comprised of investigators in systems design and signal processing, product development and support, software architecture, hardware manufacturing, and audiology. The success of this project will lead to a validated set of open-source HA research tools that will allow researchers to investigate the interactions among various signal processing functions, undertake psychophysical research not possible with current tools, and generate evidence- based solutions to increase access to hearing healthcare.