The sensitive ear: towards real-time monitoring of in-ear physiological and audio signals for managing decreased sound tolerance

Abstract

People with decreased sound tolerance often wear hearing protection or headphones to block out the sounds they are sensitive to. However, conventional hearing devices indiscriminately block out all sounds, even when these sounds do not cause distress and may be useful for the wearer to hear. The resulting acoustic isolation can prevent the wearer from fully participating in their environment and could also have long term consequences, as research suggests that regular noise avoidance may exacerbate sensitivity to sound. A potential solution to over-avoidance of noise could be a hearing device specifically designed for decreased sound tolerance to attenuate sound only when necessary to reduce the wearer’s stress, by recording external sounds, processing them, and selectively transmitting them inside the earcanal. However, there are several challenges associated with developing a hearing device that selectively attenuates distressing sounds: 1) Existing devices used to manage decreased sound tolerance are not readily modifiable, and the resources required to independently develop a new device can be a barrier to prototyping new features; 2) There is limited research on how to balance usability considerations such as how effectively distressing sounds are filtered out and potential impacts on speech comprehension, which could be of particular concern for populations where decreased sound tolerance symptoms co-occur with communication challenges; 3) The presentation of decreased sound tolerance can vary greatly, for different individuals as well as over time for the same individual, and therefore, the audio processing required could depend on each wearer’s current needs. This thesis presents the findings of research efforts aiming to address the above challenges: First, existing hardware and software tools to prototype hearing device algorithms were evaluated for their suitability for decreased sound tolerance research. A processing pipeline interfacing an external personal computer with a microcontroller-based platform was developed to allow for algorithms implemented on the computer to control real-time audio processing. Second, an experimental protocol, piloted online with 50 participants, was developed to assess the latency requirements for real time audio processing based on how simulated delays affected a task necessitating speech comprehension. Third, with a dataset of 30 participants’ heartbeat sounds recorded from a hearing device during a stress-induction protocol, an automatic stress recognition system was developed and validated. These results open the possibility for a hearing device that can adapt to the wearer by simultaneously monitoring stress and controlling triggering noise for the benefit of users with decreased sound tolerances.

Date	21 Dec 2022
Original language	American English
Awarding Institution	École de technologie supérieure
Supervisor	Jérémie Voix (Supervisor) & Rachel Bouserhal (Co-supervisor)