Publications-Detail

Natural own-voice perception while wearing headphones - Active Occlusion Cancellation with Equalized Hear-Through

Authors:
Liebich, S.Thieling, L.Fabry, J.Jax, P.Vary, P.
Book Title:
Proceedings of International Workshop on Acoustic Signal Enhancement (IWAENC)
Organization:
IEEE
Publisher:
IEEE
Date:
Sep. 2018
Note:

Demonstration

Language:
English

Abstract

The perception of the own voice is a common complaint of hearing aid users, as their voice is changed when the ear canal is occluded. The effect is called the “occlusion effect” and it manifests in an amplification of low frequencies and an attenuation of high frequencies. The cause of this effect is the occlusion of the ear canal, which alters and unbalances air-conducted and bone-conducted sound transmitted into the ear canal. The same effect occurs with any hearing device, which is occluding the ear canal, e.g. in-ear headphones. In our demo we are showing an active approach to achieve a more natural perception of the own voice as well as ambient sounds in the surrounding. It uses a twofold approach with an Active Occlusion Cancellation (AOC) and an equalized Hear Through (HT). The amplification of low frequencies is tackled with the AOC component, which works by destructive interference, and the attenuated high frequencies are recovered with the HT component. The following figure illustrates the acoustic and the electronic front-end, including the AOC and the HT component. The AOC is realized by a time-invariant feedback controller K(z). The HT is equalized by the feedforward filter V(z) and both components are roughly independent due to the secondary path estimation G ̂(z). The principles will be explained in more detail during the demonstration.
The theory and algorithms have been described in [Liebich2018] and implemented on a stationary system. Due to the lack of mobility, we implemented the methods on a DSP platform. It is realized in a slightly modified manner to work with the limitations of the DSP, yet still allowing for comparable performance. The demo will include an in-ear headphone as well as a specialized DSP filter board, which is controlled by a laptop. The DSP is a specialized digital filtering DSP, with an analog-to-analog latency of roughly 40 μs.

Download

BibTeX