Collaborating Hearing Aids


Olivier Roy


Hearing aids are audio capture devices which aim at providing the hearing impaired with better speech intelligibility. Most state-of-the-art systems involve sensing devices that work independently (a bilateral system). In the near future, the hearing instrument at the left and right ear of the user will be able to communicate over a wireless link. This binaural system will promote the design of algorithms that benefit from this collaboration, potentially allowing for better noise reduction capability and improved acoustic feedback cancelation.

The first part of this project considers the scenario where a hearing aid has access to the signals recorded at both ears of the user for noise reduction purpose. We characterize the optimal trade-off between the coding rate sustainable by the wireless communication medium and the noise reduction gain provided by a multi-channel Wiener filtering approach. The considered setup is first identified as a remote source coding problem with side information at the decoder. In the literature, this problem has been solved in the scalar case and is referred to as remote, indirect or noisy Wyner-Ziv coding. Our first contribution is to evaluate the available results to continuous-time sources with memory in order to characterize the gain achieved by collaborating hearing aids as a function of the communication bitrate. We also derive the optimal rate-distortion trade-off when the side information is not taken into account at the encoder. Based on these results, gain-rate functions are computed for different scenarios of interest. In particular, closed-form formulas are provided for a very simple, yet insightful, scenario. Optimal rate allocation strategies between the two hearing aids are also derived. Finally, numerical simulations using data recorded in a realistic acoustic environment are compared with the presented theory.

Motivated by obvious power consumption limitations, the second part of this project aims at designing practical source coding schemes that allow the exchange of signals recorded at both ears of the user. We develop a practical coding method that benefit from the characteristics of the recording setup depicted in Figure 1 (a) in order to achieve this goal with low bitrates, low delay and low complexity. The proposed method builds on psychoacoustic fundamentals in order to reproduce at one hearing aid a binaural signal that is perceptually similar to the original one. This is achieved by imposing appropriate binaural cues on the signal directly recorded at one hearing device. These cues are efficiently estimated and coded using distributed coding principles. Figure 1 (b) illustrates the corresponding coding setup.

Ongoing research is focusing on the development of novel distributed source coding schemes that are based on the concept of signals with finite rate of innovation and that can be applied to binaural hearing aids.

Binaural hearing aids with wireless communication links
Figure 1: Binaural hearing aids. (a) Typical recording setup. (b) Transmission of signals from one hearing aid to the other using a wireless link.

Major Publications


O. Roy and M. Vetterli, Collaborating Hearing Aids, MSRI Workshop on Mathematics of Relaying and Cooperation in Communication Networks, 2006.
[detailed record] [bibtex]

O. Roy and M. Vetterli, Rate-Constrained Beamforming for Collaborating Hearing Aids, International Symposium on Information Theory (ISIT), pp. 2809-2813, 2006.
[detailed record] [bibtex]

O. Roy and M. Vetterli, Rate-Constrained Collaborative Noise Reduction for Wireless Hearing Aids, IEEE Transactions on Signal Processing, Vol. 57, Nr. 2, pp. 645-657, 2009.
[detailed record] [bibtex]

O. Roy and M. Vetterli, Distributed Spatial Audio Coding in Wireless Hearing Aids, IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA), pp. 227-230, 2007.
[detailed record] [bibtex]


Martin Vetterli

Project Period

October 2004 – Present

Funding Source

National Competence Center in Research on Mobile Information and Communication Systems (NCCR-MICS), a center supported by the Swiss National Science Foundation under grant number 5005-67322.