Authors: |
S. Goetze, T. Rohdenburg, V. Hohmann, B. Kollmeier, K.-D. Kammeyer |
Abstract: |
Multi-channel beamformer algorithms are promising solutions for noise reduction in hearing aids as they exploit the spatial distribution of the interfering signals and therefore in general lead to less signal distortion than single channel algorithms. Beamformers need a priori information about the microphone array and the direction of arrival of the target speech source. For head-worn arrays it is usually assumed that the user physically steers the arrays’ look direction toward the desired speech source. This may become unsatisfying for the hearing aid user for high directivity beamformers with a small main lobe and when the target signal source is moving. In this contribution an automatic steering (electronic control of the look direction) is applied based on the dual delay line approach after Liu et al.. This approach is modified to be applicable for head-mounted hearing-aid arrays. We show that the original free-field approach does not work on a head-mounted array because of the inappropriate propagation model. If we apply the true HRTF or a spherical head propagation model, the estimate is reliable within ±8 degree mean estimation error for an input SNR of 10dB or higher. However, for lower SNR the method seems to be not robust enough. |
Document type: |
Conference Paper |
Publication: |
Xiamen, China, 28. November - 1. December 2007 |
Conference: |
International Symposium on Intelligent Signal Processing and Communication Systems (ISPACS07) |
Pages: |
185-188 |
Files: |
BibTEX
|