Navigation

Underdetermined Source Separation by Combination of Directivity Gain

Proposal for a Master Thesis

Topic:

Underdetermined Source Separation by Combination of Directivity Gain

Description:

Acoustic mixtures are ubiquitous in signal processing, and recovering the constituent signal components is often highly desirable. Besides the typically convolutive demixing models, Wiener filtering also presents a viable option. In [1], the required estimates of the target signal and noise PSDs are estimated by combining the directivity gains of a set of pre-designed, fixed beamformers. However, the choice of appropriate beamformer designs remains a topic for further research.
In this thesis, the method in [1] shall be investigated and compared quantitatively for different beamformer design techniques and under different acoustic conditions. Implementation and evaluation are to be done in MATLAB.

Download

Professor:

Prof. Dr.-Ing. Walter Kellermann

Supervisior:

Michael Günther, M.Sc. (Cauerstr. 7, room 05.020, email: michael.guenther@fau.de)

Prerequisites:

Course ’Digital Signal Processing’
Experience with MATLAB

Available:

Immediately