Contribution
Sound Source Localization with Beamforming and Pitch Detection
* Presenting author
Abstract:
This presentation is concerned with the three-dimensional localization of sound sources with microphone arrays use as acoustical cameras and beamforming algorithms. Traditionally, acoustical beamforming can be used to estimate the direction of sources, but it can be extended into three dimensions to include distance estimation. Preliminary simulations showed promising results in terms of 3D localization, but performance during outdoor measurements of Unmanned Aerial Vehicles was unsuccessful.Our goal is to develop the distance estimation method further to be more roust against unfavourable environmental conditions. The SAMV (Iterative Sparse Asymptotic Minimum Variance) algorithm was chosen for its capacity to handle correlated signals, which means that in practice, it can separate direct and reflected sounds. The HPS (Harmonic Product Spectrum) algorithm is also used for pitch detection, to adaptively fit the observed frequency range onto the temporally changing frequency spectra of the received signals. These algorithms are expected to improve distance estimation, and they are tested through measurements performed in a semi-anechoic chamber.