No abstract

High mobility and an ability of gathering data from large terrains makes Unmanned Aerial Vehicles (UAVs) an excellent platform for placing visual or acoustic sensors. One recently emerging application of UAVs is search and rescue operation, during which drones are used to localize people in distress. A common approach to determine the target position is to rely on visual data recorded by cameras. However, in situations of limited visibility such as in presence of smoke, at night or when a person is trapped under debris, acoustic information can be exploited to perform the localization of people in distress. Solutions based on acoustic information gathered by drone-embedded microphone array are a promising alternative to the methods based on vision, and they are currently being widely examined for UAV applications. The main issues encountered in acoustic source localization using drones include high ego-noise and wind produced by the propellers. This paper investigates the statistical properties of drone’s ego-noise and proposes an algorithm for acoustic source localization which exploits the sparsity of sound sources in time-frequency domain. A comparison of the results obtained by the proposed method and by commonly used approaches clearly shows the benefits of using the proposed processing.

In this paper we present a method for the separation of sound source signals recorded using multiple microphones in a reverberant room.In particular, we propose a maximum a posteriori (MAP) estimator based on the multichannel nonnegative tensor factorization (NTF) model with the localization prior distribution on the mixing matrix, in which the latent data consists of the so-called sub-sources for an improved performance in a reverberant environment. For the proposed MAP estimator, we derive the sub-source based expectation maximization (EM) algorithm with the multiplicative update rules (MU) and the localization prior distribution (LP) on the mixing matrix (SSEM-MU-LP). We then perform several experiments for speech and instrumental sound sources recorded using two microphones, in determined and under-determined scenarios, and with different types of initialization of the model parameters. The results of these experiments clearly indicate a significant improvement of the proposed algorithm with the localization prior over the state-ofthe-art NTF-based source separation algorithms, which can reach up to 50% in the signal-to-distortion ratio.

No abstract

No abstract