An Automated Acoustic System to Monitor and Classify Birds

Collisions between aircraft and birds have become an increasing concern for human health and safety. More than four hundred people and over four hundred aircraft have been lost globally in recent years, according to a FAA report. To minimize the number of birdstikes, microphone arrays have been used to monitor birds near the airport or some critical locations in the airspace. However, the range of existing arrays is only limited to a few hundred meters. Moreover, the identification performance in low signal-to-noise environment is not satisfactory. Under the support of the US Air Force, Intelligent Automation, Incorporated (IAI) and the University of Missouri at Columbia, propose a novel system to improve bird monitoring and recognition system in noisy environments. First, a microphone dish concept (microphone array with many concentric rings) is proposed that provides very directional and long range (a few thousand meters) acquisition of bird sounds, can simultaneously pick up and track sound from different directions, and the cost of the dish will be a few hundred dollars. Second, an efficient recognition algorithm is proposed which uses Hidden Markov Model (HMM) and Gaussian Mixture Models (GMM). The overall system is suitable for real-time monitoring and recognition for a large number of birds. Here we will summarize some preliminary results of our proposed method. First, we will give a brief overview of the proposed system, which consists of several major parts: microphone dish and data acquisition system, Direction of Arrival (DOA) estimation, beamformer to eliminate interferences, and bird classifier. Second, we will describe a new wideband DOA estimation algorithm and provide a comparative study between estimation results using linear array and our circular array. Third, beamforming algorithm will be introduced through a comparative study between the linear array and our circular array. A new beamforming algorithm for dish array has been developed. It was found that the dish array has several key advantages over the linear array, including less ambiguity angles, more consistent performance, etc. Fourth, bird classification results using GMM method will be presented. Fifth, the development of a prototype microphone dish will be included. A dish array consisting of 64 microphone elements has been developed and used to collect sound data in laboratory and in an open space. Sixth, experimental results will be described to show the performance of the software and hardware. Main Results 1. Overall System Description Figure 1 shows the proposed system which consists a microphone dish, a data acquisition system, and software processing algorithms such as direction finder, beamformer, and bird sound classification. A/D conversion Direction finder Beamformer Bird sound segmentation Bird verification