Research On Evaluation Index Of Wild Bird Song Based On Microphone Array

Biodiversity assessment remains a major challenge for ecological conservation,and traditional assessment methods are costly and have limited applicability.Considering the wide existence of acoustic signals in biological communication,passive acoustic monitoring has become an effective and convenient method for monitoring species diversity.As an environmental indicator that is extremely sensitive to environmental changes,birds have become the primary research object of passive acoustic monitoring.But the increasing volume of acoustic monitoring data makes it more difficult to analyze manually or identify bird species automatically.In recent years,ecological researchers have proposed acoustic indices that do not target individual organisms,but rather communities,allowing rapid assessment of species richness in terms of acoustic complexity.Therefore,acoustic indices are widely used as an important tool for rapid biodiversity assessment.In view of the above situation,this thesis focuses on the performance analysis of the acoustic indices for the wild bird vocalizations.Based on the field data collected using the microphone array in the Nanjishan Wetland of Poyang Lake,speech for synchronous calibration has been separated,furthermore,the correlation analysis of acoustic indices with time,weather and bird vocalization behavior was carried out based on pure birdsong monitoring data.Specifically,three control computational experiments were designed using recordings from the Xeno-canto wild birdsong database,to measure the relationship objectively between acoustic indices and bird species richness under different vocalizations sound unit shapes,frequency of vocalizations occurrence,and vocalizations intensity.In the experiments,the above conditions as well as number of bird species were precisely controlled.Experimental results demonstrated that the acoustic diversity index(ADI)is more robust than other acoustic indices in assessing bird species richness,and is moderately correlated with species richness.At the same time,the effect of threshold on ADI under different SNR is analyzed in detail.Then,the performance of the acoustic indices in the complex acoustic environment was evaluated by using the continuous monitoring data of the wild birds collected from the Nanjishan Wetland of Poyang Lake.In order to identify the individual in time,ornithologists use speech to synchronize the information of species and behavior of birds in the field.Since the speech signal mixed with the birdsong,speech will lead to the deviation of the acoustic indices,the speech separation is realized by using the spatial filtering of the microphone array.Based on the above data,the diurnal variation of the acoustic indices,the relationship between the acoustic indices and weather,and the relationship between the acoustic indices and bird vocalization behavior were studied.Experimental results show that ADI can clearly reflect the changes in the complex acoustic activities of wetland birds,but it will fail in heavy rain,and changes in birds vocalizations behavior will affect the acoustic indices.The thesis has provided a good support for the rapid biodiversity assessment based on acoustic indices and has a great application value in the field of ecological civilization construction.