Speeding up training of automated bird recognizers by data reduction of audio features

PeerJ

article

Allan G. de Oliveira¹  Thiago M. Ventura¹  Todor D. Ganchev¹  Lucas N.S. Silva¹  Marinêz I. Marques¹  Karl-L. Schuchmann¹ 

[1] Computational Bioacoustics Research Unit ,(CO.BRA), National Institute for Science and Technology in Wetlands ,(INAU), Universidade Federal de Mato Grosso;Institute of Computing, Universidade Federal de Mato Grosso;Faculty of Computing and Automation, Technical University of Varna;Institute of Bioscienses, Universidade Federal de Mato Grosso;Postgraduate Program in Ecology and Biodiversity Conservation, Institute of Biosciences, Universidade Federal de Mato Grosso;Postgraduate Program in Zoology, Institute of Biosciences, Universidade Federal de Mato Grosso;Zoological Research Museum Alexander Koenig and University of Bonn

关键词: Data representation;    Data reduction;    Random sampling;    Uniform sampling;    Piecewise aggregate approximation;   

DOI  :  10.7717/peerj.8407

学科分类：社会科学、人文和艺术（综合）

来源: Inra

【 摘 要 】

Automated acoustic recognition of birds is considered an important technology in support of biodiversity monitoring and biodiversity conservation activities. These activities require processing large amounts of soundscape recordings. Typically, recordings are transformed to a number of acoustic features, and a machine learning method is used to build models and recognize the sound events of interest. The main problem is the scalability of data processing, either for developing models or for processing recordings made over long time periods. In those cases, the processing time and resources required might become prohibitive for the average user. To address this problem, we evaluated the applicability of three data reduction methods. These methods were applied to a series of acoustic feature vectors as an additional postprocessing step, which aims to reduce the computational demand during training. The experimental results obtained using Mel-frequency cepstral coefficients (MFCCs) and hidden Markov models (HMMs) support the finding that a reduction in training data by a factor of 10 does not significantly affect the recognition performance.

【 授权许可】

CC BY