Recommendations for acoustic recognizer performance assessment with application to five common automated signal recognition programs

Automated signal recognition software is increasingly used to extract species detection data from acoustic recordings collected using autonomous recording units (ARUs), but there is little practical guidance available for ecologists on the application of this technology. Performance evaluation is an important part of employing automated acoustic recognition technology because the resulting data quality can vary with a variety of factors. We reviewed the bioacoustic literature to summarize performance evaluation and found little consistency in evaluation, metrics employed, or terminology used. We also found that few studies examined how score threshold, i.e., cut-off for the level of confidence in target species classification, affected performance, but those that did showed a strong impact of score threshold on performance. We used the lessons learned from our literature review and best practices from the field of machine learning to evaluate the performance of five readily-available automated signal recognition programs. We used the Common Nighthawk (Chordeiles minor) as our model species because it has simple, consistent, and frequent vocalizations. We found that automated signal recognition was effective for determining Common Nighthawk presence-absence and call rate, particularly at low score thresholds, but that occupancy estimates from the data processed with recognizers were consistently lower than from data generated by human listening and became unstable at high score thresholds. Of the five programs evaluated, our convolutional neural network (CNN) recognizer performed best, with recognizers built in Song Scope and MonitoR also performing well. The RavenPro and Kaleidoscope recognizers were moderately effective, but produced more false positives than the other recognizers. Finally, we synthesized six general recommendations for ecologists who employ automated signal recognition software, including what to use as a test benchmark, how to incorporate score threshold, what metrics to use, and how to evaluate efficiency. Future studies should consider our recommendations to build a body of literature on the effectiveness of this technology for avian research and monitoring. Recommandations pour l'évaluation des performances de reconnaissance acoustique et application à cinq programmes courants de reconnaissance automatisée de signaux sonores RÉSUMÉ. Les logiciels de reconnaissance automatisée de signaux sonores sont de plus en plus utilisés pour extraire les données de détection des espèces d'enregistrements acoustiques récoltés au moyen d'unités d'enregistrement autonomes (ARU en anglais), mais il existe peu d'instructions pratiques sur l'utilisation de cette technologie pour les écologistes. L'évaluation de la performance est une étape importante dans l'utilisation d'une technologie de reconnaissance acoustique automatisée parce que la qualité des résultats peut varier en fonction de divers facteurs. Nous avons passé en revue la littérature sur la bioacoustique afin de résumer les critères d'évaluation de la performance, et avons trouvé que l'évaluation, les paramètres choisis et la terminologie utilisée étaient inconsistants. Nous avons aussi constaté que peu d'études examinaient dans quelle mesure le seuil du score, c'est-à-dire la limite du niveau de confiance de la classification de l'espèce cible, influait sur la performance; toutefois, les chercheurs qui l'ont fait ont observé que le seuil du score avait un fort effet sur la performance. Nous avons appliqué les leçons apprises de notre revue de la littérature et les meilleures pratiques dans le domaine de l'apprentissage automatique pour évaluer la performance de cinq programmes de reconnaissance acoustique automatisée rapidement et facilement utilisables. Nous avons choisi l'Engoulevent d'Amérique (Chordeiles minor) comme espèce-modèle, parce que ses vocalisations sont simples, invariables et fréquentes. Nous avons réalisé que la reconnaissance automatisée était efficace pour déterminer la présence-absence de l'engoulevent et sa fréquence de chant, particulièrement à des seuils de score bas. Par contre, l'occurrence calculée à partir des données traitées par reconnaissance automatisée était systématiquement plus faible que celle calculée à partir des résultats issus d'experts ayant écouté les enregistrements, et elle devenait instable à des seuils de score élevés. Des cinq programmes évalués, notre reconnaisseur « Convolutional neural network » (CNN) est celui qui a le mieux performé; les reconnaisseurs intégrés dans Song Scope et MonitoR ont aussi bien performé. Les reconnaisseurs RavenPro et Kaleidoscope ont été moyennement performants et ont produit plus de faux positifs que les autres reconnaisseurs. Enfin, nous proposons six recommandations générales destinées aux écologistes qui utilisent les logiciels de reconnaissance acoustique automatisée, y compris quoi faire comme test de performances, comment incorporer un seuil de score, quels paramètres utiliser et comment en évaluer l'efficacité. Les recherches à venir devraient prendre en compte notre recommandation à l'effet de concevoir un corpus sur l'efficacité de cette technologie pour la recherche et les suivis aviaires.

[1]  Paul Roe,et al.  Timed Probabilistic Automaton: A Bridge between Raven and Song Scope for Automatic Species Recognition , 2013, IAAI.

[2]  Paul Roe,et al.  Acoustic features for multi-level classification of Australian frogs , 2015, 2015 10th International Conference on Information, Communications and Signal Processing (ICICS).

[3]  Tuomas Virtanen,et al.  Convolutional recurrent neural networks for bird audio detection , 2017, 2017 25th European Signal Processing Conference (EUSIPCO).

[4]  Rohini Balakrishnan,et al.  Examining the Effectiveness of Discriminant Function Analysis and Cluster Analysis in Species Identification of Male Field Crickets Based on Their Calling Songs , 2013, PloS one.

[5]  Carlos M. Travieso,et al.  Automatic Taxonomic Classification of Fish Based on Their Acoustic Signals , 2016 .

[6]  Nilson Evilasio de Souza Filho,et al.  Automatic classification of Turdus rufiventris song notes by spectrographic image template matching , 2014 .

[7]  Therese M. Donovan,et al.  Assessment of Error Rates in Acoustic Monitoring with the R package monitoR , 2016 .

[8]  H. Jaafar,et al.  Comparative Study on Different Classifiers for Frog Identification System Based on Bioacoustic Signal Analysis , 2022 .

[9]  Hervé Glotin,et al.  Clusterized Mel Filter Cepstral Coefficients and Support Vector Machines for Bird Song Identification , 2014 .

[10]  D Margoliash,et al.  Template-based automatic recognition of birdsong syllables from continuous recordings. , 1996, The Journal of the Acoustical Society of America.

[11]  Thierry Aubin,et al.  Screening large audio datasets to determine the time and space distribution of Screaming Piha birds in a tropical forest , 2016, Ecol. Informatics.

[12]  Lukas Machlica,et al.  Automatic recognition of bird individuals on an open set using as-is recordings , 2016 .

[13]  Paul Roe,et al.  Application of image processing techniques for frog call classification , 2015, 2015 IEEE International Conference on Image Processing (ICIP).

[14]  Todor Ganchev,et al.  Audio parameterization with robust frame selection for improved bird identification , 2015, Expert Syst. Appl..

[15]  Wei Chu,et al.  Noise robust bird song detection using syllable pattern-based hidden Markov models , 2011, 2011 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

[16]  Jeff Houlahan,et al.  Designing better frog call recognition models , 2017, Ecology and evolution.

[17]  David W. Armitage,et al.  A comparison of supervised learning techniques in the classification of bat echolocation calls , 2010, Ecol. Informatics.

[18]  Sidarta Ribeiro,et al.  Machine Learning Algorithms for Automatic Classification of Marmoset Vocalizations , 2016, PloS one.

[19]  Héctor Corrada Bravo,et al.  Automated classification of bird and amphibian calls using machine learning: A comparison of methods , 2009, Ecol. Informatics.

[20]  L. D. Vignolo,et al.  Furnariidae species recognition using speech-related features and machine learning , 2016 .

[21]  T. Mitchell Aide,et al.  Audio segmentation using Flattened Local Trimmed Range for ecological acoustic space analysis , 2016, PeerJ Comput. Sci..

[22]  Abeer Alwan,et al.  Dynamic time warping and sparse representation classification for birdsong phrase classification using limited training data. , 2015, The Journal of the Acoustical Society of America.

[23]  T. Ganchev,et al.  AUTOMATIC ACOUSTIC IDENTIFICATION OF SINGING INSECTS , 2007 .

[24]  Therese M. Donovan,et al.  A comparison of acoustic monitoring methods for common anurans of the northeastern United States , 2016 .

[25]  Dong-Hyun Lee,et al.  Learning multi-labeled bioacoustic samples with an unsupervised feature learning approach , 2013 .

[26]  Wei-Ho Tsai,et al.  Bird Species Identification Based on Timbre and Pitch Features of Their Vocalization , 2014, J. Inf. Sci. Eng..

[27]  Juan J. Noda Arencibia,et al.  Methodology for automatic bioacoustic classification of anurans based on feature fusion , 2016, Expert Syst. Appl..

[28]  Dan Stowell,et al.  Automatic large-scale classification of bird sounds is strongly improved by unsupervised feature learning , 2014, PeerJ.

[29]  Abeer Alwan,et al.  Evaluation of a Sparse Representation-Based Classifier For Bird Phrase Classification Under Limited Data Conditions , 2012, INTERSPEECH.

[30]  Todor Ganchev,et al.  Automated acoustic detection of Vanellus chilensis lampronotus , 2015, Expert Syst. Appl..

[31]  Hjalmar S. Kühl,et al.  Assessing the performance of a semi‐automated acoustic monitoring system for primates , 2015 .

[32]  Chin-Chuan Han,et al.  Automatic recognition of animal vocalizations using averaged MFCC and linear discriminant analysis , 2006, Pattern Recognit. Lett..

[33]  R. Warren,et al.  Application of autonomous recording units to monitor gobbling activity by wild turkey , 2015 .

[34]  Philip K. McKinley,et al.  Ensemble extraction for classification and detection of bird species , 2010, Ecol. Informatics.

[35]  Chenn-Jung Huang,et al.  Frog classification using machine learning techniques , 2009, Expert Syst. Appl..

[36]  Fabien Ringeval,et al.  Bird sounds classification by large scale acoustic features and extreme learning machine , 2015, 2015 IEEE Global Conference on Signal and Information Processing (GlobalSIP).

[37]  Steve Kelling,et al.  Towards the Automatic Classification of Avian Flight Calls for Bioacoustic Monitoring , 2016, PloS one.

[38]  L. Venier,et al.  Using automated sound recording and analysis to detect bird species‐at‐risk in southwestern Ontario woodlands , 2014 .

[39]  Christian Dietz,et al.  A continental-scale tool for acoustic identification of European bats , 2012 .

[40]  Michael Towsey,et al.  A practical comparison of manual and autonomous methods for acoustic monitoring , 2013 .

[41]  Ian Agranat,et al.  IDENTIFYING ANIMAL SPECIES FROM THEIR VOCALIZATIONS , 2009 .

[42]  Priti P. Rege,et al.  Classification of Bird Species based on Bioacoustics , 2013 .

[43]  Malay Kishore Dutta,et al.  Automatic classification of frogs calls based on fusion of features and SVM , 2015, 2015 Eighth International Conference on Contemporary Computing (IC3).

[44]  T. Mitchell Aide,et al.  Real-time bioacoustics monitoring and automated species identification , 2013, PeerJ.

[45]  I. Potamitis Automatic Classification of a Taxon-Rich Community Recorded in the Wild , 2014, PloS one.

[46]  Daniel J. Mennill,et al.  Comparison of manual and automated methods for identifying target sounds in audio recordings of Pileated, Pale-billed, and putative Ivory-billed woodpeckers , 2009 .

[47]  Ruben Gonzalez,et al.  Effects of compression and window size on remote acoustic identification using sensor networks , 2010, 2010 4th International Conference on Signal Processing and Communication Systems.

[48]  Naoya Oosugi,et al.  Semi-Automatic Classification of Birdsong Elements Using a Linear Support Vector Machine , 2014, PloS one.

[49]  Frank Kurth,et al.  Detecting bird sounds in a complex acoustic environment and application to bioacoustic monitoring , 2010, Pattern Recognit. Lett..

[50]  Olaf Jahn,et al.  Automated Sound Recognition Provides Insights into the Behavioral Ecology of a Tropical Bird , 2017, PloS one.

[51]  Klaus Riede,et al.  Automatic bird sound detection in long real-field recordings: Applications and tools , 2014 .

[52]  Germán Castellanos-Domínguez,et al.  Enhancing the dissimilarity-based classification of birdsong recordings , 2016, Ecol. Informatics.

[53]  Abeer Alwan,et al.  A robust automatic bird phrase classifier using dynamic time-warping with prominent region identification , 2013, 2013 IEEE International Conference on Acoustics, Speech and Signal Processing.

[54]  Louis Ranjard,et al.  Integration over song classification replicates: song variant analysis in the hihi. , 2015, The Journal of the Acoustical Society of America.

[55]  Aki Härmä Automatic identification of bird species based on sinusoidal modeling of syllables , 2003, ICASSP.

[56]  Louis Ranjard,et al.  Unsupervised bird song syllable classification using evolving neural networks. , 2008, The Journal of the Acoustical Society of America.

[57]  Xiaoli Z. Fern,et al.  Acoustic classification of multiple simultaneous bird species: a multi-instance multi-label approach. , 2012, The Journal of the Acoustical Society of America.

[58]  Janez Grad,et al.  Predicting species identity of bumblebees through analysis of flight buzzing sounds , 2017 .

[59]  Claudia Isaza,et al.  Automatic recognition of anuran species based on syllable identification , 2014, Ecol. Informatics.

[60]  Petr Doležel,et al.  Possibilities of Feedforward Multilayer Neural Network Classifier as a Detector of Pest Birds in Vineyards , 2015 .

[61]  Mark Bush,et al.  Classifying Frog Calls Using Gaussian Mixture Models , 2015, ISVC.

[62]  David A. Nicholson,et al.  Comparison of machine learning methods applied to birdsong element classification , 2016, SciPy.

[63]  Paul Roe,et al.  Detection of Rain in Acoustic Recordings of the Environment , 2014, PRICAI.

[64]  E. D. Chesmore,et al.  Automated identification of field-recorded songs of four British grasshoppers using bioacoustic signal recognition , 2004, Bulletin of Entomological Research.