A sparse representation-based classifier for in-set bird phrase verification and classification with limited training data

Tan, Lee Ngee; Kossan, George; Cody, Martin L.; Taylor, Charles; Alwan, Abeer

doi:10.1109/icassp.2013.6637751

Cited by 16 publications

(43 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Since my ultimate goal was to study the sequential arrangements of phrase types in the songs of Cassin's Vireos, I could not sacrifice accuracy or noise robustness for the speed permitted by automated classification algorithms. Still, to ensure that my subjective classifications represented objectively identifiable and discrete phrase types, I verified a subset of my annotations using a supervised machine learning algorithm implemented by Tan et al (2013); this is a sparse-representation algorithm that uses spectrographic features to perform classification. The algorithm was designed specifically for use on Cassin's Vireo phrases and has been shown to perform at a high level on recordings with minimal background noise (Tan et al 2012(Tan et al , 2013.…”

Section: Phrase Identification and Annotationmentioning

confidence: 99%

Composition and sequential organization of song repertoires in Cassin’s Vireo (Vireo cassinii)

Hedley

2015

J Ornithol

View full text Add to dashboard Cite

The rules governing bird song sequences vary considerably across the avian phylogeny, and modifications to these rules represent one of the many ways in which bird song varies interspecifically. Cassin's Vireo (Vireo cassinii) is one species that shows a highly structured syntax, with clearly non-random patterns of sequential organization in their songs. Here I present a description of Cassin's Vireo song sequences from the Sierra Nevada Mountains in California and employ network analysis to quantify transition patterns within the songs. Repertoire sizes varied between 44 and 60 phrase types per bird for the 13 birds analyzed here. The repertoire was subdivided into 'themes' containing between two and seven phrase types. The birds sang the phrase types in a given theme for a time before eventually introducing a new theme; in this manner the repertoire was revealed relatively slowly over time. Theme composition within a bird's repertoire did not change within or between singing bouts throughout the season. The tendency to sing in themes was corroborated by network analysis, which revealed small-world structure in the songs. Phrase types were widely shared within the population. I discuss these findings as they compare with the singing styles of other species, both closely and distantly related.

show abstract

Section: Phrase Identification and Annotationmentioning

confidence: 99%

Composition and sequential organization of song repertoires in Cassin’s Vireo (Vireo cassinii)

Hedley

2015

J Ornithol

View full text Add to dashboard Cite

show abstract

“…The segmented phrases of Cassin's Vireo by the proposed method were also tested on the bird phrase classifier. The sparse representation-based (SR) classifier [7] and the support vector machine (SVM) classifier [8] were considered in the experiment. In SR classifier, 7 training tokens per phrase were used.…”

Section: Discussionmentioning

confidence: 99%

Change point detection methodology used for segmenting bird songs

Wang

Hudson

Tan

et al. 2013

2013 IEEE China Summit and International Conference on Signal and Information Processing

Self Cite

View full text Add to dashboard Cite

A bird phrase segmentation method using entropy-based change point detection is proposed. Spectrograms of bird calls are usually sparse while the background noise is relatively white. Therefore, considering the entropy of a sliding time-frequency block on the spectrogram, the entropy dips when detecting a signal and rises when the signal ends. Rather than applying a hard threshold on the entropy to determine the beginning and ending of a signal, a Bayesian change point detection is used to detect the statistical changes in the entropy sequence. Tests on a database of Cassin's Vireo (Vireo cassinii), our proposed segmentation method with spectral subtraction or a novel spectral whitening method as the front-end generates more accurate time labels, lower the false alarm rate than the conventional time-domain energy detection method and achieves high phrase classification rate.

show abstract

“…All experiments in this paper use the audio recordings described in [2]. Song fragments (phrases) for classification were obtained from recordings of Cassin's Vireo (Vireo cassinii).…”

Section: Sound Datamentioning

confidence: 99%

“…Automatic recognition systems of bird sounds are needed, among other things, to annotate large amounts of birdsong recording data [1]. Automatic bird-phrase recognition is challenging due to within class variability, limited training data, and noisy environments [2]. Two spectrograms with identical class labels may look different due to time misalignment and frequency variation.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A robust automatic bird phrase classifier using dynamic time-warping with prominent region identification

Kaewtip

Tan

Alwan

et al. 2013

2013 IEEE International Conference on Acoustics, Speech and Signal Processing

Self Cite

View full text Add to dashboard Cite

In this paper, we present a novel approach to birdsong phase classification using template-based techniques suitable even for limited training data and noisy environments. The algorithm utilizes dynamic time-warping and prominent (high-energy) time frequency regions of training spectrograms to derive templates. The algorithm is evaluated on 32 classes of Cassin's Vireo bird phrases. Using only three training examples per class, our algorithm yields a phrase accuracy of 96.23%, outperforming other classifiers (e.g. 85.21 % classification accuracy of SVM). In the presence of additive noise (10 dB SNR degradation), the proposed classifier does not degrade significantly, compared to others.

show abstract

A sparse representation-based classifier for in-set bird phrase verification and classification with limited training data

Cited by 16 publications

References 15 publications

Composition and sequential organization of song repertoires in Cassin’s Vireo (Vireo cassinii)

Composition and sequential organization of song repertoires in Cassin’s Vireo (Vireo cassinii)

Change point detection methodology used for segmenting bird songs

A robust automatic bird phrase classifier using dynamic time-warping with prominent region identification

Contact Info

Product

Resources

About