English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

In this paper, we propose an effective sound event detection (SED) methodbased on the audio spectrogram transformer (AST) model, pretrained on thelarge-scale AudioSet for audio tagging (AT) task, termed AST-SED. PretrainedAST models have recently shown promise on DCASE2022 challenge task4 where theyhelp mitigate a lack of sufficient real annotated data. However, mainly due todifferences between the AT and SED tasks, it is suboptimal to directly utilizeoutputs from a pretrained AST model. Hence the proposed AST-SED adopts anencoder-decoder architecture to enable effective and efficient fine-tuningwithout needing to redesign or retrain the AST model. Specifically, theFrequency-wise Transformer Encoder (FTE) consists of transformers with selfattention along the frequency axis to address multiple overlapped audio eventsissue in a single clip. The Local Gated Recurrent Units Decoder (LGD) consistsof nearest-neighbor interpolation (NNI) and Bidirectional Gated Recurrent Units(Bi-GRU) to compensate for temporal resolution loss in the pretrained AST modeloutput. Experimental results on DCASE2022 task4 development set havedemonstrated the superiority of the proposed AST-SED with FTE-LGD architecture.Specifically, the Event-Based F1-score (EB-F1) of 59.60% and Polyphonic Sounddetection Score scenario1 (PSDS1) score of 0.5140 significantly outperform CRNNand other pretrained AST-based systems.

AST-SED: An Effective Sound Event Detection Method Based on Audio Spectrogram Transformer