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 work, we aim to improve the expressive capacity of waveform-baseddiscriminative music networks by modeling both sequential (temporal) andhierarchical information in an efficient end-to-end architecture. We presentMuSLCAT, or Multi-scale and Multi-level Convolutional Attention Transformer, anovel architecture for learning robust representations of complex music tagsdirectly from raw waveform recordings. We also introduce a lightweight variantof MuSLCAT called MuSLCAN, short for Multi-scale and Multi-level ConvolutionalAttention Network. Both MuSLCAT and MuSLCAN model features from multiple scalesand levels by integrating a frontend-backend architecture. The frontend targetsdifferent frequency ranges while modeling long-range dependencies andmulti-level interactions by using two convolutional attention networks withattention-augmented convolution (AAC) blocks. The backend dynamicallyrecalibrates multi-scale and level features extracted from the frontend byincorporating self-attention. The difference between MuSLCAT and MuSLCAN istheir backend components. MuSLCAT's backend is a modified version of BERT.While MuSLCAN's is a simple AAC block. We validate the proposed MuSLCAT andMuSLCAN architectures by comparing them to state-of-the-art networks on fourbenchmark datasets for music tagging and genre recognition. Our experimentsshow that MuSLCAT and MuSLCAN consistently yield competitive results whencompared to state-of-the-art waveform-based models yet require considerablyfewer parameters.

MuSLCAT: Multi-Scale Multi-Level Convolutional Attention Transformer for Discriminative Music Modeling on Raw Waveforms