Revisiting SSL for sound event detection: complementary fusion and adaptive post-processing

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Cui, Hanfang, Song, Longfei, Li, Li, Xu, Dongxing, Long, Yanhua
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915464005812224
author Cui, Hanfang
Song, Longfei
Li, Li
Xu, Dongxing
Long, Yanhua
author_facet Cui, Hanfang
Song, Longfei
Li, Li
Xu, Dongxing
Long, Yanhua
contents Self-supervised learning (SSL) models offer powerful representations for sound event detection (SED), yet their synergistic potential remains underexplored. This study systematically evaluates state-of-the-art SSL models to guide optimal model selection and integration for SED. We propose a framework that combines heterogeneous SSL representations (e.g., BEATs, HuBERT, WavLM) through three fusion strategies: individual SSL embedding integration, dual-modal fusion, and full aggregation. Experiments on the DCASE 2023 Task 4 Challenge reveal that dual-modal fusion (e.g., CRNN+BEATs+WavLM) achieves complementary performance gains, while CRNN+BEATs alone delivers the best results among individual SSL models. We further introduce normalized sound event bounding boxes (nSEBBs), an adaptive post-processing method that dynamically adjusts event boundary predictions, improving PSDS1 by up to 4% for standalone SSL models. These findings highlight the compatibility and complementarity of SSL architectures, providing guidance for task-specific fusion and robust SED system design.
format Preprint
id arxiv_https___arxiv_org_abs_2505_11889
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Revisiting SSL for sound event detection: complementary fusion and adaptive post-processing
Cui, Hanfang
Song, Longfei
Li, Li
Xu, Dongxing
Long, Yanhua
Audio and Speech Processing
Artificial Intelligence
Sound
I.5.4; I.2.10; H.5.5
Self-supervised learning (SSL) models offer powerful representations for sound event detection (SED), yet their synergistic potential remains underexplored. This study systematically evaluates state-of-the-art SSL models to guide optimal model selection and integration for SED. We propose a framework that combines heterogeneous SSL representations (e.g., BEATs, HuBERT, WavLM) through three fusion strategies: individual SSL embedding integration, dual-modal fusion, and full aggregation. Experiments on the DCASE 2023 Task 4 Challenge reveal that dual-modal fusion (e.g., CRNN+BEATs+WavLM) achieves complementary performance gains, while CRNN+BEATs alone delivers the best results among individual SSL models. We further introduce normalized sound event bounding boxes (nSEBBs), an adaptive post-processing method that dynamically adjusts event boundary predictions, improving PSDS1 by up to 4% for standalone SSL models. These findings highlight the compatibility and complementarity of SSL architectures, providing guidance for task-specific fusion and robust SED system design.
title Revisiting SSL for sound event detection: complementary fusion and adaptive post-processing
topic Audio and Speech Processing
Artificial Intelligence
Sound
I.5.4; I.2.10; H.5.5
url https://arxiv.org/abs/2505.11889