ToS: A Team of Specialists ensemble framework for Stereo Sound Event Localization and Detection with distance estimation in Video
Davide Berghi, Philip J. B. Jackson
TL;DR
The paper tackles 3D sound event localization and detection with distance estimation in video (3D SELD) by proposing a Team of Specialists (ToS) ensemble. ToS combines three specialized subnetworks—spatio-linguistic, spatio-temporal, and tempo-linguistic—to jointly reason across semantic, spatial, and temporal dimensions, leveraging OWL-ViT for visual embeddings and CLAP for audio semantics. Across real and synthetic data on the DCASE2025 Task 3 Stereo SELD dataset, ToS achieves state-of-the-art performance, with substantial gains in F1 metrics and improved localization (DOA error) and distance estimation, at the cost of increased computational overhead due to ensembling. The work highlights the value of modular, specialist-driven ensembles for multimodal SELD and points to future directions including pre-training curricula and task-aligned specialization to further boost performance.
Abstract
Sound event localization and detection with distance estimation (3D SELD) in video involves identifying active sound events at each time frame while estimating their spatial coordinates. This multimodal task requires joint reasoning across semantic, spatial, and temporal dimensions, a challenge that single models often struggle to address effectively. To tackle this, we introduce the Team of Specialists (ToS) ensemble framework, which integrates three complementary sub-networks: a spatio-linguistic model, a spatio-temporal model, and a tempo-linguistic model. Each sub-network specializes in a unique pair of dimensions, contributing distinct insights to the final prediction, akin to a collaborative team with diverse expertise. ToS has been benchmarked against state-of-the-art audio-visual models for 3D SELD on the DCASE2025 Task 3 Stereo SELD development set, consistently outperforming existing methods across key metrics. Future work will extend this proof of concept by strengthening the specialists with appropriate tasks, training, and pre-training curricula.