Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper addresses the fragmentation in sound effects datasets caused by varying tagging schemes and metadata structures, which complicates SFX classification and generation. By introducing a modular dataset relabeling framework based on the Universal Category System (UCS), the authors achieve high automatic conversion rates of existing datasets and facilitate the merging of previously siloed data. The resulting EnvSound-UCS dataset, comprising 58,057 sound clips from multiple sources, exemplifies the framework's effectiveness and provides a unified resource for future research.
A unified sound effects dataset that resolves tagging inconsistencies and enhances comparability across research efforts is now available, featuring over 58,000 clips.
Sound effects (SFX) datasets and libraries often employ distinct tagging schemes, taxonomies, and metadata structures. This creates challenges for research on SFX classification and generation because incompatible taxonomies lead to siloed datasets that might require individualized approaches, result in non-comparable outcomes, and prevent data merging strategies. We propose a modular dataset relabeling framework that adopts the Universal Category System (UCS), an industry-standard hierarchical taxonomy for sound effects, as a shared structural foundation. This open-source framework enables us (i) to convert tags of existing datasets to UCS with a rule-based multi-stage pipeline and conflict resolution to achieve high automatic conversion rates, (ii) to suggest a stratified dataset split for the new labels, and (iii) to combine multiple datasets. To showcase the practical utility, we introduce the EnvSound-UCS dataset, a publicly available unified UCS-compliant dataset of environmental sounds with 58,057 sound clips from three sources: AudioSet, FSD50K, and ESC-50.