Search papers, labs, and topics across Lattice.
The DS@GT ARC team tackled the BirdCLEF+ 2026 challenge by establishing a robust supervised baseline using a combination of a frozen Perch v2 backbone, a trained HGNetV2-B0 network, and a non-bird prototypical head, achieving a leaderboard score of 0.936. They then explored the efficacy of token-based representations by comparing codec representations from neural audio codecs with semantic embeddings from foundational models. The findings reveal that while traditional supervised methods perform well, token-based encoders also show competitive potential in the context of bioacoustic classification tasks.
Token-based representations can rival traditional supervised methods in detecting animal vocalizations, challenging the dominance of established architectures.
This paper details the DS@GT ARC team's approach to BirdCLEF+ 2026, multi-label detection of animal vocalizations in soundscapes from the Pantanal wetlands. The 2026 edition adds about an hour of labeled soundscapes, shifting the task toward supervised pipelines fit to the labeled set. First, we build a competitive supervised baseline that ensembles a frozen Perch v2 backbone, a trained HGNetV2-B0 sound-event-detection network, and a non-bird prototypical head, reaching a private leaderboard score of 0.936 at rank 1894 within a 90-minute CPU budget. Second, we ask whether token-based representations can compete, contrasting codec representations from neural audio codecs against semantic representations from foundational embeddings. We compare two bioacoustic specialist models against four token-based encoders trained on AudioSet. The repository for this work can be found at https://github.com/dsgt-arc/birdclef-2026.
