R. Hahnloser

WhaleSonic: Recordings of cetacean vocal behaviour with hydrophone arrays in Northern Norway

WhaleSonic provides a raw acoustic dataset of cetacean vocalizations in Northern Norway, supporting multichannel algorithms for source separation and communication analysis.

Positive Transfer of the Whisper Speech Transformer to Human and Animal Voice Activity Detection

This paper uses a pre-trained speech transformer for voice activity detection in both human and animal audio, improving accuracy with longer context and generalizing to new species.

Benchmarking nearest neighbor retrieval of zebra finch vocalizations across development

This study benchmarks nearest neighbor retrieval for zebra finch vocalizations, finding high performance in adults but lower in juveniles, with improved results using fixed-size templates and Spearman distance.

Extracting extended vocal units from two neighborhoods in the embedding plane

The paper introduces a semi-supervised method to segment and cluster vocalizations by identifying two density-based regions in low-dimensional embeddings, achieving lower false positive error rates.

Full-duplex acoustic interaction system for cognitive experiments with cetaceans

The paper describes a full-duplex acoustic system for interacting with cetaceans in the wild, achieving 18 dB echo suppression and enabling bidirectional vocal communication.

Learning to perform auditory discriminations from observation is efficient but less robust than learning from experience

This study compares observer learning and experimenter learning in zebra finches, finding that observer learning is faster but less robust than learning from experience.