Bioacoustic Models & LLMs
Machine learning models, foundation language models (LLMs), neural audio codecs, and bioacoustic classifiers applied to non-human acoustic sequence analysis.
Perch
Lab / Provider: Google Research
Open-source machine learning model for bioacoustic feature extraction and classification across extensive avian and wildlife soundscapes.
DolphinGemma
Lab / Provider: Google DeepMind
Compact bioacoustic sequence model (~400M parameters) developed by Google DeepMind in collaboration with the Wild Dolphin Project and Georgia Tech. Applies the SoundStream neural audio codec tokenizer with Gemma-derived sequence modeling to predict acoustic token transitions in recordings of wild Atlantic spotted dolphins.
SoundStream
Lab / Provider: Google Research
Neural audio codec developed by Google Research (Zeghidour et al., 2021) utilizing fully convolutional encoder-decoder architecture with Residual Vector Quantization (RVQ) and adversarial STFT discriminators to discretize audio into multi-rate acoustic token streams.
AVES
Lab / Provider: Earth Species Project
Animal Vocalization Encoder based on Self-Supervised Learning (AVES), developed by Earth Species Project (Hagiwara, 2023). Self-supervised audio transformer encoder pre-trained on diverse animal bioacoustic recordings for downstream acoustic classification and representation learning.
BirdNET
Lab / Provider: Cornell Lab of Ornithology / Chemnitz University of Technology
Deep convolutional neural network developed by the Cornell Lab of Ornithology and Chemnitz University of Technology. Evaluated on 984 species in its seminal 2021 study (Kahl et al.), with modern platform versions (BirdNET-Analyzer) supporting over 6,000 wildlife taxa.