Phylogenetic signal in marine mammal and bird vocalizations captured by audio foundation models: the limited benefit of domain-specific pretraining
Researchers tested four large audio models on recovering phylogenetic distance from species vocalizations, finding that the models can capture structure beyond their training labels. The study used recordings from 32 marine mammal species and an independent set of bird species.
- Large pretrained audio models can capture phylogenetic signal in species vocalizations
- The models can pick up structure beyond their training labels
- Domain-specific pretraining may have limited benefits for certain tasks
- The study provides new insights into the structure of audio embeddings
A recent study published on arXiv explored the ability of large pretrained audio models to capture phylogenetic signal in marine mammal and bird vocalizations. The researchers probed four models, including AST, CLAP, BEATs-bio, and BirdNET, with a downstream task that none of them were trained on: recovering phylogenetic distance from species vocalizations.
The study used recordings from the Watkins Marine Mammal Sound Database, which includes 1,754 recordings from 32 marine mammal species. The results showed that the geometry of the embedding space tracks the tree of life, indicating that the representation is picking up something deeper than the labels the model was optimized for.
The researchers also ran Mantel tests across two independent radiations, providing further evidence for the models' ability to capture phylogenetic signal. The findings have implications for our understanding of the structure of audio embeddings and the potential benefits of domain-specific pretraining.
The study's results suggest that learned audio embeddings can encode structure that nobody explicitly told them to encode, highlighting the potential for AI models to uncover new insights into evolutionary relationships.
can use these findings to improve the design of audio models
the study highlights the potential for AI to uncover new insights into the natural world
- phylogenetic signal
- the pattern of evolutionary relationships between species
- Mantel test
- a statistical test used to compare the similarity between two matrices
As Duke Health implements AI, oversight initiatives try to ensure ethical practices - The Duke Chronicle
Katy ISD sets new framework on artificial intelligence use in classrooms - ABC13 Houston
Artificial Intelligence Is Transforming Immigration Adjudications: What Every Employer and Applicant Needs to Know - WR Immigration
Adoption of artificial intelligence outpaces training in field epidemiology programs, new survey finds - CIDRAP
agentic artificial intelligence needs shared memory - SiliconANGLE
BusinessCursor makes its biggest India push yet ahead of SpaceX acquisition with localized pricing
AI code editor Cursor identifies India as its third-largest market and is launching localized pricing and expanded sales efforts there.
AI ToolsBeyond System Prompts: Enforcing Policy & Action Boundaries in Enterprise AI Agents
This article argues that system prompts are insufficient for controlling enterprise AI agents, proposing deterministic tool adapter validation, risk classification, and human-in-the-loop gates as more robust solutions.
SecurityI Tested 7 AI OSINT Agents on My Own Digital Footprint - Here's What They Found in 4 Minutes
A test of 7 AI OSINT agents revealed significant personal data in just 4 minutes. The agents were able to uncover information despite the tester's attempts at good opsec.
AI ToolsResurrecting the Panasonic WJ-MX50 in WebGPU
A developer has successfully ported a 1990s video mixer to WebGPU, showcasing the versatility of modern web technologies.
Microsoft unveils AI security tools it says outperform competing platforms
Microsoft has unveiled a suite of AI security tools that it claims outperform competing platforms, while also being more cost-effective.
AI ToolsNine Months of Nagging, Zero Reading
A GitHub Action has been developed to analyze nine months of AI commit history, revealing insights into AI's writing habits.