Hands-free voice translation
Real-time, voice-activated species detection and translation in 5–7 seconds. No manual input required in the field.
Wild Animal Voice Engine
Listen deeply. Decode the voices of Earth.
WAVE is an AI-powered bioacoustics platform that translates, visualizes, and researches animal communication in real time — whale songs, bird calls, elephant rumbles, and wolf howls, decoded in 5–7 seconds.
12,000+
species covered
5–7s
real-time translation
< 30s
custom model training
Built forField researchersConservationistsEducatorsOpen-source developers
Watch
See WAVE in motion
A short look at how WAVE listens across oceans, forests, and savannas — and turns raw sound into meaning.
Features
Bioacoustic AI, built for real research
Powered by the NatureLM-audio and Perch 2.0 foundation models, validated against the BEANS benchmark.
Custom acoustic fingerprinting
Train custom ML models in under 30 seconds with as few as 5 audio samples — identify individuals, populations, and dialects.
Deep integration & workflow hub
A central command center with automated workflows, unified search, and cross-species pattern discovery across your datasets.
Advanced visualizations
Interactive Voronoi diagrams for behavioral correlations and Superlet transforms for ultra-high-resolution spectrograms.
Edge-ready deployment
Under-2MB models run on-device with TinyML for offline monitoring at remote stations and on hydrophones.
Land & marine coverage
One platform for terrestrial and underwater bioacoustics — from rainforest canopies to deep-ocean soundscapes.
Technology
The Technical Triad
Three layers turn raw sound into structured, comparable meaning.
Bio-logging & edge AI
Hydrophones, loggers, and sensors capture signal from dense environmental noise, with inference at the edge.
The Superlet Transform
High-resolution signal processing adapted from neuroscience. It revealed the Chagos pygmy blue whale song is pulsed, not continuous.
Foundation models
NatureLM-audio and Perch 2.0 map vocal patterns without supervision — models trained on one species can help organize another's.
How it works
From raw sound to structured meaning
Three layers turn an animal's call into something a researcher — or a model — can compare, search, and reason about.
Capture
Bio-logging sensors & edge AI
WAVE starts where the animals are. Hydrophones, acoustic loggers, and on-device inference pull clean signal out of dense environmental noise — on land and underwater — without shipping raw audio back to a server.
- On-device (edge) inference with sub-2MB TinyML models
- Works offline at remote stations and on marine hydrophones
- Real-time species presence across land and sea
Resolve
The Superlet Transform
Adapted from neuroscience, the Superlet Transform produces ultra-high-resolution spectrograms — resolving detail that conventional methods blur. It's how WAVE showed the Chagos pygmy blue whale song is pulsed, not continuous.
- Higher time-frequency resolution than standard wavelet analysis
- Reveals fine structure inside calls (pulses, harmonics, transitions)
- Powers WAVE's Research Mode spectrograms
Understand
Foundation models (NatureLM-audio, Perch 2.0)
WAVE builds on open bioacoustic foundation models that map vocal patterns without supervision. Because they learn shared acoustic structure, a model trained on one species can help organize another's vocalizations.
- Zero-shot species detection across thousands of species
- Unsupervised mapping of vocal repertoires
- Validated against the BEANS benchmark
Video library
Go deeper on the science
Long-form explainers on the ideas and engineering behind WAVE.
Translating Ocean Voices
~8 min
How WAVE extracts meaning from marine soundscapes — separating whale song from ocean noise and mapping it to behavior.
Unlocking the Ocean's Secrets
~7 min
A field-to-model look at underwater bioacoustics: hydrophones, edge AI, and real-time species presence across marine environments.
Sound as Shape: The High-Dimensional Geometry of Animal Communication
~5 min
Why animal vocalizations are best understood as geometry — embedding calls into high-dimensional space to reveal structure across species.
Decoding the Umwelt: Engineering the Wild Animal Voice Engine
~6 min
The engineering behind WAVE — the Technical Triad of bio-logging sensors, the Superlet Transform, and foundation models working together.
Blog
Field notes & research
Plain-language deep-dives on the science, the data landscape, and how WAVE turns sound into meaning.
Acoustic monitoring and the future of protected areas
A protected area is only as good as our ability to know what's happening inside it. Continuous listening is quietly becoming one of the most powerful tools for managing and defending the world's wild places.
Read article →
A short history of listening to animals
From wax cylinders to foundation models, the story of how humans learned to record, visualize, and finally analyze animal sound at scale — and why each leap changed what questions we could even ask.
Read article →
Do animals have names? Signature calls and the puzzle of individual identity
Some dolphins seem to have signature whistles that work a bit like names. It's one of the most tantalizing findings in animal communication — and a perfect case study in how to be excited without overclaiming.
Read article →
The vanishing chorus: insects and the quiet crisis we can hear
Insects are the backbone of terrestrial soundscapes and the base of countless food webs. As their numbers fall, the world is getting quieter — and acoustic monitoring may be one of our best ways to measure the loss.
Read article →
The base-rate problem: why rare species fool even good detectors
A detector that's 99% accurate can still be wrong most of the time when the target is rare. This counterintuitive math is one of the most important — and most ignored — ideas in acoustic monitoring.
Read article →
The hydrophone: the humble instrument that opened the ocean
Before you can analyze a whale's voice, something has to hear it underwater. The hydrophone is the unglamorous device that made marine bioacoustics possible — and its quirks shape everything downstream.
Read article →Use Cases
Where WAVE goes to work
Field research
Wet, dirty, hands-full conditions where manual input is impractical — WAVE listens and labels for you.
Continuous monitoring
Always-on audio streams at monitoring stations, with automated alerts for conservation threats.
Individual identification
Population-specific dialect recognition and rare-species detection with limited training data.
Education & outreach
Live demonstrations with immediate feedback that make interspecies communication tangible.
Resources
The animal-communication data landscape
WAVE connects the field's foundational datasets and archives. Search live across iNaturalist and GBIF, or explore the major public sources below.
Xeno-canto
Public APIXeno-canto Foundation
800k+ recordings
Bird & wildlife sounds, global, community-contributed
Open ↗Macaulay Library
Browse / requestCornell Lab of Ornithology
3.2M+ audio recordings
World's largest animal media archive; 96% of bird species
Open ↗Watkins Marine Mammal Sound Database
Open datasetWoods Hole Oceanographic Institution
15k+ annotated clips, 60+ species
Marine mammal vocalizations, freely downloadable
Open ↗iNaturalist Sounds (iNatSounds)
Public APIiNaturalist
230k+ audio files, 5,500+ species
Citizen-science multi-taxa audio, open archive
Open ↗Earth Species Library
Open datasetEarth Species Project
Multi-modal ML datasets + BEANS benchmark
Preprocessed bioacoustic datasets for machine learning
Open ↗Project CETI
Open datasetCetacean Translation Initiative
Large-scale sperm whale acoustic + behavioral data
Decoding sperm whale communication; open research
Open ↗Animal Sound Archive
Browse / requestMuseum für Naturkunde Berlin
120k+ recordings, ~1,800 bird + 580 mammal species
Third-largest bioacoustic collection worldwide
Open ↗GBIF
Public APIGlobal Biodiversity Information Facility
Billions of occurrence records
Species occurrence & biodiversity data, with API
Open ↗Building a research integration? WAVE is designed to plug into these sources. Talk to us about data access →
Try it
The Animal Sound Translator
Upload an animal sound or name a species. WAVE identifies it, profiles the acoustics, and cross-references validated research data — reference recordings, species facts, and more. Honest by design: it connects real data, it doesn't fake sentence-level translation.
New: the Sound Map arranges clips from multiple species by acoustic similarity — distance means how alike they sound.
Roadmap
Where we're headed
Now
Real-time engine
Voice translation, custom fingerprinting, visualizations, and the integration hub — all live.
Next
Open-source community
Grow the MIT-licensed engine and developer docs on GitHub; welcome contributors.
Later
Agentic swarm
An institutional AGI framework of cooperating specialized agents for ecosystem-scale monitoring.
Q&A
Straight answers about what WAVE can — and can\'t — do
Bioacoustic AI is full of hype. We\'d rather earn trust with honest answers.
Not in the sci-fi sense — and we won't pretend otherwise. WAVE identifies *which species* is calling with real confidence scores, profiles the acoustics, and connects each sound to validated research data. Mapping calls to specific *meaning* is an open scientific frontier; we surface the evidence honestly rather than inventing sentences.
Membership
Join the WAVE
Get early access to the platform, priority support, and a voice in the roadmap. Reserve your membership below.
Researcher
Free
- Core translation engine
- Community support
- MIT open-source engine
Most popular
Pro
Membership
- Custom fingerprinting
- Workflow automation
- Priority support
- Early features
Institution
Let's talk
- Multi-site datasets
- Dedicated onboarding
- Custom integrations
Connect
Talk to the WAVE team
Researching a species? Building on the open-source engine? Exploring a partnership or membership? Tell us what you're working on.