Upload video, audio and images, or import a dataset or a public corpus. Bring the ethograms, codings and annotations you already have from BORIS, ELAN, Praat, Raven, SLEAP or DeepLabCut.

Anima Pluralis: animal behaviour research workspace
Each animal has a world of their own, with their own capacities, experiences, interests and suffering. Anima Pluralis exists for the people trying to understand those worlds, through science, technology, art, and where they meet.

Who we are
We are a non-profit initiative that builds a workspace with the tools and models researchers need to study animal behaviour and communication. Ethology, bioacoustics, linguistics, ethnography, philosophy, computational methods and the arts all have a place here.
The workspace
Everything your research needs, from labeling, to access to models and tools, to the final export, in one place. Choose a step to look closer.
02Sync
Put recordings made at the same time on one clock: several cameras, a microphone or hydrophone, GPS or a collar. The Workspace proposes the offsets, for example by matching the audio, and you confirm them.
03Add context
Every recording keeps its location and time. Add what matters for interpretation, such as weather, season or group composition, and the relations around each moment, such as kinship, social bonds or the people present.
04Explore
Cluster calls, gestures or movements by similarity to see the repertoire in your data: what recurs and what is rare. Search by example or by description, and name the groups you recognise.
05Identify
Keep a roster of known individuals and follow them across recordings, sites and seasons. The Workspace proposes matches and finds more footage of each animal. An identity holds only once you confirm it.
06Label
Code behaviour from your ethogram as events and bouts for each animal, or draw boxes around calls on the spectrogram. Colleagues code blind in their own passes, then check their inter-rater agreement and settle the differences.
07Analyze
Start from your research question and build an Analyzer around it. Choose the models it runs through, such as Perch v2, BatDetect2, NatureLM-audio, SpeciesNet, SAM 3, DeepLabCut, SLEAP, RTMPose, Molmo, Qwen3-VL or Gemma, and run it on your dataset to find every moment your question needs, as clips, events and measurements.
08Review
Check a random sample of detections, each beside its source, and get a precision estimate for the whole run. Confirm, correct or reject. A suggestion becomes a label only through your verdict.
09Compare
Which method works best for your species and setting? Freeze your labels as a reference, test models and settings on the same recordings, and compare precision and recall side by side.
10Export
Take your work wherever it goes next: tables for Excel, R or Python, files for BORIS, Raven, DeepLabCut or SLEAP, and video clips of each moment or one compilation per recording, with faces blurred if needed. Every export records the models, versions and settings behind it.

Work together
We designed Anima Pluralis for collaboration, between disciplines and between colleagues.
- Invite your group
- Bring colleagues, students and supervisors into the same dataset, each with their own role.
- Share the work
- Give colleagues recordings to code, model suggestions to check, or identity matches to confirm.
- Label blind
- Several people code the same recordings without seeing each other's labels. Then you compare, see where you agree, and talk through the differences.
- Tag each other in interesting moments
- Spot a call, a gesture or an encounter worth discussing? Leave a note on it and ask a colleague to take a look.
- Work across disciplines
- An ethnographer and a computer scientist can study the same dataset, each with their own methods, and learn from each other's insights.
What stays yours
Your recordings stay with you. The models run on a computer you choose, and only the people you invite can see your data.
That computer needs a graphics card (GPU). You can use one in your lab or at your university, or rent one online when you have a lot of footage.
Need help with the technical side? Our computer scientists can help you set up the computer and build your Analyzer. They only see your recordings on request, with your permission.
Can go to
- The computer you choose
- The people you invite
- Our team, on request
Never goes to
- AI companies
- Training other models

Works with about 70 extensions, among them:
BORIS, ELAN, Praat, Raven, SLEAP, DeepLabCut, CVAT, Label Studio, Perch, YAMNet, NatureLM-audio, SpeciesNet, MiewID, SAM, MMPose, Molmo, Qwen-VL, Gemma, V-JEPA, AudioMoth, GUANO, Camtrap DP, Movebank, GPX, and the one you ask for.
Add extensions
for what your
research needs
Every study has its own questions, and no workspace can foresee them all. When something is missing, such as a measurement, an export format or a correction for your footage, describe it in your own words. The Assistant builds it as an Extension: a small add-on that works like every other tool in the Workspace.

Ask
Describe what you need: what goes in, what should come out, and how you'll know it works.
Build
The Assistant writes the code and checks that it runs, in a sandbox with no network access.
Review
An owner or administrator of your organisation decides whether it's installed.
Test
Measure how well it works on your own footage, like any other method. Every result notes which version produced it.
The Atelier
In the Atelier, we connect animal behaviour and communication researchers with artists and filmmakers who share a fascination for the more-than-human world.
For researchers
Would you like to be connected with an artist or filmmaker? Tell us about your study, and we will introduce you to someone from our network.
Get in touch →For artists and filmmakers
Do you make work about animals and their worlds? Join our pool and we will let you know when a research group is looking for someone like you.
Join the pool →A spectrogram and a film can tell the same story in different ways.
Together, they help more people see an animal's world.
