You can hear that something is wrong with your animal. This tells you what kind of wrong.
Record ten seconds. Animind AI measures how worked up the animal is and whether the sound is a distressed one or a happy one — the two things a vocalisation reliably carries — and gives you a plain sentence, the measurements behind it, and an honest confidence.
The sentence is that measurement written in plain language, not a decoding of words — there are no words to decode. When the recording is too noisy or the signal too weak to support an answer, it says so instead of guessing.
Garden, another dog present
Play initiation
The signals are consistent with an invitation to play rather than with conflict.
High confidence
Alternative interpretation
Attention seeking directed at a personWeakly indicated
Evidence
- Repeated play bowVideo · Forequarters low, hindquarters raised · weight 0.86
- Wide pitch variation in barksAudio · F0 spread of 214 Hz · weight 0.71
- High, loose tail carriageVideo · Above the spine, relaxed · weight 0.62
- Another dog oriented toward the subjectContext · Reported by observer · weight 0.49
Contradicting signal
Brief piloerection along the shoulders
Animind estimates how an animal sounds, and compares it against how that animal usually sounds. It is not a translation, and no accuracy figure has been established for it — this platform has not yet been tested against enough confirmed outcomes to state one. It does not diagnose, and it does not replace a veterinarian.
What is inside
Everything an account gets, and where each of it lives.
Recording is the part people come for. The rest is what makes a recording worth having: a baseline to compare it with, a schedule that knows what is due, and a plan for the day something goes wrong.
Every day
- HomeRecord something, and see what has happened since you last looked.
- Record & translateHold the button, get a sentence back. The quickest way in.
- AnimalsEveryone you live with, and how each of them usually sounds.
- HistoryEverything you have recorded, and what each one said.
- TrendsHow they have been over weeks, not just this moment.
- Detailed observationThe full observation form, with context fields.
Care
Emergency
Converse
What this is for
The questions you already ask out loud.
Most of them are ordinary, and ordinary is the point — knowing how your animal usually sounds is what makes the day it sounds wrong stand out. Some of them matter a great deal: a missed illness, a bite that was signalled first. The same reading answers both.
“What is my cat actually asking me for?”
Cats direct most of their vocalising at people and invent their own vocabulary for it. The same call can mean the bowl, the door, or nothing but company, and you learn the difference by living with them.
What you get: Whether the call is a keyed-up one or a settled one, and where it sits against the calls this cat has made before — so you can tell the demand from the greeting.
“Did she enjoy that, or was she tolerating it?”
The new visitor, the bath, the other dog at the park. Animals put up with a great deal politely, and the difference between fine and enduring is easy to miss when you want the answer to be fine.
What you get: A reading you can take twice — during and after — and compare, plus the evidence behind each so you can judge it yourself.
“Is this a hurt noise or an annoyed noise?”
The two sound alike to us and mean completely different things. Pain that reads as grumpiness is pain that goes untreated for weeks.
What you get: Whether the call is high-arousal and negative, with the pitch, harshness and call rate that led to that reading shown underneath.
One reading, four channels
Frequency alone tells you almost nothing.
A bark is not a word and pitch is not a meaning. What carries information is the combination: the acoustic shape of the call, the posture it came from, the situation around it, and how this individual usually behaves.
Audio
Signal processing over the recording, bounded by what the species actually produces.
- · Pitch contour and stability
- · Harmonic vs noisy structure
- · Rhythm and repetition
- · Call count and duration
Vision
Posture, orientation, movement and the body-language channels that matter for the species.
- · Body orientation and gaze
- · Ear, tail and crest position
- · Movement intensity
- · Distance to people and animals
Context
What was happening around the animal, in a closed vocabulary the engine can actually weigh.
- · Before, during and after
- · Food, strangers, offspring
- · Location and confinement
- · Time, weather, group size
History
How this individual normally behaves, so unusual can be told apart from normal.
- · Per-feature baselines
- · Deviation from normal range
- · Confirmed past outcomes
- · Known triggers
Personalisation
The same call means different things from different animals.
Every profile builds its own behavioural baseline. Once it exists, the interesting question stops being “what does this call mean” and becomes “is this unusual for this animal”.
Maturity is capped deliberately. A baseline sharpens an interpretation; it never turns one into a certainty, and the interface says which is which.
Personalisation maturity
New profile
No baseline. Species-level patterns only, which is a poor substitute for knowing an individual.
Early learning
A handful of observations. Patterns are visible but not yet stable enough to weigh heavily.
Developing baseline
Enough range for deviations to start meaning something.
Strong baseline
Stable across varied contexts. Unusual behaviour becomes distinguishable from normal.
Longitudinal profile
Months of observations across seasons. Slow changes become visible.
Supported species
Twelve species today, built to add more as data allows.
Each species carries its own acoustic range, its own meaningful body-language channels, its own excluded behaviours and its own honest statement of what consumer hardware can and cannot capture.
Domestic dog
Canis lupus familiaris
The most studied companion species in behavioural science, and the one where audio, posture and context most reliably combine into an interpretable signal.
Domestic cat
Felis catus
Highly individual vocal repertoires make cats the clearest demonstration of why a per-animal baseline matters more than a species-level average.
African bush elephant
Loxodonta africana
Long-distance infrasonic coordination makes elephants the clearest case for treating context and group movement as first-class signals rather than audio alone.
Lion
Panthera leo
Territorial broadcast calling and coordinated pride behaviour, where the interesting signal is usually the group response rather than the call.
Rhesus macaque
Macaca mulatta
Dense social signalling with rapid facial expressions, where the value is in the interaction structure rather than any single call.
Chimpanzee
Pan troglodytes
Structured, multi-phase calls layered over rapid gestural communication — the species where temporal sequence carries the most meaning.
Western lowland gorilla
Gorilla gorilla gorilla
Quiet, close-range communication punctuated by rare high-intensity displays — a species where absence of vocalisation is itself informative.
Horse
Equus ferus caballus
A prey species whose most informative channels are postural and facial rather than vocal, and where separation is the dominant driver of calling.
Cattle
Bos taurus
A welfare-monitoring species where the value is longitudinal: deviation from an individual or herd baseline matters more than any single call.
African grey parrot
Psittacus erithacus
A vocal-learning species where mimicry must be separated from species-typical calls before any interpretation is attempted.
Common bottlenose dolphin
Tursiops truncatus
The clearest illustration of honest capture limits: without a hydrophone the primary communication channel is simply not observed, and the product says so.
Grey wolf
Canis lupus
Long-range territorial chorus calling where the group structure of the signal matters far more than any individual howl.
Who uses it
One platform, four very different jobs.
Pet owners
Build a real record of how your animal behaves, and notice change early rather than in hindsight.
For pet owners →Researchers
Projects, subjects, ethograms, annotation with inter-rater reliability, and exports with provenance.
For researchers →Veterinarians
Longitudinal behavioural evidence and baseline deviation to bring into a consultation.
For veterinarians →Conservation teams
Multi-animal operations, group interaction analysis and governance that survives an audit.
For conservation teams →Limits, stated up front
The things this product will not do.
A probabilistic system earns trust by being explicit about where it stops. These are not buried in a policy page — they ship with every result.
- It does not translate. There is no sentence an animal said, and the product never presents one.
- It does not diagnose. Pain-related outputs are observational flags for a veterinarian, nothing more.
- It does not encourage approach. Wild-species results carry distance guidance and suppress contact advice.
- It plays back only a cue you recorded for your own animal — never another animal’s call, which is the method the playback literature uses to provoke alarm responses.
- It returns “insufficient evidence” rather than a confident guess, and does so often.
Record where the animal is
The iOS and Android apps capture audio and video, run quality checks on device, hold drafts offline and upload when a connection returns. Field observations do not wait for signal.
Work with us on the science
Every species adapter states where its behavioural priors came from. We are looking for field groups, ethologists and bioacousticians willing to hold those priors to account and to help build annotated datasets with proper consent and provenance.