Eight seconds of recording. Nine layers of analysis. One honest readout.
Every recording passes through a sequence of steps. Each step is inspectable. Nothing is hidden.
Hold your phone 10 centimetres from your dog's nose. Press the button. Eight seconds of sound are captured at 48,000 samples per second — the same quality as a studio microphone.
The first 1.5 seconds capture only room noise — the hum of air conditioning, traffic, conversation. This "fingerprint" is subtracted from the rest of the recording, leaving only the dog's sounds.
The cleaned sound is broken into individual events — breaths, coughs, snores. Each event is measured: pitch, harshness, duration, frequency content. These measurements are compared against thousands of reference recordings from veterinary databases.
The result is not a diagnosis. It is a structured readout: what was heard, how it compares to normal for this breed, and whether it warrants a conversation with a veterinarian.
From normal breathing to pathological sounds. Each event is identified, measured, and contextualised.