The system takes a sound signal and breaks it into smaller parts, or segments. Each segment is processed to create a normalised representation of the sound in the time domain. This is like a map of the sound over time.
This map is then fed into a trained neural network which generates scores and probabilities for each type of sound event in each segment. This is like guessing what type of sound each segment is and how sure it is about the guess.
This clever kit can distinguish a doorbell ding from a baby's wail, a pooch's yap from a cough, and even the tinkle of breaking glass—talk about eavesdropping
Now, this neural network they've developed is not just any eavesdropper. It's a precision instrument that assigns scores and probabilities to every sound it detects, akin to a discerning judge in a choir competition. To demonstrate its prowess, it refines these numbers to provide you with the likelihood of a burglar's presence or a cat's misadventure with a vase.
This gizmo isn't just for catching the odd midnight snack raider. It's got grander designs. It'll be on guard in your bright house for the sound of glass going smash, ready to ring the alarm if some blighter tries to break in. And if the little one starts bawling, it'll be on it faster than you can say "nappy change."
In the sickbay, it'll listen to your ticker and your coughs, playing doctor like it's got a stethoscope and a medical degree.
Vole tells us that it can also warn about Mother Nature's temper tantrums. That's right, if there's a storm brewing or the earth is about to wobble, this AI will be the first to let you know. If only it could predict the lottery numbers, we'd all be quids in.