Web app · Lifestyle
SoundSpotter
Teach your browser the doorbell — then read back everything it heard
Train any sound in about 30 seconds from five microphone samples, get a per-sound vibration pattern the moment it fires, and scroll a timestamped event log: the oven at 7:14pm, the dog at 3:02am. Runs entirely in this tab, free, no install.
Stored in your browser only. Nothing you do in the web version is sent to our servers. Clear your browser data and it's gone — so export a backup from the in-app menu if you want to keep it.
In your browser
What you can do in SoundSpotter on the web
Train any sound in 30 seconds
Five short microphone samples build a 32-band acoustic fingerprint on the spot — your doorbell, the oven beep, the cat at 5am. The audio itself is never recorded or stored, only the shape.
Six pre-loaded starters
Doorbell, Microwave, Smoke Alarm, Knock, Baby Cry and Dog Bark ship in the library with the app's own haptic defaults — train each one on your actual appliance.
Per-sound vibration patterns
One short pulse, three short pulses, or SOS (long-short-long) — the same three patterns as the iPhone app, played through the browser's vibration motor with a preview button.
Live listening with a match meter
A running spectrum, a level bar and a best-match read-out show exactly what the recogniser is hearing, with strict / balanced / loose sensitivity and a 4-second cooldown per sound.
Persistent event log
Every detection is stamped with time, sound and confidence, grouped by day, charted per day and by hour of day, and exportable as CSV.
Nothing leaves this browser
No account, no server, no upload. Sounds, fingerprints and events live in this browser's local storage and can be exported as a JSON backup.
Screens
Every screen, rebuilt for the browser
-
1
Sound library
Last heard, today's detection count, and every tracked sound with its training state and last firing.
-
2
Sound detail
Training state, acoustic fingerprint, vibration pattern, listening switch and every firing of one sound.
-
3
Train a sound
Name it, pick a vibration, then capture five microphone samples with a live spectrum and progress bar.
-
4
Vibration pattern
Pick one short pulse, three short pulses or SOS for a sound, with a timeline and a preview buzz.
-
5
Listen
Live recognition with a pulsing wave mark, level meter, best-match read-out and detections as they fire.
-
6
Events
The full detection log grouped by day, with detections per day, which sound fired, and an around-the-clock chart.
-
7
Settings
Sensitivity, vibration and chime on detection, library counts, CSV and JSON export, restore starters, erase.
Try it with sample data: Trains Doorbell, Microwave, Smoke Alarm and Dog Bark, adds two custom sounds (Oven timer, Cat at 5am) and fills three weeks with 33 realistic detections. You can start fresh at any time.
Ready? It's free.
Sign in once and every All Things AI web app opens instantly. No install, no subscription, no data leaving your browser.
Your say
Is SoundSpotter useful? Tell us what to build next.
Sign in to vote.
Tell us what to build next.
Sign in to leave a comment or vote — it takes a few seconds.
Sign in to commentNo comments yet — be the first to say what this app should do next.
Questions
Is the web version of SoundSpotter free?+
Yes. Sign in with a free All Things AI account and use every screen — no subscription, no in-app purchases.
Where is my data stored?+
Only in this browser's local storage. Nothing you enter is sent to our servers, so we can't see it, back it up, or restore it. Clearing site data, a private window, or another device starts empty. Export a backup from the in-app menu to keep it.
Do I need an account?+
You can read about the app without one, but you must sign in to open it. Sign-in is free and takes a few seconds with Google.
How is the web version different from the SoundSpotter iPhone app?+
Live listening uses your microphone through the browser and only while this tab is open and on the Listen screen — the iPhone app keeps a background classifier alive when the app is closed. Recognition here is a 32-band spectral fingerprint matched by cosine similarity, not the iPhone's CoreML transfer-learning classifier, so it works best on distinctive tones (doorbells, beeps, alarms) and less well on speech-like sounds. The six pre-loaded starters arrive untrained: without Apple's built-in 300-sound classifier they each need one 30-second training pass on your actual doorbell or alarm. Vibration uses the browser Vibration API, which iPhone Safari does not support — patterns are saved and drawn on screen there but not felt. No notifications: the iPhone app can banner-alert you on a detection, while a web page can only show the hit while the tab is open. Everything is stored in this browser only — clearing site data erases it, and it does not sync to another device.