soundwalk.Explore the source ↗

The working experience · 1 minute 17 seconds

Give the world a minute.
Bring back a sound.

A local sound suggestion. A listening direction you choose. One small detail, in your own words.

English narration & captions · real browser interactionsLicensed shoreline sample · scripted observation · timer excerptsWatch on YouTube ↗

A little field trip for your ears.

  1. Prepare the local model, then capture five seconds of your surroundings.
  2. Choose one sound to follow. Read its invitation and let the page dim for a minute.
  3. Write what you noticed. Save your field note or take home an image.

At your desk, Try a shoreline sample takes you through the same listening ritual using a credited sea-wave recording.

Inside this film.

Recorded from the working application in Chromium. YAMNet classifies the licensed sample locally; the complete sixty-second rehearsal runs before the scripted observation is entered. The edit shows excerpts of that minute.

The card keeps its sample label and the words “Scripted example”. Open the exported card →

The application and model work offline after the readiness message appears. This film is streamed online. How SoundWalk works →

English transcript

A wave. A bird nearby. A gap in the traffic. SoundWalk gives you one small reason to stop and listen.

Outdoors, you record five seconds. Here, I’m using the credited shoreline sample. The open YAMNet model runs in this browser and suggests water.

You choose what to follow. This invitation asks you to pick out one small splash within the larger wash of sound.

Then the page dims for sixty seconds. Read once, look up, and listen. This film skips ahead through the minute.

When you return, write one thing you noticed. Here is a scripted example. The card keeps the model’s suggestion, your choice, and your words.

Save an image, or keep your field notes on this device. After preparation, the model works offline. Try SoundWalk, and bring back a sound.

Film credits. Original application, artwork, and browser recording by Blucca with Codex. The English voice is synthesized. Sea-wave audio: HerbertBoland’s “Branding_kort.wav”, via ESC-10 clip 1-28135-A-11, CC BY 3.0; looped and mixed during the listening sequence. Preparation, interactions, and the listening interval are edited for length. Full asset credits · Download the film.