Custom visual codes for physical-to-digital interactions.
MusiCode is inspired by Spotify Codes, but replaces the fixed barcode-like format with a flexible visual recognition workflow. A symbol, object, logo, or image can become a code that opens a link when scanned through an Apple Shortcuts workflow.
The idea is simple: instead of generating a special machine-readable graphic, use multimodal recognition to match a real-world visual against a reference image.
- Turn a custom visual into a link trigger
- Use existing images, logos, objects, or printed designs as codes
- Avoid a rigid barcode format
- Create physical-to-digital interactions with only GitHub, Shortcuts, and an Apple Intelligence-capable device
- Make links easy to update by editing
url.txt
flowchart LR
A["Reference image in GitHub"] --> B["Shortcut downloads reference"]
C["Camera capture"] --> D["Apple visual model"]
B --> D
D --> E{"Match?"}
E -->|Yes| F["Open URL from url.txt"]
E -->|No| G["Show error"]
img.jpg: reference visual codeurl.txt: destination URLbanner-placeholder.png: project banner
- Fork this repository.
- Replace
img.jpgwith the visual code you want to recognize. - Edit
url.txtwith the destination link. - Copy the raw GitHub URLs for both files.
- Install the Apple Shortcut.
- On first launch, paste the raw file URLs when prompted.
Shortcut:
- Music and playlist links
- Personal business cards
- Posters and event flyers
- Brand activations and merchandising
- Treasure hunts and interactive installations
- Linking physical objects to digital experiences
- Apple cloud model: recommended for the best speed and accuracy
- Apple local model: works with lower accuracy
- ChatGPT Vision: can work, but adds latency and depends on user access
MusiCode depends on Apple Intelligence-capable devices and the Shortcuts app. Check Apple's current compatibility list before using it in production.
MusiCode is a creative prototype. It is useful for experiments, demos, and small physical-to-digital interactions, but recognition accuracy depends on the visual design, lighting, camera framing, and the model used by the shortcut.
You may not copy this project and claim authorship unless you have substantially modified it. Credits to the original author are required.