wave-share

wave-share, created by Georgi Gerganov, is a proof-of-concept for establishing a WebRTC connection between nearby devices using sound instead of the internet or a signaling server. It targets situations where two browsers on different devices, each with a microphone and speakers, want to exchange files directly over the local network without installing an app or relying on existing network infrastructure for the initial handshake.

WebRTC requires peers to exchange connection details through a process called signaling, but the specification does not define how that exchange happens. This project performs signaling entirely through audio tones: one device encodes its WebRTC session data into sound and broadcasts it, a nearby device captures and decodes those tones, and responds with its own encoded answer, after which a direct WebRTC connection forms over the local network for actual data transfer.

The underlying data-over-sound technique was later extracted into a standalone library called ggwave, which can be embedded into other projects. wave-share itself is written in C++ and compiled to run in the browser, and is released under the MIT license, serving mainly as a demonstration of serverless, peer-to-peer signaling rather than a polished file-sharing product.