gibberlink
Basic Information
GibberLink is a demonstration and developer-facing project that shows two conversational AI agents switching from natural language to a compact data-over-sound protocol when they detect the other side is an AI. The repository documents the viral demo that won a hackathon and drew press coverage and provides the API and supporting files needed to reproduce the behavior. It ties conversational agents from a speech-focused provider to the ggwave sound-encoding library so agents can encode messages into an audio-based channel and decode them on the other end. The README points to a browser demo, a recorded video of the interaction, authors and contact information, and a wiki with reproduction steps so others can run or extend the experiment.
Links
Stars
4594
Github Repository
App Details
Features
A site and recorded demo showing two independent conversational agents switching from English to a compact ggwave sound protocol when they identify each other as AI. Integration examples that pair conversational agent prompts with the ggwave data-over-sound library. An API in the repository that enables agents to encode and decode messages into the sound protocol. Documentation noting the demo scenario (a hotel booking caller and receptionist) and pointers to reproduction steps. Mention of the hackathon award and media coverage providing context and outreach. References to derived work and an invitation for contributors who build on the project.
Use Cases
This repository helps developers and researchers reproduce and experiment with an agent-to-agent communication channel that uses audio as a data transport. It provides a concrete example of prompting conversational agents to detect peer AI and switch to a more efficient encoded language, plus an API to integrate the ggwave sound protocol with speech-based AI. The materials make it easier to explore agent coordination, covert or compressed inter-agent messaging, and multimodal communication experiments. It also includes reproduction instructions and a browser demo to validate behavior, enabling others to extend the concept or adapt the protocol for their own agent systems.
-
$$$$ -
$$$$