A developer has released a small browser-based audio experiment designed to play back sound it hears after a delay of roughly two seconds, a deliberately irritating feedback effect intended to discourage people from playing loud media in shared spaces.
The project, called STFU, was published on GitHub by Pankaj Tanwar. In the repository description, Tanwar says the idea arose at Mumbai airport after another traveler watched short-form videos at full volume and laughed loudly. Unwilling to confront the person directly, he looked for a technical response instead.
Tanwar says he asked Anthropic’s Claude to create the application and received a working version from a single prompt. The resulting tool uses the Web Audio API, the browser interface that allows web applications to capture, process and output sound. In this case, the program listens through a device microphone and returns the captured audio approximately two seconds later.
That delay is the project’s essential mechanism. Immediate playback can merge with the original sound or create familiar acoustic feedback. A delayed copy instead produces an audible echo that competes with whatever the nearby device is playing. Tanwar describes the effect informally and does not present controlled testing, usage data or a scientific account of how consistently it changes another person’s behavior.
The repository is similarly candid about the project’s origins. Tanwar says it was first called “make-it-stop,” but he adopted the STFU name after seeing that another developer, identified as Tim Darcet, had built something similar. He characterizes the program as a side project created with the Web Audio API and permits others to use it as they wish.
Several practical and ethical constraints follow from the design even though the supplied material does not document them in detail. The application necessarily depends on microphone access and on a device capable of playing the captured sound loudly enough to be noticed. The evidence does not state whether audio is recorded or retained, how permissions are presented, which browsers are supported, or whether the project has a formal software licence. Those points should be checked before treating it as more than an experiment.
The project also does not establish that delayed playback is a safe or effective substitute for speaking to someone or seeking help from venue staff. Its documented contribution is narrower: it packages browser audio capture and delayed output into a compact demonstration motivated by a common public-space annoyance.
STFU is therefore best understood as a playful proof of concept and an example of rapid AI-assisted prototyping. The repository supports Tanwar’s account of what he built and why, but not broader claims about neuroscience, behavior modification or success in real-world use.
