What it does
earshot gives AI coding agents a sense of hearing. Agents can now write the code that renders a film, a product demo or a game, but they cannot tell a premium key press from a cheap one, notice that a hit lands 80 ms after the cut, or hear that the music swells in the wrong place. earshot turns sound into evidence an agent can reason about: a description of what a file sounds like in words, how well it fits a description, the best moment inside a long recording, whether it is a real recording or synthesized in code, a second-by-second listening transcript of a whole film, an annotated picture of it, and a scorecard against professional launch films. With that, an agent can make its own sounds in code and improve them by listening, or translate a recording into code it can vary. Version 0.4 ships a bank of 252 sound effects stored as code, such as UI pops, knocks, whooshes and chimes, which render by id in seconds with no models and no network, plus a launch kit that mixes effects against a music bed. It runs locally on CPU as a CLI, a Python API and an MCP server, needs uv and ffmpeg, and downloads about 1.1 GB of open models on first use. MIT licensed.
Who it’s for
Developers and creators who make video, demos or games with coding agents.
Use cases
- Letting an agent check the sound design of a code-rendered product video
- Generating UI sound effects in code for an app or game
- Finding the best moment inside a long recording

