earshot

Earshot gives coding agents the ability to hear, review and generate sound, running on CPU only.

Why it matters

Coding agents can now build videos, demos and games but can't hear the result. earshot gives them a way to listen to audio, judge it, and generate sound effects in code, all locally on CPU.

Highlights

• describe, match, rank and find commands turn audio into words and scores an agent can reason about, such as finding the best key-press moment in a long recording.
• timeline and sheet produce a second-by-second listening transcript and an annotated PNG (frames, events, spectrogram, loudness, sync markers).
• critique scores a launch video against a profile of 7 professional launch films and returns timestamped fixes.
• forge synthesizes sounds from code recipes and improves them by listening; fit turns a recording into a recipe you can vary.
• Runs as a CLI, a Python API and an MCP server. CPU only, no GPU; an optional Gemini critique runs only with your key and explicit consent.

Taken from the project's own README, docs or launch page. Claims about speed or accuracy are the maker's unless stated otherwise.

Worth knowing: Each CLI call loads models (about 20-40 s); the README recommends the MCP server or Python API for repeated use.

Quick facts

Links

• Repository: https://github.com/Leonxlnx/earshot
• Release notes: https://github.com/Leonxlnx/earshot/releases

Related ranking

See how established tools compare in Best AI Coding Agents & IDEs. New releases aren't scored until we review them against the rubric.

Checked 2026-10-17 from the tool's own pages. Prices and star counts change, so confirm on the official site. Listing here is not a ranking or an endorsement.