earshot
earshot (Leonxlnx/earshot): Gives a coding agent a way to judge audio it cannot hear: a local CPU tool (CLAP and AST models) ranks sound effects or music against a text description, finds the cleanest moment in a recording, scores whether audio sounds recorded or code-synthesized, measures loudness, builds a timeline of a video's soundtrack and A/B-compares mix changes. It can also make sounds in code and refine them by listening, and ships a sound-effect bank. The skill tells the agent never to say how audio sounds without quoting an earshot reading. Works as a CLI or an MCP server; audio stays on the machine except with the optional Gemini-based llm command. MIT.
Stars
2.3k
Votes
—
Score
33.5238
Source: Leonxlnx/earshot · skills/earshot/SKILL.md
npx skills add Leonxlnx/earshot --skill earshot
Related skills
| 01 | algorithmic-artanthropics/skills | Media | 182k | 5260.1 |
↗ | anthropics/skills · Media · 5260.1 |
| 02 | audio-jinglenexu-io/open-design | Media | 102k | 5009.8 |
↗ | nexu-io/open-design · Media · 5009.8 |
| 03 | algorithmic-artdavila7/claude-code-templates | Media | 33k | 4515.0 |
↗ | davila7/claude-code-templates · Media · 4515.0 |
| 04 | braglatent-spaces/brag | Media | 17k | 4224.8 |
↗ | latent-spaces/brag · Media · 4224.8 |
| 05 | brag-slimlatent-spaces/brag | Media | 17k | 4224.8 |
↗ | latent-spaces/brag · Media · 4224.8 |