March 1, 2026
Robots with reverb? Drama ensues
Show HN: Audio Toolkit for Agents
AI sidekicks learn to mix: devs cheer, tinkerers want live‑sound chops
TLDR: A new audio toolkit lets AI assistants edit and analyze music with one‑click chains, from trimming to “master track.” The crowd’s hyped, pushing for easy hosting and dreaming of live‑sound control via mixing‑board protocol OSC, splitting buzz between studio automation now and stage‑ready AI soon.
Hacker News just handed AI assistants a mixing desk. SAS Audio Processor dropped a bundle of 25 studio tricks — think trim, normalize, auto‑detect song key, even turn a melody into piano notes (MIDI) — all plug‑and‑play for chatty code helpers like Claude via a simple tool bridge called MCP. Translation: your AI can now say “hold my reverb” and actually do it. The repo is live for the curious sas‑audio‑processor and its plan‑runner DeclarAgent.
The comments? A vibe. Builders immediately schemed: enmerk4r wants this “in an A2A agent,” aka bots helping other bots, while Oatcake21 yelled “ship it” to a hosted service so agents can recommend it to regular users. hosaka cheered the broader trend of tools built for agents from day one. Then bradleyy tossed in a spicy curveball: could this act over OSC (Open Sound Control, the language many mixing boards speak) to help “amateur live sound”? Cue a mini split: studio automation fans drooling over the one‑click “master track” chain vs. weekend sound techs dreaming of an AI that rides faders at the church gig.
No flame war, but plenty of playful swagger. Jokes flew about AI interns becoming AI DJs, and the unofficial feature request was clear: a big Make It Sound Good button. If this thing learns live sound, the memes might start mixing themselves.
Key Points
- •SAS Audio Processor offers 25 MCP-exposed audio tools covering processing, effects, analysis, MIDI, and composite pipelines.
- •All tools accept WAV input and produce structured JSON output; the CLI emits line-delimited JSON and JSON errors.
- •Setup for Claude Code includes installing DeclarAgent (Go), installing sas-audio-processor (Python), and adding MCP server config.
- •Other MCP clients (Claude Desktop, Cursor, Windsurf, Copilot) use the same MCP configuration; DeclarAgent reads YAML plans and runs the sas-processor CLI.
- •Composite tools (e.g., master-track, sample-prep, tempo-match, full-analysis, melody-extract) chain multiple operations for common tasks.