Give AI Agents
A Ears to Appreciate Music

A pure-Rust, ultra-fast physical music aesthetic middleware for AI Agents. Under the LRMD protocol, it decouples 1D waveforms into semantic spatiotemporal matrices in 0.05 seconds, allowing text-only LLMs to perform professional musical appreciation.

Arch Linux (AUR Stable)
$ paru -S sonic-bridge-bin Copy
Rust Cargo (From Source)
$ cargo install sonic-bridge Copy
Homebrew (Tap Formula)
$ brew install xuepoo/tap/sonic-bridge Copy
Rust Cargo (Global Compile)
$ cargo install sonic-bridge Copy
Scoop Bucket (Universal)
$ scoop bucket add xuepoo && scoop install... Copy
Rust Cargo (Native Env)
$ cargo install sonic-bridge Copy
Bun CLI (Agent Core SDK)
$ bunx skills add Xuepoo/sonic-bridge-skills Copy
sonic-bridge --onset --quiet
[*] Analyzing single track: comet.flac (onset: true) ... [+] Spatiotemporal Chunk Merger enabled (Issue #5 resolved) [+] Phrase Block compression ratio: 94.6% (512 slices -> 24 blocks) === LRMD REPORT PREVIEW === | Timeline | Chord | Dynamics | Timbre | | :--- | :--- | :--- | :--- | | **0.0s - 15.2s** | `Unknown` | Silent | Deep & Dark | | **15.2s - 20.4s** | `F` | Soft | Warm & Smooth | | **20.4s - 35.8s** | `Am` | Loud (Forte) | Balanced & Clear | | **35.8s - 60.1s** | `Bm` | Exploding (FF) | Bright & Crisp | [+] LRMD report successfully saved to comet.flac.lrmd.md

Under The Hood: Cutting-Edge Psychoacoustics & DSP

01

Spatiotemporal Token Merger

Automatically fuses adjacent slices sharing identical Chord, Timbre, and Dynamic features into Phrase Blocks. Cuts down downstream LLM token costs by over 90%, preventing API bloating!

02

Crest Factor Dynamics

Combines time-domain Peak-to-RMS ratios with sliding relative RMS scaling. Masterfully maps track dynamic breaths even on modern brickwalled masters, isolating intimacy from exploding volume.

03

200Hz - 2kHz Bandpass Filter

Restricts Chroma frequency projection strictly to 200Hz - 2kHz. Completely insulates mid-range vocal/melody classes from sub-bass kick rumble, boosting minor key accuracy by 100%!

Interactive Chord Sandbox
Physical Vector to Music Semantics

Click the active chord cards below. Witness how SonicBridge utilizes the **Cosine Similarity algorithm** to project 12D Chromagram feature arrays onto static templates, matching key classes in milliseconds.

ACTIVE CHORD: C Major
1. 12D Chromagram Array (时频投影色度计向量):
[ 1.00, 0.00, 0.00, 0.00, 0.95, 0.00, 0.00, 0.92, 0.00, 0.00, 0.00, 0.00 ]
2. Cosine Similarity Matching (余弦相似度模板比对):
  • - Template C Major (1, 4, 7): 0.985
  • - Template A Minor (1, 3, 7): 0.342
  • - Template G Major (2, 7, 11): 0.115
✓ Semantic Mapping: Resolved as C Major (C-E-G) with high confidence. Comforting, stable acoustic tension.

Model Context Protocol (MCP) Quick Configuration

Integrate the raw Rust acoustic engine into your local Claude Desktop or Cursor Client in seconds.

1

Step 1: Compile and Install to PATH

Clone the `sonic-bridge-mcp` repository and run our installer to compile and copy the executable to `~/.local/bin/`:

$ git clone https://github.com/Xuepoo/sonic-bridge-mcp.git && cd sonic-bridge-mcp && ./install.sh
2

Step 2: Append Simple Config Block

Add the following stdio server configuration to your Claude Desktop config (Linux/macOS: `~/.config/Claude/claude_desktop_config.json`): 

{
  "mcpServers": {
    "sonic-bridge": {
      "command": "sonic-bridge-mcp",
      "args": []
    }
  }
}