Quelle  SKILL.md   Sprache: unbekannt

---
name: songsee
description: Generate spectrograms and feature-panel visualizations from audio with the songsee CLI.
homepage: https://github.com/steipete/songsee
metadata:
  {
    "openclaw":
      {
        "emoji": "��",
        "requires": { "bins": ["songsee"] },
        "install":
          [
            {
              "id": "brew",
              "kind": "brew",
              "formula": "steipete/tap/songsee",
              "bins": ["songsee"],
              "label": "Install songsee (brew)",
            },
          ],
      },
  }
---

# songsee

Generate spectrograms + feature panels from audio.

Quick start

- Spectrogram: `songsee track.mp3`
- Multi-panel: `songsee track.mp3 --viz spectrogram,mel,chroma,hpss,selfsim,loudness,tempogram,mfcc,flux`
- Time slice: `songsee track.mp3 --start 12.5 --duration 8 -o slice.jpg`
- Stdin: `cat track.mp3 | songsee - --format png -o out.png`

Common flags

- `--viz` list (repeatable or comma-separated)
- `--style` palette (classic, magma, inferno, viridis, gray)
- `--width` / `--height` output size
- `--window` / `--hop` FFT settings
- `--min-freq` / `--max-freq` frequency range
- `--start` / `--duration` time slice
- `--format` jpg|png

Notes

- WAV/MP3 decode native; other formats use ffmpeg if available.
- Multiple `--viz` renders a grid.