songsee

songsee wraps the songsee CLI binary to give your AI agent audio visualization capabilities. Here's the typical flow:Point songsee at an audio file (MP3, WAV, or other formats via ffmpeg): songsee track.mp3Choose one or more visualization panels with --viz — e.g., spectrogram,mel,chroma,hpss,selfsim,loudness,tempogram,mfcc,flux. Multiple panels are rendered as a grid.Customize the output with flags for color palette (--style), image dimensions (--width / --height), FFT settings (--window / --hop), frequency range (--min-freq / --max-freq), and time slicing (--start / --duration).Export as jpg or png to a file or stdout.WAV and MP3 are decoded natively; other formats are handled automatically if ffmpeg is available on the system.

Multiple Visualization Types — Supports spectrogram, mel filterbank, chroma, HPSS, self-similarity matrix, loudness, tempogram, MFCC, and spectral flux panels.Multi-Panel Grid — Combine any mix of visualizations into a single image with one command.Time Slicing — Extract and visualize any segment of audio using --start and --duration flags.Color Palettes — Choose from five styles: classic, magma, inferno, viridis, and gray.Stdin Support — Pipe audio directly into songsee for seamless integration with shell pipelines.Flexible Output — Export as JPEG or PNG with configurable width and height.FFT Control — Tune window size, hop length, and frequency range for precise analysis.Broad Format Support — Native MP3/WAV decoding; other formats handled via ffmpeg if installed.

songsee CLI — The core binary that performs audio analysis and rendering. Installed automatically by this skill via Homebrew (steipete/tap/songsee). Required.ffmpeg — Enables decoding of audio formats beyond MP3 and WAV (e.g., FLAC, OGG, AAC). Optional — the skill works without it for MP3 and WAV files.

Music Production Review — Visualize frequency content, dynamics, and rhythm features of a mix to identify mastering issues or compare versions.Audio Research & Analysis — Generate multi-panel feature grids (chroma, MFCC, tempogram) for academic or ML dataset inspection.Podcast & Speech QA — Quickly check a recorded segment for noise, clipping, or silence by slicing and visualizing specific time windows.Batch Pipeline Integration — Pipe audio from other tools into songsee via stdin to automate visualization in CI or data processing workflows.