🌊 songsee — FFT so pretty, your ears will be jealous.

Features

• 9 visualization modes: spectrogram, mel, chroma, hpss, selfsim, loudness, tempogram, mfcc, flux
• 6 color palettes: classic, magma, inferno, viridis, gray, clawd
• Auto-contrast: per-panel percentile normalization for readable heatmaps
• Combine modes: stack multiple visualizations in one grid image
• Universal input: WAV, MP3, or anything ffmpeg can handle
• Fast: native Go, no Python dependencies
• Flexible output: PNG or JPEG, customizable dimensions

Install

brew install steipete/tap/songsee

go install github.com/steipete/songsee/cmd/songsee@latest

Quick Start

# Basic spectrogram
songsee track.mp3

# Mel spectrogram with magma palette
songsee track.mp3 --viz mel --style magma

# All 9 modes combined
songsee track.mp3 --viz spectrogram,mel,chroma,hpss,selfsim,loudness,tempogram,mfcc,flux

# Custom output
songsee track.mp3 --viz hpss,chroma --style inferno -o viz.png --width 2560 --height 1440

Visualization Modes

Mode	Description
spectrogram	Time × frequency magnitude
mel	Perceptual frequency scale
chroma	12-bin pitch class
hpss	Harmonic vs percussive separation
selfsim	Self-similarity matrix
loudness	Volume over time
tempogram	Tempo variation
mfcc	Timbre fingerprint
flux	Spectral change detection

Palettes

classic · magma · inferno · viridis · gray · clawd 🦞

Options

--output        Output path (default: input name + extension)
--format        jpg or png (default: jpg)
--width         Output width (default: 1920)
--height        Output height (default: 1080)
--window        FFT window size (default: 2048)
--hop           Hop size (default: 512)
--min-freq      Minimum frequency in Hz
--max-freq      Maximum frequency in Hz
--start         Start time in seconds
--duration      Duration in seconds
--style         Palette name
--viz           Visualization list (repeatable or comma-separated)

---

Built by @steipete