Sound Explorer

A spectrogram viewer and audio player for exploring field recordings, built by LOM. Load a WAV file, see its spectrogram, zoom into time-frequency regions of interest, and play them back at different speeds -- slowing down a 384 kHz bat call or speeding through hours of dawn chorus.

Try it now: mrkva.github.io/sound-explorer

No install needed. Works on desktop and mobile. All processing happens locally in your browser -- your files never leave your device.

Quick start

1. Open mrkva.github.io/sound-explorer
2. Drag & drop a WAV file onto the page, or click Open Files
3. Scroll to zoom in/out on the spectrogram
4. Click on the spectrogram to seek to that position
5. Press Space to play/pause
6. Change Play as speed to slow down ultrasonic content (e.g. 0.25x makes a 384 kHz recording audible)
7. Select a region by Shift-clicking, then Export it as a WAV file

Live audio input

1. Click Live to capture audio from your microphone or sound card
2. Click Rec to record a segment
3. Click Stop Live -- the recording loads automatically for playback and export
4. Use Play, Speed, and Export just like with any WAV file

Keyboard shortcuts

Key	Action
Space	Play / Pause
Left / Right	Seek -1s / +1s
Shift + Left / Right	Seek -10s / +10s
Scroll wheel	Zoom in / out
S	Zoom to selection
T	Trim to selection
U	Untrim (restore full view)
L	Toggle loop
Esc	Clear selection

Features

• Spectrogram -- parallel FFT across all CPU cores, tile cache for instant scrolling, six colormaps (viridis, magma, inferno, grayscale, green, hot), linear/logarithmic frequency scale, adjustable brightness/contrast and dynamic range, FFT sizes from 128 to 32768
• FFT window functions -- Hann (default for files), Blackman-Harris (default for live), Hamming, and Flat-top
• Tape-speed playback -- change playback speed (0.0625x--4x) with proportional pitch shift, like speeding up or slowing down a tape. Makes ultrasonic content audible or lets you skim through long recordings
• Live audio input -- capture audio from a microphone or sound card and view the spectrogram in real time. Select input device, adjust time window (2--30s), record segments and export as WAV with BWF timecode metadata
• Time-frequency navigation -- scroll to zoom, drag to pan, click to seek, frequency range presets (Full, Birds, Voice, Low, Mid), overview minimap for quick navigation
• Selection and export -- select a region, loop it, export as WAV. Preserves BWF timecode. Can export at altered speed for time-expanded recordings. Trim view to selection
• Annotations -- label time regions with notes, displayed as overlays on the spectrogram, export/import as JSON
• Multichannel -- mono downmix, per-channel view, split L|R display for stereo files
• VU meter -- per-channel level metering with peak hold, dBFS scale
• Spectrum analyser -- real-time frequency spectrum display in a sidebar panel with log-frequency axis, save and label snapshot lines for comparison, adjustable frequency range, resizable panel, fullscreen mode
• PNG export -- export spectrogram or spectrum analyser as publication-quality PNG with frequency/amplitude axes, time labels, wall-clock row (when available), settings info, and Sound Explorer branding
• BWF/timecode -- parses Broadcast Wave Format metadata for wall-clock timestamps, displayed on spectrogram axis and preserved through export
• Session metadata -- metadata forms (location, equipment, conditions, notes) with iXML and FRM sidecar support
• Dark/light mode -- toggle with the moon/sun button, saved between sessions
• Offline support -- works offline as an installable PWA after first visit

Supported formats

Format	Bit depth	Sample rates
PCM WAV	16-bit, 24-bit, 32-bit	Any (tested up to 384 kHz)
IEEE Float WAV	32-bit	Any
WAVE_FORMAT_EXTENSIBLE	All above	Any
BWF (Broadcast Wave)	All above	Any

Only WAV files are supported. If you have MP3, FLAC, or other formats, convert them to WAV first (e.g. using FFmpeg: ffmpeg -i input.mp3 output.wav).

Browser support

Works in Chrome, Firefox, Edge, and Safari. Chrome recommended for best performance with large files and live audio input.

Two versions

Sound Explorer comes in two versions:

• Web -- runs in the browser, no install needed. Drag and drop WAV files, works offline as a PWA. Best for quick exploration and live audio input.
• Desktop (Electron) -- handles large files and multi-file sessions. Load an entire night of recordings as one continuous zoomable spectrogram with wall-clock time navigation.

Both versions share the same core: FFT computed in parallel Web Workers, tile-cached spectrogram rendering, tape-speed playback, and colormap-based visualization.

Desktop-only features

• Multi-file sessions: load a folder of WAVs stitched into one timeline via BWF timecode
• Subsampled overview mode for 30+ minute recordings
• iXML metadata read/write, Field Recording Metadata (FRM) sidecar support
• Audio output device selection
• Batch annotation export
• Timecode offset correction (-12h to +12h)

Running locally

Web version

cd apps/web
python3 -m http.server 8765
# open http://localhost:8765

Desktop version

cd apps/desktop
npm install
npm start

See apps/desktop/README.md for build and usage details.

Documentation

• Quick Start Guide -- get up and running in 5 minutes
• Architecture -- module overview, data flow, shared code
• Web App -- web-specific features, PWA, mobile UX
• Rebuild Spec -- detailed specification for recreating the desktop app

Repository structure

sound-explorer/
  docs/            <- shared documentation
  apps/
    desktop/       <- Electron version (large files, multi-file sessions)
    web/           <- browser version (GitHub Pages, PWA)

Both versions share the same FFT engine (Cooley-Tukey radix-2) and colormap code, kept as local copies in each app's source folder.

How it was made

Vibecoded with Claude Code. No frameworks, no bundlers, no npm (except Electron), no TypeScript, no build step. Plain ES modules served as static files.

License

MIT