GLORB

Glorb

Glorb is a browser-based generative audio app powered by a Python synthesis engine. Every sound is built from scratch in real-time — no samples, no loops, pure mathematics.

Choose a playback mode to control how events are arranged, then a preset to choose their sound source. The two layers can be combined independently.

What you hear

• Abstract blips and blops that chirp, sweep and pop like a living circuit board
• 8-bit chiptune melodies — square waves stepping through arpeggios with a crisp retro bite
• Rain, fire and insects — procedural ambient textures that breathe and shift
• Sci-fi effects — laser shots, warp drives, energy shields and matter transporters
• Deep bass hits — 808 kicks with falling pitch, subsonic thuds and growling harmonics
• Metallic bells — tubular bells, marimbas, singing bowls and Balinese gamelan gongs
• Lo-fi nostalgia — vinyl crackle, cassette hiss, tape dropout and power-line buzz
• Dial-up modem screech — DTMF tones, FSK handshakes, the full connect sequence
• Generative arpeggios stepping through minor, pentatonic and modal scales
• Digital glitch — bit-crushed aliasing, grain stutter, buffer freeze and DC spikes
• Horror textures — tremolo drones, eerie sweeps, ghost whispers and jump scares
• Underwater soundscapes — sonar pings, bubble streams and hydrophone hiss
• Foley and haptics — footsteps, paper rustle, clicks, taps and motor rumble
• Cats and birds — formant-shaped meows, pulsing purrs, melodic peeps, whistles and warbles

Synthesis Engine

The backend runs on NumPy and SciPy in Python. All waveforms are computed as floating-point arrays at 44 100 Hz stereo, then exported as 16-bit or 32-bit WAV and streamed to the browser.

• Additive synthesis — multiple sine partials summed with individual envelopes
• FM synthesis — frequency modulation for metallic, bell, and sci-fi timbres
• Subtractive synthesis — filtered white/pink noise via Butterworth bandpass and lowpass filters
• Granular synthesis — stochastic grain clouds: small overlapping windows of noise or tone, scattered in time and pitch
• Physical modelling — mass-spring-damper resonators for plucked strings, bells, and impact transients
• Karplus-Strong — delay-line feedback for plucked and percussive string sounds
• Noise shaping — shaped random walks for breath, wind, rain, and texture layers

Signal Processing

• ADSR envelopes — attack / decay / sustain / release on every voice
• Reverb — multi-tap Schroeder comb + allpass reverberator
• Echo / delay — feedback delay lines with variable decay
• Soft saturation — tanh waveshaper for analogue warmth
• Pitch wobble — LFO-modulated frequency drift for organic movement
• Stereo panning — randomised left/right placement per sound event

3 Playback Modes

• Continuous — free-running generative soundscapes with organic timing
• Arp — arpeggiated events with scale, root, waveform, and BPM controls
• Groove — rhythmically quantised events using five groove templates and adjustable BPM

28 Presets

• Glorb — abstract blip/bloop electronic tones
• Retro — 8-bit chiptune square and pulse waves
• Nature — rain, fire, insects, and forest textures
• Sci-Fi — phasers, warp drives, and laser bursts
• Haptic — tactile vibration pulses and clicks
• Radio — AM/FM static, morse, and transmission artefacts
• UI Pack — notification pings, confirm tones, and interface clicks
• Foley — footsteps, paper, keys, and everyday impacts
• Underwater — sonar pings, bubble trains, deep pressure tones
• Weather — thunder, lightning crackle, rain intensity layers
• Bell — marimba, tubular bells, and resonant metal strikes
• Bass — deep sub-bass pulses and 808-style low-end hits
• Glitch — buffer corruption, bit-crush artefacts, digital stutter
• Pinball — flippers, bumpers, drain, and mechanical clicks
• Horror — dissonant drones, breath, reversed tails, and tension builders
• Granular — scattered grain clouds across noise and pitched sources
• Lo-Fi — vinyl crackle, tape wow/flutter, degraded telephone bandwidth
• Modem — DTMF tones, FSK handshake, connect screech
• Insects — crickets, cicadas, grasshoppers, and water bugs
• Gamelan — inharmonic metal bars, detuned pairs, gong resonance
• Glass — brittle shards, singing rims, and crystalline resonances
• Clockwork — ticks, ratchets, springs, and tiny mechanisms
• Creature — imaginary chirps, calls, growls, and breath
• Cat — meows, mewls, trills, purrs, and hisses
• Birds — melodic peeps, whistles, trills, warbles, and calls
• Electricity — arcs, transformer hum, relays, and rising charge
• Cave — drips, stones, subterranean wind, and deep rumbles
• Synth — dedicated triangle, saw, or square oscillator voice for Arp playback

Visualiser Backgrounds

Seven audio-reactive backgrounds selectable from the settings drawer: Dots (dot-grid pulse), Wave (oscilloscope), Particles (floating field), Rings (expanding circles), Matrix (falling dot trails), Blobs (three frequency-reactive orbs), and None.

Browser Playback

Audio is decoded and played via the Web Audio API. Playback loops continuously until stopped. The waveform display and visualiser use Canvas 2D and a real-time AnalyserNode FFT.

Sharing

Click the 🔗 button to copy a shareable link. The URL preserves the playback mode, preset, playback-specific controls, knobs, duration, quality, theme, and background.

Privacy

The public Glorb website uses Umami for anonymous, aggregate usage statistics. It records generation attempts and outcomes, playback, ten-second listening engagement, downloads, sharing, and interactions with the optional support request so the project can be improved. Glorb does not send knob values, URLs, filenames, audio, exact error messages, or its private support-usage estimate. Analytics is disabled for local installations, repository forks, preview deployments, and the Railway backend.

Offline / Terminal

Clone the repo and install dependencies, then run the local server:

pip install -r requirements-server.txt
python server.py          # open http://localhost:5000

Or generate WAV files directly from the terminal without the browser:

python main.py                        # 10 s glorb → blipblop.wav
python main.py -d 30                  # 30 s glorb
python main.py --play                 # render & play immediately
python main.py --list                 # list playback modes and presets

Batch-render selected standalone presets or export a Freesound upload sheet:

python generate.py --all --duration 60 --freesound

Help keep GLORB generating

Free · open source · no ads

Every sound you generate is synthesized on a paid server. If GLORB has been useful to you, a small contribution helps cover those costs and keeps it freely available to everyone.

Supporting is entirely optional. Nothing is locked, and your sound is never interrupted.