Mantice is a procedural ambient drone generator. The name comes from the Italian word
for organ bellows — the invisible breath engine that gives life to the pipes. Like a mantice,
this tool provides the continuous, living airflow beneath your sound.
Unlike sample-based instruments or DAW plugins, Mantice generates audio from mathematical
processes — every sound is computed in real-time from oscillators, noise, and grain clouds.
No two renders are ever exactly alike (unless you lock the seed).
Architecture
Layers
A Mantice preset consists of 1–5 independent layers, each with its own
frequency, timbre, spatial position, and movement. Layers are mixed together to create
complex, evolving textures. Think of layers as individual instruments in an ensemble —
each contributes a different voice to the whole.
FM Synthesis Engine
Each FM layer uses Frequency Modulation synthesis as its core.
Multiple detuned voices per layer create rich, beating textures. The FM index controls
harmonic complexity — low values produce pure tones, higher values create metallic,
bell-like timbres.
Subtractive Synthesis Engine
Subtractive layers use classic waveform oscillators — sawtooth, square, or triangle —
instead of FM. Two detuned oscillator pairs (±detune cents) create the thick, beating quality
of a Reese bass. A third sub-oscillator one octave below adds weight. The output is then
shaped by the per-layer filter for the full subtractive sound design workflow.
Per-Layer Filter & LFO
Every layer has five filter modes: LP (low-pass), HP (high-pass),
BP (band-pass), Comb, and Formant.
LP/HP/BP are biquad filters with resonance and optional LFO modulation (sine/triangle/square) on the cutoff.
Comb creates metallic ringing via feedforward delay — the Cutoff sets the delay frequency, Feedback controls resonance intensity.
Formant shapes the sound into vowel resonances (A, E, I, O, U) using three parallel bandpass filters at human vocal formant frequencies — ideal for chant-like and throat-singing textures.
Harmonics & Noise
On top of FM, each layer can add harmonic overtones (up to 16 partials
with configurable decay) and filtered noise (white, pink, or brown)
for breath-like textures and organic warmth.
Granular Clouds
Layers can also be granular — extracting tiny overlapping slices
(grains) from real audio recordings. The included samples (singing bowls, gongs,
bowed glass, throat singing, water, wind) become raw material for evolving textures
impossible with pure synthesis. Control grain size, density, pitch spread, and
read position to sculpt the cloud.
Spatial Motion
Each layer has a position in 3D space — horizontal (left/right, front/back)
and vertical (elevation). Layers can move through space using trajectories: orbit, drift,
bounce, or random. This creates an immersive, living soundfield — especially powerful
with headphones.
Chorus
A per-layer chorus effect adds width and shimmer through modulated
delay copies. Slow rates create gentle swaying; faster rates add vibrato-like movement.
Flanger
A per-layer flanger modulates a short delay (comb filter) with an LFO,
creating the classic jet-plane sweeping sound. Rate controls sweep speed, Depth controls
modulation intensity, Feedback controls resonance. Runs sample-by-sample for artefact-free
audio. Applied after Chorus in the per-layer chain.
Phaser
A per-layer phaser uses cascaded all-pass filter stages to create
notch sweeps through the spectrum. Center Hz sets the sweep range, Stages controls the
number of notches (more stages = deeper effect), Feedback adds resonance.
Applied after Flanger in the per-layer chain.
Layer Sub-tabs & Global FX
Layer parameters are organised into four sub-tabs: Synth (oscillator type and tuning),
Filter (filter type + LFO), Space (quadrant, trajectory, elevation),
and FX (chorus → flanger → phaser, distortion). Global processing lives in the
Global FX card below the layers, ordered by signal chain:
Reverb → Shimmer → Earth & Air → Binaural.
Per-layer signal chain: Synthesis → Filter → Distortion → Panner → Chorus → Flanger → Phaser
Global chain: DC Block → Saturation → Reverb → Shimmer → Earth & Air → Master EQ/Comp → Limiter → Binaural
Global Processing
Saturation
A master soft saturation stage using 4× oversampled tanh waveshaping.
Running internally at 4× the output sample rate eliminates aliasing artefacts on harmonically
rich material. Adds analog tape-like warmth and gently compresses peaks. At low values it's
transparent; higher values add harmonic richness and density.
Reverb (FDN + IR)
An 8-line Feedback Delay Network with Hadamard mixing matrix,
per-line damping, and stereo decorrelation. Choose from algorithmic spaces (cathedral, cave, hall,
chamber, plate) or switch to IR mode — real impulse responses convolved
with the signal for authentic acoustic spaces. The reverb glues layers together
into a unified acoustic space.
Shimmer FX
A global pitch-shifted feedback tail applied after the reverb.
The reverb output is pitch-shifted (by a selectable interval — octave up, fifth, fourth, etc.),
then fed back into the reverb input, creating an ethereal upward bloom.
Feedback controls how long the tail sustains. Works best with reverb enabled.
Binaural Beats
Optional binaural entrainment creates a perceived beat frequency
by slightly detuning left/right channels. Frequencies correspond to brainwave bands:
Delta (0.5–4 Hz, sleep), Theta (4–8 Hz, meditation), Alpha (8–13 Hz, focus),
Beta (13–30 Hz, alertness). Requires headphones.
Earth & Air
Earth adds sub-bass tectonic rumble (10–40 Hz) — felt more than heard.
Air adds high-frequency breath and wind turbulence. Together they
extend the frequency range beyond what the layers provide, grounding the drone
in physical space.
Parameter Automation
Any global or per-layer parameter can evolve over the duration of a render or stream.
Open the Automation card and enable any parameter with the toggle.
Each row shows a compact inline breakpoint editor — up to three nodes,
each with a timing position (t%), a value, and a curve shape.
Click the shape label between nodes to cycle through lin (linear),
exp (exponential), and S (S-curve). + mid inserts
a midpoint at the largest gap; − mid removes it. Values before the
first node and after the last node are held constant.
Five global templates — Journey, Arc, Breathe, Meditate, Sunrise —
appear as quick-set pills at the top of the card. One click wires up a coherent
multi-parameter automation so you don't have to start from scratch. Use
✕ clear all to reset.
Automation only runs when a duration is set (Render or timed Stream).
It has no effect in open-ended infinite playback.
Just Intonation (JI)
Enable JI mode in the Synth tab to switch from 12-TET layer roots
to exact rational-fraction tuning. All layer frequencies are derived
from a single tonic via pure intervals (e.g. 3/2 for a perfect fifth, 5/4 for a major
third), so intervals between layers beat at zero — the defining quality of sustained
sacred drone music.
A small ● dot in the layer header confirms JI is active on that layer.
Existing presets with decimal ratios load without change.
Workflow
Preview streams audio in real-time via WebSocket — tweak parameters
and hear changes instantly. Render computes the full duration at high
quality for export. Use the seed to reproduce exact results, or leave
it random for variation.
The Generator creates random presets biased by mood. Choose which
synthesis types to include — FM and/or Subtractive — then click Generate.
Mutate takes the current preset and introduces controlled variation
(use the Amount slider to set how wild).
Presets are stored as YAML files — human-readable, version-controllable, and shareable.
Use Save/Save As to create your own library.
Download Mantice and run it entirely on your machine — no internet required after cloning.
▸ Installation
# Clone the repository
git clone https://github.com/bassimatte/mantice.git
cd mantice
# Install Python dependencies
pip install -r requirements.txt
Requires Python 3.10+ and pip. Works on Windows, macOS, and Linux. No GPU needed.
▸ Launch Web UI
# Start the server and open browser
python main.py --gui
# Or start server only
python -m engine.web_server
# → http://127.0.0.1:8432
▸ Generate WAV Files from Terminal
# Render all presets to WAV (default 60s each)
python main.py
# Render a specific preset
python main.py --name "Breathing Cathedral"
# Custom duration (seconds)
python main.py --name "Om" --duration 300
# Export as FLAC or OGG
python main.py --name "Solar Choir" --format flac
# Set a seed for reproducible output
python main.py --name "Gong Bath" --seed 42
# Generate a random preset and render it
python main.py --generate
# Generate with a mood bias
python main.py --generate --mood dark
# Mutate an existing preset
python main.py --mutate "Breathing Cathedral" --amount 0.3
# List all available presets
python main.py --list
▸ Real-Time Preview (Speakers)
# Preview a preset through speakers
python main.py --preview --name "Cavern of Echoes"
# Infinite drone (Ctrl+C to stop)
python main.py --preview --infinite
Comb: feedforward FIR comb — metallic ringing and spectral combing. Formant: 3-band vowel resonance (A=800/1200/2500, E=400/2000/2800, I=270/2300/3000, O=570/850/2500, U=380/950/2200 Hz). Filter is applied after synthesis, before chorus/flanger/phaser on all layer types.
Layer — Distortion
Setting
Range
Default
Description
Drive
0.0–5.0
0.0 (off)
Waveshaping intensity — 0 bypasses the stage
Type
soft / hard
soft
Soft = tanh (warm, analog-like); Hard = hard clip (aggressive)
Applied after the filter stage. Adds harmonic saturation; at high drive values creates heavy distortion. Keep Drive below 1.0 for subtle warmth.
Layer — Harmonics
Setting
Range
Default
Description
Harmonics
1–16
4
Number of overtone partials per voice
Harmonic Decay
0.0–1.0
0.7
Amplitude decay per harmonic (0.7 = each partial 70% of previous)
How long the shimmer tail sustains — higher = longer bloom
Pitch-shifted feedback tail fed after reverb — creates an ethereal upward bloom. Works best with Reverb enabled. Applied in the global chain after Reverb, before Earth & Air.
Binaural Beats
Setting
Range
Default
Description
Method
detune / carrier
detune
Detune splits voices L/R; Carrier adds a sine pair
High-pass cut-off — removes sub-rumble below this frequency
Bass Gain
−12–+12 dB
0
Low-shelf boost/cut around Bass Freq
Bass Freq
40–500 Hz
100
Low-shelf transition point (visible when Bass ≠ 0)
Lo Mid Gain
−12–+12 dB
0
Bell boost/cut in the low-mid range
Lo Mid Freq
100–1000 Hz
250
Bell centre frequency (visible when Lo Mid ≠ 0)
Lo Mid Q
0.3–5.0
1.0
Bell width — higher Q = narrower, more surgical (visible when Lo Mid ≠ 0)
Hi Mid Gain
−12–+12 dB
0
Bell boost/cut in the high-mid range
Hi Mid Freq
500–8000 Hz
2500
Bell centre frequency (visible when Hi Mid ≠ 0)
Hi Mid Q
0.3–5.0
1.0
Bell width (visible when Hi Mid ≠ 0)
Air Gain
−12–+12 dB
0
High-shelf boost/cut for presence and sparkle
Air Freq
2000–16000 Hz
10000
High-shelf transition point (visible when Air ≠ 0)
5-band master EQ. Bass and Air are shelving filters (±6 dB/oct). Lo Mid and Hi Mid are constant-Q peaking (bell) filters — same formulas used in professional DAW EQs (Audio EQ Cookbook). Freq and Q sub-controls are hidden when a band's gain is at 0 to keep the UI clean. Filters run in series: Low Cut → Bass → Lo Mid → Hi Mid → Air.
Master Compressor
Setting
Range
Default
Description
Threshold
-40–0 dB
0
Level above which compression begins
Ratio
1–20
2.0
Compression ratio (e.g. 4:1)
Attack
1–500 ms
50
Time to reach full compression
Release
10–2000 ms
200
Time to recover after signal drops
Knee
0–12 dB
0
Soft-knee width for gradual onset
Makeup Gain
-6–12 dB
0
Output gain after compression
Feed-forward peak compressor on the master bus. Use a soft knee (4–8 dB) for transparent limiting; fast attack/slow release for pumping effects.
Transport
Key
Action
A
Start/stop playback
S
Stop all audio
R
Render full audio (offline)
Layers
Key
Action
1–5
Switch to layer 1–5
File Operations
Key
Action
Ctrl+S
Save preset (overwrite)
Ctrl+Z
Undo parameter change
Ctrl+Shift+Z
Redo parameter change
View
Key
Action
Ctrl + / Ctrl -
Zoom in / out
Ctrl 0
Reset zoom to 110%
Tips
Tip
Description
🔒 Lock seed
Click the lock icon next to seed to get reproducible results across previews
📁 File menu
Click "File ▾" at the bottom of the sidebar for Save, Save As, Restore, Export
🎛 Collapse cards
Click any section header (Settings, Binaural, Reverb, Earth&Air) to collapse/expand
📱 Mobile
Use the ☰ hamburger menu on small screens to toggle the sidebar
Community & Feedback
Special thanks to CAT-FOX_ALEX on Freesound for ongoing feedback, suggestions and feature ideas that have shaped Mantice.
Granular Sample Library
All samples are licensed Creative Commons 0 (public domain) from freesound.org.