Axon 3 is a drum synthesizer plugin by Audio Damage for macOS, Windows, Linux, and iOS, driven by a sequencer built like a small neural network. Seven neurons each count the trigger pulses they receive, and when a neuron's count passes its threshold it plays its drum voice and sends a pulse to every neuron it is wired to, so a few connections and threshold settings build interlocking rhythms without programming a single step. Trigger likelihood and ratcheting add variation, and pitch quantizing to 45 scales makes melodic patterns easy. Each neuron's voice started as a model of the Audio Damage Boomtschak analog drum module and adds FM, two waveshapers, a resonant filter, and a six-oscillator cymbal source from Tattoo, and the voices can modulate one another over shared FM and AM busses. A full mixer, a stereo delay, a Rough Rider compressor on the main output, MIDI in and out, and four stereo outputs complete it. Available as VST3, AU, AAX, and CLAP for desktop, and AUv3 for iPad. Priced at $89 USD with no DRM, no subscription, and a perpetual license.

Platforms & Downloads

Fully functional demo. Save disabled.

Features

• Neural network sequencer. Seven neurons, each with a threshold from 0 to 16 and a preload count for its starting point. The clock pulses the first neuron on every 1/8, 1/16, or 1/32 note, and each neuron that fires passes a pulse to the neurons it is wired to.
• Trigger likelihood and ratchets. Set how likely each neuron is to play when it fires, how likely it is to ratchet, the ratchet size, and up to eight repeats, plus each neuron's output velocity and velocity randomization.
• Clock and reset. Follow the host transport or run from the internal clock at 20 to 240 BPM, and reset the network on play or every 1, 2, 4, or 8 measures to keep patterns in phrase.
• Seven drum voices. Each neuron plays its own voice: a sine or triangle oscillator from 20 Hz to 1.5 kHz with a pitch envelope and an amplitude envelope, an FM modulator with a ratio from 0.25:1 to 6:1 and its own decay, a click transient, and two waveshapers.
• Cymbal and noise source. A six-oscillator metallic source carried over from Tattoo, blended with white noise and tunable 12 semitones either way, with its own attack, decay, and curve controls.
• Filter. Low-pass, band-pass, or high-pass with resonance, on the noise alone or on the oscillator and noise together, modulated by the noise envelope, the bend envelope, or the oscillator.
• FM and AM busses. Any voice can send to and receive from shared FM and AM busses, so the seven voices can modulate one another and combine into a single complex percussion synthesizer.
• Pitch and scale quantizing. Snap any voice's tuning to notes in the current scale, chosen from 45 with a root note and a transpose of up to 24 semitones either way. MIDI output notes follow the same scale.
• Mixer. Level, pan, mute, solo, and delay send for every voice, with each voice assignable to the main output or one of three stereo aux outputs.
• Stereo delay. Left and right delay times from 1 ms to 2 seconds, each free or tempo-synced, with feedback, color, width, and level controls.
• Rough Rider compressor. On the main output, with sensitivity, ratio, attack from 0.01 to 100 ms, release from 10 ms to 1 second, up to 30 dB of makeup gain, and a bypass switch.
• MIDI in and out. Trigger each neuron or its voice from a MIDI note, send a MIDI note from each neuron to drive other instruments, choose whether a neuron plays its voice, sends MIDI, or both, and assign notes with MIDI learn.
• Randomizer. Randomize everything, the network wiring, all seven thresholds, all seven voices, the pitches, or just the current voice, and copy a voice to paste it into another or into all of them.
• Preset browser. Folders, copy and paste, and import and export of single presets or whole folders as ZIP files, with 21 factory presets and 67 designer presets in three artist banks.
• Universal preset system. Cross-platform XML-based presets compatible across all plugin formats and operating systems, including the iPad edition. Save over the Default preset to start every new instance from your own settings.
• Resizable vector UI. HiDPI and Retina aware, scales cleanly from a laptop panel to a large display, and new instances open at the size you last used.

Sound Design Ideas

Start from the Initialize preset, wire the first neuron to two others with thresholds of 2 and 3, and listen to the pattern shift as their counts drift apart. Turn on pitch quantize for a few voices, pick the minor pentatonic scale, and let the network play a bassline against the drums. Lower the trigger likelihood on the hats and add a little ratchet likelihood for rolls that come and go. Send one voice to the FM buss and let another receive it for clangy, bell-like hits, or set the noise section's mix toward its metallic oscillators with a short decay for tight hats. Set a neuron to MIDI only to drive a drum rack in your DAW, put the kick on its own aux output for separate processing, and reset every four measures to keep the pattern in phrase.

Frequently Asked Questions