The attached manual pages describe the Robaux Decision Tree, a trigger/gate routing and clock-divider module. It is not a pitch generator by itself, but it is very useful for building melodic structure when paired with oscillators, quantizers, envelopes, sequencers, shift registers, sample & hold, or precision adders.
At its core, the Decision Tree:
This creates branching rhythmic paths.
It can operate in: - Mono/Mono - Poly/Mono - Poly/Poly - and their Latch versions
It also has: - a random ↔ repeating 16-step pattern control - a new random sequence button - an auto-reset / hidden reset behavior - clock divider modes: - Classic Divider - 2/3/5 Divider - Spread
Melody in Eurorack is often built from several layers:
The Decision Tree is strongest in the last three categories: - it decides when notes happen - it creates branching phrase structures - it can generate different melodic lanes - it can create repeating or semi-random phrase skeletons - it can divide one clock into multiple related time layers
So, together with other melodic modules, it becomes a very powerful melodic organizer.
Use one pitch source and let the Decision Tree decide which voice gets played.
Patch: - clock or trigger sequencer → input a - same pitch CV source → several oscillators or voices - outputs b/c/d → different envelope generators or LPGs for 3 different voices
Each incoming note event is sent to a different voice. If each voice has: - a different timbre - different octave - different quantizer scale - different filter setting
then a single pitch line becomes a distributed melody.
This is especially effective in Mono/Mono mode if you want one clear note event per step.
Instead of only routing triggers to voices, route them to different pitch-generation chains.
Different branches of the tree can produce: - different notes - different octave ranges - different scales - different note lengths
This turns the module into a melodic decision engine.
The manual says knob n smoothly adjusts between: - completely random - and a repeating 16-step pattern
This is excellent for melody because it gives you a continuum between: - generative unpredictability - stable motif repetition
Patch a regular clock into a, and let the outputs trigger: - a quantized random CV source - a step sequencer reset/advance - envelope triggers for note articulation
Then use knob n to move between: - evolving melodic fragments - recognizable looping phrases
A lot of modular melodies fail by being either: - too random to remember - too repetitive to stay interesting
The Decision Tree is designed to sit between those extremes.
Because there are 12 outputs arranged in a hierarchy, you can assign different outputs to different melodic functions.
You can create melodies where: - some events trigger the main pitch - some add ornament notes - some trigger higher-octave doubles - some trigger secondary voices
This is a good way to create musically legible complexity instead of pure randomness.
The latch modes hold the signal until a new input appears. The manual specifically notes this makes the module good as a random switch.
For melody, this is extremely useful.
Use the module to hold one selected route active, so one melodic branch stays selected until the next trigger changes it.
Patch examples: - send one CV source into several switched destinations - hold one drone note active - keep one transposition lane enabled - select one of several sequencers for a phrase
In Latch Mono/Mono or Latch Poly/Mono: - a route stays active longer - phrases feel more intentional - melodic fragments persist before changing
This can create: - sustained pedal tones - held harmonic states - phrase-based melodic switching
In: - Poly/Mono, one or more main outputs can fire, but each selected branch still picks one suboutput - Poly/Poly, multiple outputs can fire across the whole tree
These modes are ideal when your system has: - multiple oscillators - a chord voice - several quantized pitch sources - multiple envelope/VCA paths
The tree produces: - dyads - triad fragments - staggered chord tones - clustered melodic bursts
In latch versions, these can become sustained harmonic textures.
The Decision Tree excels in generative patches where one timing source branches into many related events.
A single incoming pulse creates: - main melody - variation - harmony - resets and phrase structure
This is one of the strongest uses of the module in melodic music.
The divider section is very valuable for melody because it creates multiple related trigger streams at different rates.
According to the manual: - b = /2 - e = /4 - f = /8 - g = /16
And the lower outputs: - c h i j k l m d act like an 8-step clock sequencer, cycling one after the other.
The manual specifies: - b e f g = /2 /4 /8 /16 - c h i j = /3 /6 /12 /24 - d k l m = /5 /10 /20 /40
This is excellent for creating multi-timescale melodies.
Even with simple pitch material, the differing divisions create: - phase relationships - evolving counterpoint - non-repeating melodic interplay
Pitch content can remain simple, while timing complexity creates the sense of intelligent melodic motion.
This mode is especially good with: - quantized random voltages - shift-register melodies - fixed 8-step sequencers - transposition inputs
The manual says Spread mode creates successive divisions: - 1/2, 1/3, 1/4, 1/5, 1/6, 1/7, etc.
It also explicitly suggests using it at audio rate for polyphonic chords using one square VCO.
Feed an audio-rate square wave into the input and use the outputs as subharmonic tones.
This can create: - chordal intervals - pseudo-organ harmonies - subharmonic melodic structures
If you switch or sequence these outputs through VCAs, you can build melodies from one oscillator’s divided frequencies.
At clock rate, use the spread outputs to trigger different voices at increasingly slower rates: - frequent notes in one voice - slower harmony notes in another - very slow root movement in a third
This gives a natural melodic hierarchy: - ornament - phrase - harmonic anchor
The hidden reset input uses jack m in a special mode, and is useful for returning to the first step in loop mode.
This matters for melody because reset is how you create: - downbeats - phrase boundaries - recurring motifs - synchronized loops with other sequencers
When activated, m is no longer available as an output, so it’s a tradeoff: - fewer melodic branches - more phrase control
Usually worth it if your patch needs stronger song structure.
Goal: one pitch source, three melodic personalities
Use Mono/Mono for clean hocketing.
Use Poly/Poly for chord fragments.
Goal: random notes that become recurring motifs
This gives a melody that sounds composed rather than chaotic.
Goal: main notes plus decorative notes
Now the tree creates melodic detail around the core line.
Goal: three related melodies with different pulse rates
In 2/3/5 Divider mode: - /2 outputs → bassline sequencer advance - /3 outputs → lead sequencer advance - /5 outputs → transposition or accent logic
This produces slowly evolving melodic counterpoint.
Goal: melody/harmony from one oscillator
In Spread mode: - square VCO → input - outputs → separate VCAs/mixer/filter paths - use external gates to open selected divided outputs rhythmically
This creates dark, organ-like melodic harmonies from one source.
The Robaux Decision Tree is especially strong for:
By itself, it does not generate pitch CV.
To make it fully melodic, pair it with:
- quantizers
- random CV sources
- sequencers
- sample & hold
- oscillators
- switch modules
- precision adders
- envelope/VCA voice chains
Think of it as a melody traffic controller, not the melody source itself.
It pairs especially well with: - quantizers - Turing Machine / random CV modules - sequencers with reset/advance inputs - sample & hold - shift registers - envelope generators - logic modules - VCAs/LPGs - precision adders - multi-oscillator voice setups
The Decision Tree can be a powerful melodic module when used as a rhythmic branching brain. It shines when you want melodies that feel: - alive - structured - varied - pseudo-composed
Its strongest musical role is to distribute timing and phrase decisions across several pitch paths, turning ordinary melodic sources into evolving, layered musical systems.