Aftertone & Signal
Back

LEARN LATTICE

Choose how you want to begin. There's no wrong door.

"I JUST WANT SOUNDS"

No theory. No diagrams. Just sound. You'll understand more than you think by the end.

01
Load "Singing Bowl"

Open the preset browser. Find "Play" and select "Singing Bowl".

Try this:
Play a note and hold it. The tone changes over the next few seconds as the network settles into its resonant frequencies.
02
Find the moment of emergence

Look at the big Feedback knob. It's set high, around 88%.

Try this:
Turn Feedback down to 40% and play a note. Then bring it back up past 70%. There's a threshold where the sound stops decaying and starts sustaining. That's the network feeding itself.
03
Break the loop

See the node tabs at the bottom? Each one is a point in the network.

Try this:
While a sound is sustaining, double-click Node 3 to mute it. Some frequencies drop out, others shift. Double-click again to unmute and they come back.
04
Freeze time

Find the Freeze control. It has three modes: Full, With Input, and Partial.

Try this:
Build up a sustained sound, then switch Freeze to "Full". The sound holds indefinitely. Try "With Input" and play new notes on top of the frozen sound.
05
That's the core of it

Feedback controls sustain. Nodes shape the sound. Muting restructures the network. Freezing holds state. Everything else builds on these.

"I WANT TO UNDERSTAND"

Lattice isn't a delay. It's a network. That distinction changes everything.

What is a feedback network?

A traditional delay has one path: signal goes in, gets delayed, comes out. Feedback sends some output back to the input, but it's still one loop.

A feedback network has multiple points (nodes) that all feed into each other. Sound doesn't travel in a circle. It travels everywhere simultaneously.


    Traditional delay:          Feedback network:

    ┌─────────┐                 ┌─────────┐
    │  INPUT  │                 │ Node 1  │◄───────┐
    └────┬────┘                 └────┬────┘        │
         │                           │             │
         ▼                      ┌────┴────┐        │
    ┌─────────┐                 ▼         ▼        │
    │  DELAY  │◄──┐        ┌─────────┐ ┌─────────┐ │
    └────┬────┘   │        │ Node 2  │ │ Node 3  │ │
         │        │        └────┬────┘ └────┬────┘ │
         ▼        │             │           │      │
    ┌─────────┐   │             └─────┬─────┘      │
    │ OUTPUT  │───┘                   │            │
    └─────────┘                       └────────────┘
              

When sound enters a network, it takes every possible path. Different paths have different delay times, filters, and drive settings. The outputs interfere with each other, sometimes reinforcing, sometimes canceling. This interference creates timbres that no single delay can produce.

What happens inside each node?

Each node is a small signal processor: delay, then filter, then drive. The delay sets the node's contribution to the network's timing and pitch. The filter shapes which frequencies pass through. The drive adds harmonics and saturation.

Because every node feeds every other node, these small differences compound. One node with a bandpass filter emphasizes a frequency range. Two nodes with different bandpass settings create complex resonances. The interactions scale with the number of nodes.

Why does feedback past 70% sound different?

Below ~70% feedback, energy dissipates faster than it accumulates. The sound decays.

Above ~70%, the network reaches equilibrium. Energy circulates indefinitely. High-feedback networks are selective: frequencies that align with the delay times get reinforced, others get canceled.

The network decides which harmonics survive based on its topology, not your input.

What does muting a node actually do?

Muting a node removes it from the network. But because every node feeds every other node, removing one changes all the paths that passed through it.


    Before muting Node 2:       After muting Node 2:

    ┌─────────┐                 ┌─────────┐
    │ Node 1  │◄───────┐        │ Node 1  │◄───────┐
    └────┬────┘        │        └────┬────┘        │
         │             │             │             │
    ┌────┴────┐        │             │             │
    ▼         ▼        │             │             │
┌─────────┐ ┌─────────┐│        ┌ ─ ─ ─ ─ ┐ ┌─────────┐
│ Node 2  │ │ Node 3  ││          Node 2    │ Node 3  ││
└────┬────┘ └────┬────┘│        └ ─ ─ ─ ─ ┘ └────┬────┘│
     │           │     │             ╳           │     │
     └─────┬─────┘     │             ╳           │     │
           │           │             ╳───────────┘     │
           └───────────┘                               │
                                     └─────────────────┘
              

Some frequencies lose their reinforcement path and fade. Others find new paths and become more prominent. Muting is not subtraction — it's restructuring.

What makes freeze different in a network?

In a traditional delay, freeze captures a buffer: a recording of what just happened.

In a network, freeze captures state: the current energy distribution across all nodes. The sound continues to circulate and interfere with its input disconnected.

"With Input" mode keeps the input connected, so new sounds blend into the frozen network. "Partial" mode bleeds energy slowly, letting the frozen state evolve.

Where does the sound come from?

A feedback network is passive. It shapes and sustains sound, but it needs energy from somewhere. That's the Exciter.

You can feed it external audio, or use the built-in sources: a short impulse, a burst of noise, or a simple oscillator tone. The choice affects the initial character, but the network transforms it.

Why one LFO with multiple targets?

Lattice has a single LFO, but it can modulate different things: delay times, filter cutoffs, feedback amount, or node gains.

This keeps the interface simple while still allowing movement. Modulating delay times creates pitch drift and chorus effects. Modulating filters creates timbral sweeps. Modulating feedback creates swells and pulses.

One source of motion, applied where you need it.

Why link nodes together?

With up to 8 nodes, adjusting each one individually can be tedious. Node linking lets you group nodes so they move together.

Common uses: keeping stereo pairs matched, adjusting multiple delay times proportionally, or creating macro controls that affect several nodes at once.

Going deeper

For detailed parameter ranges, filter techniques, and drive strategies, see the Sound Design guide. It covers the technical details that turn understanding into craft.

"I WANT TO BUILD"

Start with a goal. Work backward. Each recipe teaches a principle.

Shimmering pad that evolves forever
Ethereal, slowly-changing texture that never quite repeats
  1. Start with 4 nodes, all connected (full mesh)
  2. Set delay times to prime-number relationships: 47ms, 71ms, 103ms, 149ms
  3. Enable LP filters on all nodes, cutoff around 2-4kHz, resonance 20-40%
  4. Feedback to 85%, Damping to 30%
  5. LFO targeting filter cutoff, slow rate (0.1Hz), depth 40%
  6. Exciter: Oscillator, Sine wave, long decay (2-3 seconds)
Why it works: Prime-number delays never align, so the pattern never repeats. The slow LFO sweep creates movement without disruption. High feedback sustains; damping prevents harshness.
Metallic, bell-like tones
Pitched, resonant, with inharmonic overtones like a struck metal object
  1. Use 3 nodes with very short delay times: 2-10ms range
  2. Set delays to non-integer ratios: 2.3ms, 4.7ms, 7.1ms
  3. All filters off, or HP at low cutoff to remove mud
  4. Add subtle drive (Soft, 20-30%) on one node for harmonic richness
  5. Feedback at 75-85%, Spread at 50%
  6. Exciter: Impulse with short decay, or Noise burst (1-5ms)
Why it works: Very short delays create pitched resonance (like Karplus-Strong synthesis). Non-integer ratios produce inharmonic partials, the signature of struck metal. The network "rings" at frequencies determined by delay times.
Rhythmic delay with gaps
Tempo-synced echoes with intentional holes in the pattern
  1. Use 4 nodes with tempo-synced delays: 1/8, 1/8T, 1/4, 1/4.
  2. Start with all nodes connected
  3. Mute Node 2 (the 1/8T) to create an asymmetric gap
  4. Feedback at 60-70% (enough for repeats, not infinite)
  5. Optional: BP filter on one node to emphasize mid frequencies
  6. Exciter: Off (use external audio input)
Why it works: Muting one node removes that delay time from the pattern while keeping the network structure. The triplet gap (1/8T) creates rhythmic interest. Use node muting as a performance tool. Unmute during a breakdown.
Infinite drone from nothing
Self-sustaining texture that requires no input after initial trigger
  1. Use 6-8 nodes for maximum complexity
  2. Delay times between 100ms and 500ms, varied
  3. Feedback at 95-100%
  4. Add light LP filtering (8-12kHz cutoff) to prevent harshness
  5. Exciter: Noise burst, medium duration (20-50ms)
  6. Trigger once, then switch Freeze to "Full"
Why it works: A single noise burst excites all frequencies. The network selects which survive based on delay time interference. Freezing locks the system at the moment of maximum richness. The drone lives indefinitely, evolving subtly as phases shift.
Lo-fi warped tape effect
Degraded, wobbly, vintage-sounding processing
  1. Use 2-3 nodes for simplicity
  2. Delay times around 80-150ms (classic tape echo range)
  3. Enable Bit drive on all nodes, amount 40-60%
  4. LP filter at 4-6kHz to remove digital harshness
  5. LFO targeting delay times, rate 2-4Hz, depth 30-50%
  6. Feedback at 50-65%
Why it works: Bit crushing adds digital artifacts that feel analog when combined with filtering. LFO on delay time creates the pitch wobble of stretched tape. Moderate feedback gives repeats without infinite sustain.
Reverse-swell pad
Sounds that build up before the note, like reversed reverb
  1. Use 5-6 nodes with long delay times: 500ms to 2 seconds
  2. Stagger delays so energy arrives at different times
  3. Feedback at 70-80%
  4. Exciter: Oscillator with slow attack (200-400ms)
  5. Set Mix to 100% wet (no dry signal)
  6. Play notes slightly before the beat
Why it works: Long delays with staggered times create a diffuse buildup. The slow attack exciter feeds energy gradually. Playing ahead of the beat means the swell peaks on the beat. 100% wet hides the original attack.