Stereo output & wet/dry parallel
How the master Stereo toggle reshapes the signal path, what the Dry knob is for on high-gain rigs, and the Auto-Pan pedal as a quick way to hear the difference.
Axion's signal path is mono-by-default. Two master-section controls let you change that: the STEREO toggle and the Dry knob, both sitting in the Cab panel header / footer. They're independent — flip one without touching the other.
The STEREO toggle
The small STEREO pill in the top-right of the Cab panel switches every stereo-capable node in the rig between mono (1-channel, explicit) and stereo (2-channel, explicit). Off by default to preserve the mobile CPU budget.
What flips when you turn it on:
- Noise gate — envelope-per-channel tracking so a stereo signal doesn't get a duck on one side when only the other triggered.
- Chorus, delay, EQ — internal nodes upgrade to two channels, preserving stereo content that arrives at their input.
- Reverb, cab — the convolvers do pseudo-stereo from mono: the input is up-mixed to L=R, then each channel convolves independently. With the bundled mono IRs the output is still L=R, but downstream wide effects now have two channels to work with. A future stereo IR upgrade will unlock true spatial separation.
What doesn't change:
- Amp + NAM stay mono internally. They're the highest CPU draw in the rig, and stereo NAM would double the kernel cost per sample. This is consistent with how every commercial NAM-bearing plugin ships — mono kernel, stereo-from-mono in the post. The cab handles spatialising the post-amp signal.
NOTE With STEREO off, the cab is forced to channelCount: 1. So even if you flip an "always stereo" effect like Auto-Pan upstream, its output collapses to mono at the cab. Switching STEREO on opens the door for those spatial effects to be heard.
What it actually costs
Numbers on the Xiaomi 11 Lite 5G NE (our mid-tier mobile baseline) running a full chain at high gain:
| Mode | Sustained CPU |
|---|---|
| STEREO off | baseline (≈60–65% with NAM, ≈40% analytical) |
| STEREO on | +15–25% absolute |
The cost is dominated by the two convolvers (reverb + cab) going from 1-channel to 2-channel convolution. The cheap-fix nodes (chorus / delay / EQ / gate) only add ~1.4–1.6× their mono cost.
If you don't need stereo and you're on a tight CPU budget, leave it off. If you're on desktop or you want spatial depth, leave it on.
Per-preset persistence
The STEREO state is part of the preset. Saving with STEREO on means it loads on the next session at the right state; loading a preset that has it off will switch back automatically. Older presets that predate this field load as mono (the historical default).
The Dry knob
Right next to the Cab's Mix knob, the Dry knob blends a parallel path of the unprocessed DI signal (your raw guitar input, pre-pedalboard) back into the master output. The wet rig still plays at full level; Dry adds the dry signal on top.
The use case is high-gain rigs that lose pick attack and low-end definition because the saturator squashes transients. A 15–25% dry blend recovers the punch without softening the gain itself.
A pattern that holds up for metal-style tones:
- Set your high-gain tone as you normally would. Pick attack feels a bit smudged.
- Open Dry to ~20%.
- The chug definition returns. Low end tightens. The gain saturation hasn't moved.
Default value is 0 — you only hear what you ask for. The blend is a single linear gain on the DI signal; there's no compression, no EQ, no phase alignment to worry about (the DI is in phase with the wet rig at the master sum point).
Like STEREO, the Dry value is per-preset.
The Auto-Pan pedal — a way to hear stereo work
Auto-Pan is the easiest way to confirm the STEREO toggle is doing what it should. It's a normal pedal — add it to the chain like any other.
A sine LFO sweeps a stereo panner between L and R:
- Rate — 0.1 to 6 Hz. Around 2 Hz is the "obvious side-to-side" sweet spot for testing.
- Depth — 0 to 100 %. 100 % is full L↔R; lower values stay narrower.
What you should hear:
- STEREO on — dramatic side-to-side panning. Headphones make this obvious; speakers show it depending on placement.
- STEREO off — the cab sums L+R back to mono, the spatial motion collapses, and what's left is a faint amplitude wobble at the LFO rate (the equal-power pan law dips ~3 dB at extremes). The pan motion you'd expect is gone.
Toggle STEREO live with Auto-Pan running to hear the transition directly.
Practical recipes
A few combinations that tend to land:
| Goal | STEREO | Dry | Effects to lean on |
|---|---|---|---|
| Modern high-gain rhythm | on | 20% | gate + amp + cab; reverb low |
| Spacious clean | on | 0% | chorus + delay + reverb |
| Tight retro chug | on | 25% | gate + amp + cab; dry restores chunk |
| Ambient soundscape | on | 0–10% | delay (long) + reverb (high mix); Auto-Pan slow |
| CPU-tight mobile rig | off | 0% | everything mono — saves convolver cost |
What's not in yet
- Stereo input — for two-channel sources like acoustic-piezo blends or stereo synths into Axion. The audio graph is ready (every stereo-capable node carries the upstream channel count); the missing piece is asking
getUserMediafor two channels and routing them. Deferred until there's user demand. - Stereo cab IRs — bundled IRs are mono today. The convolvers happily accept stereo IRs; a premium pack of true-stereo cabs is a follow-on. Until then, mono IR + STEREO on produces pseudo-stereo (L=R).
- True stereo NAM — defer indefinitely. Doubling the LSTM kernel work per sample for the marginal stereo-spread benefit doesn't make sense.
See Signal flow diagram for where the dry tap sits in the graph and how the looper plugs in alongside.