Amps, pedals and cabinets

OpenRig has five built-in blocks that play a file: a Neural Amp Modeler capture or an impulse response. Each one only offers the files that make sense in it.

Block	Plays	Typical use
Amp	NAM captures of an amp head	Followed by a Cabinet
Amp + Cab	NAM captures of a combo, or an amp with its cab and mic	Alone, no Cabinet after it
Pedal	NAM captures of pedals, preamps and studio gear	Before the amp
Cabinet	Impulse responses of a speaker cabinet	After an Amp
Space	Impulse responses of a room or a reverb	At the end of the chain

The neural blocks run NeuralAudio, the engine of the NAM LV2 plugin, natively: the model is part of the chain and saved in presets.

Marshall JCM800 2203x by @pamukparty on TONE3000, in an Amp block.

Amp and Pedal controls

Control	Range	What it does
Input	−24 to +24 dB	Drive into the model
Output	−24 to +24 dB	Level after the model
Normalize	on / off	Use the levels recommended by the capture, so captures play at similar loudness
Gate	on / off (off)	Noise gate
Threshold	−100 to 0 dB (−80)	Level under which the gate closes
Bass, Middle, Treble	0 to 10 (5 is flat)	Tone stack, on Amp and Amp + Cab only
Quality	Full / Light	Size of the model, only for captures that hold two sizes

The gate and the tone stack work like those of the NAM plugin, so the same settings sound the same:

• The gate listens to the clean input of the block and turns down the model’s output: it follows your playing, not the hiss the capture adds. One gate on the first high-gain block is usually enough.
• The tone stack is three filters after the model: low shelf at 150 Hz, peak at 425 Hz and high shelf at 1.8 kHz. Centred, it costs nothing. Pedals have none: their capture already includes their tone knob.

Quality

TONE3000 serves recent (A2) captures as one file holding two sizes. Light costs about a quarter of Full, for a slightly less accurate tone: a good choice for a booster before the amp, or to fit more blocks. It switches while you play, without a dropout. A full A2 capture uses about 12 % of one Raspberry Pi 5 core at 128 frames.

Cabinet and Space controls

Control	Range	What it does
Level	−24 to +24 dB	Level of the convolved signal
Mix	0 to 100 %	Dry / wet balance: 100 % on a Cabinet, 35 % by default on a Space
Normalize	on / off	Scale the response to unity gain over 80 Hz – 5 kHz
Low cut	20 to 500 Hz	High-pass on the wet signal (boomy cab, muddy reverb)
High cut	1 to 20 kHz	Low-pass on the wet signal (fizz, bright rooms)

Impulse responses are .wav files at any sample rate, up to 10 seconds. They play with no latency: long room responses are convolved partly on a worker thread, so the cost of the block stays flat.

Your own files

Captures and impulse responses are read from the models folder of the data partition, subfolders included. A tone is a folder of captures of the same gear; a file on its own is a tone with a single capture.

models/
├── Plexi Crunch.nam
├── My Rig/
│   ├── clean.nam
│   └── crunch.nam
└── Tone3000/          ← downloads from the TONE3000 library

Supported files: .nam (WaveNet, LSTM, A1 and A2), RTNeural .json, AIDA-X .aidax, and .wav impulse responses. To copy files to the pedalboard, see SSH and logs.

Presets store the path of the model relative to models, so presets and models can move to another pedalboard together.

Note

Captures are trained at 48 kHz: keep the pedalboard at 48 kHz (the default). At 44.1 kHz the tone is off.