Chameleon
Key Features
- Three neural network amp voicings — Red (high gain), Gold (crunchy overdrive), and Green (crisp clean) — modeled from a vintage tube amp head
- RTNeural inference engine delivers realistic tube amp dynamics with low CPU usage comparable to commercial amp simulators
- Built-in EQ and gain controls for shaping each channel beyond the default neural network capture
- Stereo processing support for wider amp tones in a mix context
- Open-source architecture allows loading custom user-trained amp and pedal models via JSON files
- Cross-platform availability with native support on Windows, macOS, and Linux
Description
Chameleon by GuitarML is a guitar amp head simulator that uses neural network inference to recreate the dynamic response of a real vintage tube amplifier. Built on the RTNeural engine, it delivers convincing tube amp tones with CPU usage low enough to compete with commercial plugins.
The plugin offers three distinct voicings modeled from a vintage-style amp head. Red delivers high-gain saturation for heavier styles, Gold produces a warm crunchy overdrive, and Green provides crisp clean tones.
Each channel responds dynamically to your playing, preserving the touch sensitivity and harmonic complexity of a real tube circuit. The built-in EQ and gain controls let you shape each voicing beyond its default character, giving you a practical range of tones from a single plugin.
Version 1.2 introduced stereo processing, improved amp and EQ sound, and refined state saving within your DAW. It also removed the standalone app and 32-bit Windows builds in favor of a streamlined 64-bit plugin-only release.
The three tone models are stored as JSON files in your Documents folder and can be manually swapped out for user-trained models. This open architecture, combined with GuitarML's open-source codebase, means experienced users can train and load their own neural network captures of any amp or pedal using the Automated-GuitarAmpModelling framework.
Chameleon is designed to sit in your signal chain the same way a real amp head would. Pair it with an impulse response loader for cabinet simulation, add reverb and delay, and you have a complete guitar rig inside your DAW.
Video Preview
Frequently Asked Questions
Can I load my own amp models into Chameleon?
Yes. The three tone models are stored as JSON files in your Documents folder and can be replaced with custom models. You can train your own captures using the Automated-GuitarAmpModelling project and a Colab training script maintained by GuitarML.
Do I need a cabinet simulator with Chameleon?
Yes. Chameleon models an amp head only, not a speaker cabinet. For the most realistic results, place an impulse response loader after Chameleon in your signal chain to simulate a guitar cabinet and microphone setup.
How does Chameleon compare to commercial neural amp simulators?
Chameleon uses the same RTNeural inference engine found in several respected audio applications. While commercial alternatives like TONEX offer larger model libraries and additional effects, Chameleon provides convincing tube amp dynamics and touch sensitivity at no cost, with CPU usage competitive with paid options.
What changed in version 1.2?
Version 1.2 added stereo processing, improved the amp and EQ sound quality, and fixed state saving in your DAW. The standalone app and 32-bit Windows builds were removed in this update.
