Make something
you can feel.

• Mouth_Wah_Mini_Builder.py
• Project setup guide and parameter documentation

Requires: TouchDesigner, a webcam, audio interface, NumPy and the separately installed MediaPipe TouchDesigner plugin. OpenCV is optional for the overlay.

Source/prototype edition. No native VST3 or DAW plugin is included. Simulated control/DSP checks are documented; live routing and tracking depend on your setup.

• TouchDesigner .toe project
• WAV sample, MP4 demo and illustrated quick-start guide

Requires: TouchDesigner and an audio interface for live guitar.

Existing original release. See the Patreon listing for the included files and setup.

• Complete fft_wah_lab Python builders and callbacks
• Note/chord analysis, color map, and synthetic preview
• Numerical tests and detailed setup/checking guide

Requires: TouchDesigner with NumPy; OpenCV for the rendered visualizer. Audio input/output devices for live use.

Experimental source kit. Native TouchDesigner and Windows audio/device execution has not been verified here. No standalone plugin or finished .tox is supplied.

• Python microphone/Vosk transcription helper
• TouchDesigner caption builder
• Windows startup and connection-test commands
• Setup and troubleshooting guide

Requires: Windows x64, Python 3.11 recommended, TouchDesigner, microphone, and first-run internet access to install Vosk/sounddevice and download a model.

Local prototype. Windows microphone and native TouchDesigner checks remain setup-dependent. English is the default; captions are not guaranteed accurate. No radio recordings, speech model, or third-party packages are bundled.

• ChordRadiance.tox v1.0.1
• Editable engine.py
• Exact keyboard/color mappings
• Preview image and verification notes

Requires: TouchDesigner. No extra packages or assets required.

Checked in TouchDesigner 2023.12370 on macOS for mapping, finite audio data, rendering, and reload. Physical speaker playback was not listened to. This is a keyboard instrument, not a guitar chord-recognition plugin.

• water.gif — original 600 × 240 animation
• Local browser preview and usage notes

Requires: A browser or creative tool that supports animated GIFs. TouchDesigner is optional.

Rendered GIF edition at the stated native resolution. No editable .toe project, audio, HD/4K export, or custom adaptation is included. The animation is also visible as a public portfolio preview.

• film-grain.gif — original 600 × 312 animation
• Local browser preview and usage notes

Requires: A browser or creative tool that supports animated GIFs. TouchDesigner is optional.

Rendered GIF edition at the stated native resolution. No editable .toe project, audio, HD/4K export, or custom adaptation is included. The animation is also visible as a public portfolio preview.

• calibration.py with finite-value checks, clamping, smoothing, and tracking-loss behavior
• Calibration worksheet and sample data
• TouchDesigner integration guide

Requires: Python 3 for the math helper. For live tracking: your own webcam, TouchDesigner and the separately installed upstream MediaPipe plugin.

This is an original mapping helper and guide, not the MediaPipe plugin, a tracker model, or a prebuilt .tox. No OBS/phone-camera setup service is included.

• gesture_color.py — original calibrated gesture-to-RGB functions
• palette.json — twelve named RGB anchors
• TouchDesigner connection recipe

Requires: Python 3 for the mapper. TouchDesigner and your separately installed tracker for live gesture input.

Source-only starter edition. No camera tracker, historical .toe, full feedback composition, or prebuilt .tox is included. The mapping math is checked; live TouchDesigner integration is not verified.

• Editable network-plan.csv and network-template.json
• Python generator for a local HTML node-layout map
• README and handoff checklist templates

Requires: Python 3 to regenerate the diagram, and a browser to view it. No TouchDesigner installation required to use the planning templates.

Planning/template edition. It does not automatically reorganize or modify an existing .toe file. The original project screenshot is a reference, not an editable network.

• Standalone world.html with WebGL and canvas fallback
• Local launch instructions and editable source

Requires: A modern browser. WebGL for the primary renderer; the included canvas fallback is more limited.

Existing source/demo edition. RTX/OpenUSD assets, NVIDIA certification, custom scenes, website integration, and hosting are not included.

• Browser demo with 185 authored notes
• Training/validation data, source catalogs and Colab notebook
• Training, evaluation and API scripts
• CPU/browser tests and model/dataset documentation

Requires: Browser and Python for demo use. Compatible NVIDIA/CUDA/PyTorch environment and separately acquired compute/model dependencies for training.

No trained model or adapter is included. The browser retrieves starter notes; it does not generate answers with a fine-tuned model. GPU execution and quality improvement are unverified.