How quinnsmitth/neuropedal works
Neuropedal predicts guitar-pedal drive and tone settings from audio, and also includes training, evaluation, and audio-feature utilities. The primary user workflow is selecting a WAV in the GUI, preprocessing it, running inference, and displaying the prediction. The graph also preserves the documented training and analysis capabilities. Some implementation details are unsampled; edges are limited to relationships shown in excerpts, the SOURCE INDEX, or the README.
Main entry points: infer.py Inference GUI · select_path.py Data-root selection · evaluate_model.py Model evaluation · visualize_results.py Result visualization
Components 15
Prediction
- Inference GUI
src/Classification/infer.py - Audio preprocessing
src/Classification/infer.py - Mel transform
src/Classification/melSpec.py - Pedal regressor
src/Classification/model.py
Model learning
- Training dataset
src/Classification/melDataLoader.py - Training pipeline
src/Classification/melTrain.py - Trained weights
weights/guitar_model_mel_36300_100.pth
Analysis
- Model evaluation
src/Classification/evaluate_model.py - Result visualization
src/Classification/visualize_results.py
Audio utilities
- Data-root selection
src/select_path.py - MFCC extraction
src/extract_features.py - Feature CSV
- MIDI rendering
src/render_to_wav.py
Other components
- Guitarist
- FluidSynth
Connections 18
- Guitarist → Inference GUI: selects WAV
README.md - Inference GUI → Audio preprocessing: preprocesses audio
src/Classification/infer.py - Audio preprocessing → Inference GUI: returns Mel input
src/Classification/infer.py - Inference GUI → Pedal regressor: runs prediction
- Inference GUI → Guitarist: displays settings
README.md - Training pipeline → Training dataset: loads batches
src/Classification/melTrain.py - Training dataset → Mel transform: converts audio
src/Classification/melDataLoader.py - Training pipeline → Trained weights: saves weights
src/Classification/melTrain.py - Model evaluation → Pedal regressor: evaluates model
src/Classification/evaluate_model.py - Model evaluation → Mel transform: preprocesses audio
src/Classification/evaluate_model.py - Model evaluation → Data-root selection: loads data path
src/Classification/evaluate_model.py - Result visualization → Pedal regressor: runs predictions
src/Classification/visualize_results.py - Result visualization → Mel transform: preprocesses audio
src/Classification/visualize_results.py - Result visualization → Data-root selection: loads data path
src/Classification/visualize_results.py - MFCC extraction → Data-root selection: loads data path
src/extract_features.py - MFCC extraction → Feature CSV: writes features
src/extract_features.py - MIDI rendering → Data-root selection: loads data path
src/render_to_wav.py - MIDI rendering → FluidSynth: renders MIDI
src/render_to_wav.py