biosynth-emg 0.1.0

Physics-informed synthetic EMG signal generator for prosthetic control

BioSynth-EMG: Physics-Informed Generative Framework for Synthetic Myoelectric Signals

A physics-based synthetic EMG signal generator for training low-latency neural networks in prosthetic control systems.

Features

• Biomechanical Layer: Motor unit firing rate simulation using Poisson distribution
• Electrical Propagation Layer: MUAP generation with Hermite functions
• Realistic Noise: Gaussian noise, power line interference, and movement artifacts
• High Performance: Optimized for GTX 1650 (<1ms generation for 1s signal)
• Dataset Generation: 8-channel EMG with gesture labels and force values

Installation

pip install -r requirements.txt

Usage

from biosynth_emg import BioSynthGenerator

generator = BioSynthGenerator()
data = generator.generate_dataset(num_samples=1000, duration=1.0)

Paper

Based on "BioSynth-EMG: A Physics-Informed Generative Framework for Synthetic Myoelectric Signal Synthesis and Prothetic Control Benchmarking"