Underlying Data for: Edge-Computed Voice Activity Detection (VAD) for Privacy-Preserving and Energy-Efficient Smart Home Lighting

This dataset contains the empirical measurements from 200 experimental trials evaluating a localized Voice Activity Detection (VAD) system. The study assesses classification accuracy (TP, TN, FP, FN), latency (ms), and energy efficiency across quiet (30–40 dB) and noisy (55–65 dB) environments. Files include the raw CSV data and a README file with variable definitions.