Characterizing Acoustic Energy Distribution in English: A Quantitative Analysis of Intensity Dynamics in Connected Speech

This research provides a quantitative characterization of acoustic energy distribution within English speech signals. By using a curated subset of the LibriSpeech corpus (dev-clean), the study analyzes 100 high-fidelity audio samples from 10 distinct speakers to establish a normative profile of vocal intensity. In the context of this research, the analysis focuses on the relationship between global acoustic stability and individual dynamic variance. As we can see, results indicate a highly stable global mean intensity (≈ 69.7 dB), while revealing significant inter-speaker heterogeneity in dynamic range and prosodic modulation. Moreover, this study provides essential baseline data for researchers in acoustic phonetics, speech signal processing, and bio-acoustic modeling.