Enhancement of throat microphone speech using Empirical Mode Decomposition (EMD)

This paper presents a novel approach for enhancing the quality of throat microphone (TM) speech using Empirical Mode Decomposition (EMD). TM speech is known for its robustness in noisy environments but often suffers from poor intelligibility and unnatural sound due to the absence of high-frequency components. To address this issue, we propose using EMD to decompose the TM speech signal into intrinsic mode functions (IMFs) and selectively enhance components that contribute to improved speech clarity. The performance of the proposed method is evaluated using Perceptual Evaluation of Speech Quality (PESQ) scores, Signal-to-Noise Ratio (SNR), comparison of Linear Predictive Coding (LPC) spectra and Spectrogram analysis. Results demonstrate significant improvements in speech quality, making the approach a promising solution for applications requiring reliable communication in adverse conditions.