On a cepstrum-based speech detector robust to white noise

TL;DR

This paper introduces a cepstrum-based speech detection method that remains effective in white noise environments by using a scalar quantity V derived from cepstral coefficients, enabling reliable speech frame classification.

Contribution

The paper proposes a novel scalar measure V from cepstral coefficients that improves speech detection robustness against white noise.

Findings

The distribution of V effectively separates speech and noise frames at SNRs above 5 dB.

The method demonstrates robustness to white noise in speech detection.

Cepstral coefficients' distributions are significantly affected by white noise.

Abstract

We study effects of additive white noise on the cepstral representation of speech signals. Distribution of each individual cepstrum coefficient of speech is shown to depend strongly on noise and to overlap significantly with the cepstrum distribution of noise. Based on these studies, we suggest a scalar quantity, V, equal to the sum of weighted cepstral coefficients, which is able to classify frames containing speech against noise-like frames. The distributions of V for speech and noise frames are reasonably well separated above SNR = 5 dB, demonstrating the feasibility of robust speech detector based on V.