EML Online Speech Activity Detection for the Fearless Steps Challenge Phase-III

TL;DR

This paper presents an online speech activity detection algorithm designed for noisy NASA Apollo-11 audio data, capable of real-time processing with high accuracy in challenging conditions.

Contribution

The paper introduces a novel online SAD algorithm that can be trained supervised or unsupervised, achieving real-time performance on NASA Apollo-11 data.

Findings

Competitive accuracy on development and evaluation datasets

Real-time processing with a factor of 0.002 on a single CPU

Effective in noisy, degraded audio conditions

Abstract

Speech Activity Detection (SAD), locating speech segments within an audio recording, is a main part of most speech technology applications. Robust SAD is usually more difficult in noisy conditions with varying signal-to-noise ratios (SNR). The Fearless Steps challenge has recently provided such data from the NASA Apollo-11 mission for different speech processing tasks including SAD. Most audio recordings are degraded by different kinds and levels of noise varying within and between channels. This paper describes the EML online algorithm for the most recent phase of this challenge. The proposed algorithm can be trained both in a supervised and unsupervised manner and assigns speech and non-speech labels at runtime approximately every 0.1 sec. The experimental results show a competitive accuracy on both development and evaluation datasets with a real-time factor of about 0.002 using a single CPU machine.