Window Size Versus Accuracy Experiments in Voice Activity Detectors

2026-01-24

AI Summary

This study systematically investigates the joint impact of window size and hysteresis mechanisms on the accuracy of voice activity detection (VAD). Focusing on three widely used VAD algorithms—Silero, WebRTC, and RMS—the authors evaluate their performance across diverse real-world audio streams under varying window configurations and with or without hysteresis. The experiments provide the first multi-algorithm, real-scenario quantification of how these two critical parameters influence VAD effectiveness. Results demonstrate that Silero consistently outperforms the other methods by a significant margin, while the incorporation of hysteresis markedly enhances the stability of WebRTC. These findings offer practical, empirically grounded guidelines for parameter tuning in real-world VAD system deployment.

Technology Category

• Machine Learning: Active Learning
• Search and Optimization: Algorithm Configuration
• Computer Vision: Video Understanding & Activity Analysis

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

Voice activity detection (VAD) plays a vital role in enabling applications such as speech recognition. We analyze the impact of window size on the accuracy of three VAD algorithms: Silero, WebRTC, and Root Mean Square (RMS) across a set of diverse real-world digital audio streams. We additionally explore the use of hysteresis on top of each VAD output. Our results offer practical references for optimizing VAD systems. Silero significantly outperforms WebRTC and RMS, and hysteresis provides a benefit for WebRTC.

Problem

• Voice Activity Detection
• Window Size
• Accuracy
• Hysteresis
• Speech Recognition

Innovation

• Voice Activity Detection
• window size
• hysteresis
• Silero
• accuracy optimization