Elior Hadad

Title: Elior Hadad and Speech Features-Based Voice Activity Detection

Introduction

Elior Hadad is an inventor associated with Ness Ziona, Israel. His patent work focuses on improving voice activity detection and noise reduction in audio systems. The supplied information lists 1 patent and identifies his company as GM Global Technology Operations, Inc.

Latest Patents

Hadad’s listed patent is “Speech Features-Based Single Channel Voice Activity Detection Method and System for Reducing Noise from an Audio Signal.” It describes a single-channel Speech Features-Based Voice Activity Detection (SFVAD) system that produces frame-by-frame speech and noise indications and calculates speech and noise time-frequency masks.

The system extracts speech-like patterns from an input audio signal while remaining invariant to the signal’s power level. It uses voiced and unvoiced speech-feature detectors to guide noise estimation, with a cepstral-based pitch detector and a centrum calculation method intended to prevent speech content from contaminating the noise estimate. The design also aims to respond robustly to sudden changes in background noise and reduce false detections.

The patent describes applications in automatic speech recognition, virtual assistants, and hands-free calls. It also explains how the SFVAD system can control adaptation in a beam-forming system, supporting improved speech quality and noise reduction, including when transient noises occur.

Career Highlights

The provided information associates Hadad with GM Global Technology Operations, Inc. His listed invention addresses practical challenges in audio processing, including reliable speech detection and noise handling.

Collaborations

The supplied data identifies Moshe Tzur as Hadad’s coworker. It does not specify their roles or describe a particular joint project.

Conclusion

Elior Hadad’s listed patent centers on a speech-feature-based approach to single-channel voice activity detection, with techniques designed to improve noise estimation and support clearer audio in speech-recognition and communication applications.