An End-to-End Approach for Korean Wakeword Systems with Speaker Authentication

TL;DR

This paper presents an end-to-end Korean wakeword detection and voice authentication system that addresses language-specific challenges and privacy concerns, demonstrating promising accuracy and security in experimental results.

Contribution

It introduces a novel Korean wakeword training method combined with voice authentication using an open-source platform, enhancing privacy and language adaptability.

Findings

Achieved 16.79% EER in wakeword detection

Achieved 6.6% EER in voice authentication

Demonstrated effectiveness for Korean language applications

Abstract

Wakeword detection plays a critical role in enabling AI assistants to listen to user voices and interact effectively. However, for languages other than English, there is a significant lack of pre-trained wakeword models. Additionally, systems that merely determine the presence of a wakeword can pose serious privacy concerns. In this paper, we propose an end-to-end approach that trains wakewords for Non-English languages, particulary Korean, and uses this to develop a Voice Authentication model to protect user privacy. Our implementation employs an open-source platform OpenWakeWord, which performs wakeword detection using an FCN (Fully-Connected Network) architecture. Once a wakeword is detected, our custom-developed code calculates cosine similarity for robust user authentication. Experimental results demonstrate the effectiveness of our approach, achieving a 16.79% and a 6.6% Equal Error Rate (EER) each in the Wakeword Detection and the Voice Authentication. These findings highlight the model's potential in providing secure and accurate wakeword detection and authentication for Korean users.