GIST-AiTeR System for the Diarization Task of the 2022 VoxCeleb Speaker Recognition Challenge

TL;DR

The GIST-AiTeR system for the 2022 VoxCeleb challenge combines multiple speech processing models into an ensemble, achieving a low diarization error rate and ranking third.

Contribution

This paper introduces an ensemble diarization system integrating speech enhancement, VAD, multi-scaled embeddings, and overlapped speech detection for speaker diarization.

Findings

Achieved a diarization error rate of 5.12% on the challenge dataset.

Constructed four different diarization systems with various model combinations.

Ranked third in the VoxCeleb Speaker Recognition Challenge.

Abstract

This report describes the submission system of the GIST-AiTeR team at the 2022 VoxCeleb Speaker Recognition Challenge (VoxSRC) Track 4. Our system mainly includes speech enhancement, voice activity detection , multi-scaled speaker embedding, probabilistic linear discriminant analysis-based speaker clustering, and overlapped speech detection models. We first construct four different diarization systems according to different model combinations with the best experimental efforts. Our final submission is an ensemble system of all the four systems and achieves a diarization error rate of 5.12% on the challenge evaluation set, ranked third at the diarization track of the challenge.