Microsoft Speaker Diarization System for the VoxCeleb Speaker Recognition Challenge 2020

TL;DR

This paper presents Microsoft's advanced speaker diarization system for multi-talker recordings, achieving top performance in the VoxCeleb Speaker Recognition Challenge 2020 by integrating novel components like Res2Net embeddings and conformer-based separation.

Contribution

The paper introduces a novel diarization system combining Res2Net embeddings, conformer-based separation, and a modified DOVER method, tailored for real-world multi-talker audio in challenging conditions.

Findings

Achieved 3.71% DER on development set

Achieved 6.23% DER on evaluation set

Ranked 1st in the challenge

Abstract

This paper describes the Microsoft speaker diarization system for monaural multi-talker recordings in the wild, evaluated at the diarization track of the VoxCeleb Speaker Recognition Challenge(VoxSRC) 2020. We will first explain our system design to address issues in handling real multi-talker recordings. We then present the details of the components, which include Res2Net-based speaker embedding extractor, conformer-based continuous speech separation with leakage filtering, and a modified DOVER (short for Diarization Output Voting Error Reduction) method for system fusion. We evaluate the systems with the data set provided by VoxSRCchallenge 2020, which contains real-life multi-talker audio collected from YouTube. Our best system achieves 3.71% and 6.23% of the diarization error rate (DER) on development set and evaluation set, respectively, being ranked the 1st at the diarization track of the challenge.