Community Detection Graph Convolutional Network for Overlap-Aware Speaker Diarization

TL;DR

This paper introduces CDGCN, a graph-based clustering method that enhances speaker diarization by capturing local and global speaker relationships, especially for overlapping speech segments.

Contribution

The paper presents a novel graph convolutional network approach for overlap-aware speaker diarization, integrating community detection and overlapped speech detection.

Findings

Outperforms traditional clustering methods on DIHARD III corpus

Effectively detects overlapping speech segments

Improves overall diarization accuracy

Abstract

The clustering algorithm plays a crucial role in speaker diarization systems. However, traditional clustering algorithms suffer from the complex distribution of speaker embeddings and lack of digging potential relationships between speakers in a session. We propose a novel graph-based clustering approach called Community Detection Graph Convolutional Network (CDGCN) to improve the performance of the speaker diarization system. The CDGCN-based clustering method consists of graph generation, sub-graph detection, and Graph-based Overlapped Speech Detection (Graph-OSD). Firstly, the graph generation refines the local linkages among speech segments. Secondly the sub-graph detection finds the optimal global partition of the speaker graph. Finally, we view speaker clustering for overlap-aware speaker diarization as an overlapped community detection task and design a Graph-OSD component to output overlap-aware labels. By capturing local and global information, the speaker diarization system with CDGCN clustering outperforms the traditional Clustering-based Speaker Diarization (CSD) systems on the DIHARD III corpus.