Diarisation using location tracking with agglomerative clustering

TL;DR

This paper introduces a speaker diarisation method that incorporates speaker location tracking with Kalman filters into agglomerative clustering, improving accuracy in dynamic meeting scenarios.

Contribution

It relaxes the stationarity assumption in spatial location modeling by explicitly tracking speaker movements within an AHC framework.

Findings

Improved diarisation accuracy on Microsoft rich meeting data.

Outperforms methods without location tracking.

Effective in dynamic speaker movement scenarios.

Abstract

Previous works have shown that spatial location information can be complementary to speaker embeddings for a speaker diarisation task. However, the models used often assume that speakers are fairly stationary throughout a meeting. This paper proposes to relax this assumption, by explicitly modelling the movements of speakers within an Agglomerative Hierarchical Clustering (AHC) diarisation framework. Kalman filters, which track the locations of speakers, are used to compute log-likelihood ratios that contribute to the cluster affinity computations for the AHC merging and stopping decisions. Experiments show that the proposed approach is able to yield improvements on a Microsoft rich meeting transcription task, compared to methods that do not use location information or that make stationarity assumptions.