From Modular to End-to-End Speaker Diarization

TL;DR

This paper reviews the evolution from modular to end-to-end speaker diarization, introducing new models and data generation techniques that improve handling overlapped speech and multiple speakers.

Contribution

It presents a new EEND-based model called DiaPer, compares it with VBx, and introduces a synthetic data generation method for training neural diarization models.

Findings

DiaPer outperforms EEND-EDA with many speakers and overlaps.

Synthetic data improves neural diarization training.

VBx remains effective with clustering approaches.

Abstract

Speaker diarization is usually referred to as the task that determines ``who spoke when'' in a recording. Until a few years ago, all competitive approaches were modular. Systems based on this framework reached state-of-the-art performance in most scenarios but had major difficulties dealing with overlapped speech. More recently, the advent of end-to-end models, capable of dealing with all aspects of speaker diarization with a single model and better performing regarding overlapped speech, has brought high levels of attention. This thesis is framed during a period of co-existence of these two trends. We describe a system based on a Bayesian hidden Markov model used to cluster x-vectors (speaker embeddings obtained with a neural network), known as VBx, which has shown remarkable performance on different datasets and challenges. We comment on its advantages and limitations and evaluate results on different relevant corpora. Then, we move towards end-to-end neural diarization (EEND) methods. Due to the need for large training sets for training these models and the lack of manually annotated diarization data in sufficient quantities, the compromise solution consists in generating training data artificially. We describe an approach for generating synthetic data which resembles real conversations in terms of speaker turns and overlaps. We show how this method generating ``simulated conversations'' allows for better performance than using a previously proposed method for creating ``simulated mixtures'' when training the popular EEND with encoder-decoder attractors (EEND-EDA). We also propose a new EEND-based model, which we call DiaPer, and show that it can perform better than EEND-EDA, especially when dealing with many speakers and handling overlapped speech. Finally, we compare both VBx-based and DiaPer systems on a wide variety of corpora and comment on the advantages of each technique.