End-to-End Diarization for Variable Number of Speakers with Local-Global Networks and Discriminative Speaker Embeddings

TL;DR

This paper introduces an end-to-end neural network for meeting diarization that handles variable numbers of speakers, overlaps, and discriminative training, showing improved performance on simulated and real-recorded meeting data.

Contribution

The paper proposes a novel end-to-end diarization model with local-global networks and discriminative speaker embeddings for variable speaker counts, outperforming previous models.

Findings

Outperforms previous end-to-end diarization models on LibriCSS data.

Effectively handles unknown number of speakers and overlaps.

Utilizes multi-task transfer learning and sequential refinement.

Abstract

We present an end-to-end deep network model that performs meeting diarization from single-channel audio recordings. End-to-end diarization models have the advantage of handling speaker overlap and enabling straightforward handling of discriminative training, unlike traditional clustering-based diarization methods. The proposed system is designed to handle meetings with unknown numbers of speakers, using variable-number permutation-invariant cross-entropy based loss functions. We introduce several components that appear to help with diarization performance, including a local convolutional network followed by a global self-attention module, multi-task transfer learning using a speaker identification component, and a sequential approach where the model is refined with a second stage. These are trained and validated on simulated meeting data based on LibriSpeech and LibriTTS datasets; final evaluations are done using LibriCSS, which consists of simulated meetings recorded using real acoustics via loudspeaker playback. The proposed model performs better than previously proposed end-to-end diarization models on these data.