EVI: Multilingual Spoken Dialogue Tasks and Dataset for Knowledge-Based Enrolment, Verification, and Identification

TL;DR

This paper introduces EVI, a multilingual spoken dialogue dataset for knowledge-based enrolment, verification, and identification, along with formal task definitions and benchmark models to advance research in multilingual spoken dialogue systems.

Contribution

The paper formalizes three authentication tasks, provides a multilingual dataset, and establishes initial benchmarks for knowledge-based spoken dialogue authentication.

Findings

First competitive benchmarks for multilingual spoken dialogue authentication.

Challenges of multilingual natural language processing in spoken dialogue.

Directions for future research in multilingual dialogue systems.

Abstract

Knowledge-based authentication is crucial for task-oriented spoken dialogue systems that offer personalised and privacy-focused services. Such systems should be able to enrol (E), verify (V), and identify (I) new and recurring users based on their personal information, e.g. postcode, name, and date of birth. In this work, we formalise the three authentication tasks and their evaluation protocols, and we present EVI, a challenging spoken multilingual dataset with 5,506 dialogues in English, Polish, and French. Our proposed models set the first competitive benchmarks, explore the challenges of multilingual natural language processing of spoken dialogue, and set directions for future research.