End-to-End Multi-speaker ASR with Independent Vector Analysis

TL;DR

This paper introduces an end-to-end multi-speaker ASR system that integrates source separation and dereverberation using IVA, optimized jointly with ASR, and performs well across different noise conditions and speaker counts.

Contribution

It presents a novel IVA-based frontend integrated into an end-to-end ASR system, jointly optimized with the ASR loss, and capable of handling multiple speakers without retraining.

Findings

Competitive performance with neural beamforming systems.

Effective on noisy data trained only on clean mixtures.

Scales to three and four speakers without retraining.

Abstract

We develop an end-to-end system for multi-channel, multi-speaker automatic speech recognition. We propose a frontend for joint source separation and dereverberation based on the independent vector analysis (IVA) paradigm. It uses the fast and stable iterative source steering algorithm together with a neural source model. The parameters from the ASR module and the neural source model are optimized jointly from the ASR loss itself. We demonstrate competitive performance with previous systems using neural beamforming frontends. First, we explore the trade-offs when using various number of channels for training and testing. Second, we demonstrate that the proposed IVA frontend performs well on noisy data, even when trained on clean mixtures only. Furthermore, it extends without retraining to the separation of more speakers, which is demonstrated on mixtures of three and four speakers.