Internal Language Model Estimation based Language Model Fusion for Cross-Domain Code-Switching Speech Recognition

TL;DR

This paper explores the application of Internal Language Model Estimation (ILME) based fusion to improve cross-domain code-switching speech recognition, demonstrating its effectiveness across different datasets and domain combinations.

Contribution

It extends ILME-based language model fusion to cross-domain code-switching speech recognition and evaluates its effectiveness with monolingual data merging.

Findings

ILME fusion improves recognition accuracy in cross-domain CSSR

Effective for intra-domain and cross-domain tasks

Works well with merged monolingual datasets

Abstract

Internal Language Model Estimation (ILME) based language model (LM) fusion has been shown significantly improved recognition results over conventional shallow fusion in both intra-domain and cross-domain speech recognition tasks. In this paper, we attempt to apply our ILME method to cross-domain code-switching speech recognition (CSSR) work. Specifically, our curiosity comes from several aspects. First, we are curious about how effective the ILME-based LM fusion is for both intra-domain and cross-domain CSSR tasks. We verify this with or without merging two code-switching domains. More importantly, we train an end-to-end (E2E) speech recognition model by means of merging two monolingual data sets and observe the efficacy of the proposed ILME-based LM fusion for CSSR. Experimental results on SEAME that is from Southeast Asian and another Chinese Mainland CS data set demonstrate the effectiveness of the proposed ILME-based LM fusion method.