Improving Mandarin End-to-End Speech Recognition with Word N-gram Language Model

TL;DR

This paper introduces a novel decoding algorithm for Mandarin E2E speech recognition that constructs word-level lattices on-the-fly, enabling effective integration of external word N-gram LMs and achieving state-of-the-art results.

Contribution

It proposes a new decoding method that constructs word-level lattices dynamically, allowing better use of external word-level language models in Mandarin ASR.

Findings

Outperforms subword-level LMs in experiments.

Achieves state-of-the-art CER on Aishell datasets.

Reduces CER by 14.8% on a large Mandarin dataset.

Abstract

Despite the rapid progress of end-to-end (E2E) automatic speech recognition (ASR), it has been shown that incorporating external language models (LMs) into the decoding can further improve the recognition performance of E2E ASR systems. To align with the modeling units adopted in E2E ASR systems, subword-level (e.g., characters, BPE) LMs are usually used to cooperate with current E2E ASR systems. However, the use of subword-level LMs will ignore the word-level information, which may limit the strength of the external LMs in E2E ASR. Although several methods have been proposed to incorporate word-level external LMs in E2E ASR, these methods are mainly designed for languages with clear word boundaries such as English and cannot be directly applied to languages like Mandarin, in which each character sequence can have multiple corresponding word sequences. To this end, we propose a novel decoding algorithm where a word-level lattice is constructed on-the-fly to consider all possible word sequences for each partial hypothesis. Then, the LM score of the hypothesis is obtained by intersecting the generated lattice with an external word N-gram LM. The proposed method is examined on both Attention-based Encoder-Decoder (AED) and Neural Transducer (NT) frameworks. Experiments suggest that our method consistently outperforms subword-level LMs, including N-gram LM and neural network LM. We achieve state-of-the-art results on both Aishell-1 (CER 4.18%) and Aishell-2 (CER 5.06%) datasets and reduce CER by 14.8% relatively on a 21K-hour Mandarin dataset.