Internal Language Model Training for Domain-Adaptive End-to-End Speech Recognition

TL;DR

This paper introduces an internal LM training method that enhances domain adaptation in end-to-end speech recognition by enabling better external language model integration without sacrificing accuracy.

Contribution

The proposed ILMT method trains E2E models to form a standalone internal LM, improving external LM integration and domain adaptation in speech recognition systems.

Findings

Achieved up to 31.5% WER reduction on LibriSpeech

Achieved up to 11.4% WER reduction on Microsoft production data

Enhanced external LM integration without degrading ASR accuracy

Abstract

The efficacy of external language model (LM) integration with existing end-to-end (E2E) automatic speech recognition (ASR) systems can be improved significantly using the internal language model estimation (ILME) method. In this method, the internal LM score is subtracted from the score obtained by interpolating the E2E score with the external LM score, during inference. To improve the ILME-based inference, we propose an internal LM training (ILMT) method to minimize an additional internal LM loss by updating only the E2E model components that affect the internal LM estimation. ILMT encourages the E2E model to form a standalone LM inside its existing components, without sacrificing ASR accuracy. After ILMT, the more modular E2E model with matched training and inference criteria enables a more thorough elimination of the source-domain internal LM, and therefore leads to a more effective integration of the target-domain external LM. Experimented with 30K-hour trained recurrent neural network transducer and attention-based encoder-decoder models, ILMT with ILME-based inference achieves up to 31.5% and 11.4% relative word error rate reductions from standard E2E training with Shallow Fusion on out-of-domain LibriSpeech and in-domain Microsoft production test sets, respectively.