Entropy-Based Dynamic Rescoring with Language Model in E2E ASR Systems

Language models (LM) have played crucial roles in automatic speech recognition (ASR), whether as an essential part of a conventional ASR system composed acoustic model and LM, or integrated to enhance the performance novel end-to-end systems. With development machine learning deep learning, language modeling has made great progress natural processing applications. In recent years, efforts been leverage advantages LM ASR. The most common way apply integration is still shallow fusion because it can be easily implemented by zero-overhead while obtaining significant improvement. Our method further applicability without hyperparameter tuning maintaining similar performance.