Reducing Exposure Bias in Training Recurrent Neural Network Transducers

Abstract

When recurrent neural network transducers (RNNTs) are trained using the typical maximum likelihood criterion, the prediction network is trained only on ground truth label sequences. This leads to a mismatch during inference, known as exposure bias, when the model must deal with label sequences containing errors. In this paper we investigate approaches to reducing exposure bias in training to improve the generalization of RNNT models for automatic speech recognition (ASR). A label-preserving input perturbation to the prediction network is introduced. The input token sequences are perturbed using SwitchOut and scheduled sampling based on an additional token language model. Experiments conducted on the 300-hour Switchboard dataset demonstrate their effectiveness. By reducing the exposure bias, we show that we can further improve the accuracy of a high-performance RNNT ASR model and obtain state-of-the-art results on the 300-hour Switchboard dataset.

arXiv commentaccepted to Interspeech 2021

References in corpus (6)

• Decoupled Weight Decay Regularization
• SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognition
• Attention-Based Models for Speech Recognition
• Scheduled Sampling for Sequence Prediction with Recurrent Neural Networks
• Sequence Transduction with Recurrent Neural Networks
• Improving Generalization of Transformer for Speech Recognition with Parallel Schedule Sampling and Relative Positional Embedding