Self-critical Sequence Training for Automatic Speech Recognition
This addresses performance degradation in ASR for speech recognition applications, offering a method to better align training objectives with evaluation metrics, though it is incremental as it builds on existing sequence-to-sequence models.
The paper tackled the mismatches between training and testing in automatic speech recognition by proposing self-critical sequence training, which uses reinforcement learning to align training with word error rate evaluation, resulting in relative WER improvements of 8.7% and 7.8% over baselines on clean and noisy datasets.
Although automatic speech recognition (ASR) task has gained remarkable success by sequence-to-sequence models, there are two main mismatches between its training and testing that might lead to performance degradation: 1) The typically used cross-entropy criterion aims to maximize log-likelihood of the training data, while the performance is evaluated by word error rate (WER), not log-likelihood; 2) The teacher-forcing method leads to the dependence on ground truth during training, which means that model has never been exposed to its own prediction before testing. In this paper, we propose an optimization method called self-critical sequence training (SCST) to make the training procedure much closer to the testing phase. As a reinforcement learning (RL) based method, SCST utilizes a customized reward function to associate the training criterion and WER. Furthermore, it removes the reliance on teacher-forcing and harmonizes the model with respect to its inference procedure. We conducted experiments on both clean and noisy speech datasets, and the results show that the proposed SCST respectively achieves 8.7% and 7.8% relative improvements over the baseline in terms of WER.