Back from the future: bidirectional CTC decoding using future information in speech recognition
arXiv:2110.03326
Abstract
In this paper, we propose a simple but effective method to decode the output of Connectionist Temporal Classifier (CTC) model using a bi-directional neural language model. The bidirectional language model uses the future as well as the past information in order to predict the next output in the sequence. The proposed method based on bi-directional beam search takes advantage of the CTC greedy decoding output to represent the noisy future information. Experiments on the Librispeechdataset demonstrate the superiority of our proposed method compared to baselines using unidirectional decoding. In particular, the boost inaccuracy is most apparent at the start of a sequence which is the most erroneous part for existing systems based on unidirectional decoding.
arXiv commentsubmitted to ICASSP 2022
References in corpus (7)
- SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognition
- Deep Speech 2: End-to-End Speech Recognition in English and Mandarin
- Regularizing and Optimizing LSTM Language Models
- Conformer: Convolution-augmented Transformer for Speech Recognition
- Listen, Attend and Spell
- First-Pass Large Vocabulary Continuous Speech Recognition using Bi-Directional Recurrent DNNs
- Fully Convolutional Speech Recognition