Segmental Recurrent Neural Networks for End-to-end Speech Recognition
arXiv:1603.00223
Abstract
We study the segmental recurrent neural network for end-to-end acoustic modelling. This model connects the segmental conditional random field (CRF) with a recurrent neural network (RNN) used for feature extraction. Compared to most previous CRF-based acoustic models, it does not rely on an external system to provide features or segmentation boundaries. Instead, this model marginalises out all the possible segmentations, and features are extracted from the RNN trained together with the segmental CRF. In essence, this model is self-contained and can be trained end-to-end. In this paper, we discuss practical training and decoding issues as well as the method to speed up the training in the context of speech recognition. We performed experiments on the TIMIT dataset. We achieved 17.3 phone error rate (PER) from the first-pass decoding --- the best reported result using CRFs, despite the fact that we only used a zeroth-order CRF and without using any language model.
arXiv comment5 pages, 2 figures, accepted by Interspeech 2016
References in corpus (4)
Cited by in corpus (12)
- The History Began from AlexNet: A Comprehensive Survey on Deep Learning Approaches
- Frame-Semantic Parsing with Softmax-Margin Segmental RNNs and a Syntactic Scaffold
- Exploring Neural Transducers for End-to-End Speech Recognition
- Reducing Bias in Production Speech Models
- Multitask Learning with CTC and Segmental CRF for Speech Recognition
- A study of latent monotonic attention variants
- Morphological Inflection Generation with Hard Monotonic Attention
- Sequence Prediction with Neural Segmental Models
- Lexicon-Free Fingerspelling Recognition from Video: Data, Models, and Signer Adaptation
- Improving Human Text Comprehension through Semi-Markov CRF-based Neural Section Title Generation
- Phoneme Boundary Detection using Learnable Segmental Features
- Efficient Segmental Cascades for Speech Recognition