Extending Recurrent Neural Aligner for Streaming End-to-End Speech Recognition in Mandarin

By Linhao Dong, Shiyu Zhou, Bo Xu. 2018

23 citations recorded by the source corpus.

End-to-end models have been showing superiority in Automatic Speech Recognition (ASR). At the same time, the capacity of streaming recognition has become a growing requirement for end-to-end models. Following these trends, an encoder-decoder recurrent neural network called Recurrent Neural Aligner (RNA) has been freshly proposed and shown its competitiveness on two English ASR tasks. However, it is not clear if RNA can be further improved and applied to other spoken language. In this work, we explore the applicability of RNA in Mandarin Chinese and present four effective extensions: In the encoder, we redesign the temporal down-sampling and introduce a powerful convolutional structure. In the decoder, we utilize a regularizer to smooth the output distribution and conduct joint training with a language model. On two Mandarin Chinese conversational telephone speech recognition (MTS) datasets, our Extended-RNA obtains promising performance. Particularly, it achieves 27.7% character error rate (CER), which is superior to current state-of-the-art result on the popular HKUST task.

Subfields: Artificial Intelligence, Speech & Audio, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Feng Wang
2. Bo Xu
3. Linhao Dong
4. Ne Luo
5. Wubo Li
6. Wei Zou
7. Dongwei Jiang
8. Xiangang Li

In Speech & Audio

1. Feng Wang
2. Ne Luo
3. Wubo Li
4. Xiangang Li
5. Wei Zou
6. Dongwei Jiang
7. Bo Xu
8. Linhao Dong

In Signal Processing

1. Feng Wang
2. Ne Luo
3. Wubo Li
4. Bo Xu
5. Linhao Dong
6. Wei Zou
7. Dongwei Jiang
8. Xiangang Li

Citations in the corpus

• Improving Transformer-based Speech Recognition Using Unsupervised Pre-training (2019, methodology): “For LSTM based models, we follow the model structure in paper [20], which consists of 4 unidirectional LSTM layers with 480 cells in the encoder network.”
• Self-attention Aligner: A Latency-control End-to-end Model for ASR Using Self-attention Network and Chunk-hopping (2019, methodology): “The RNN-LM also follows the setting in [11], utilizes 1-layer LSTM with 640 cells.”