Transformer ASR with Contextual Block Processing
arXiv:1910.07204
Abstract
The Transformer self-attention network has recently shown promising performance as an alternative to recurrent neural networks (RNNs) in end-to-end (E2E) automatic speech recognition (ASR) systems. However, the Transformer has a drawback in that the entire input sequence is required to compute self-attention. In this paper, we propose a new block processing method for the Transformer encoder by introducing a context-aware inheritance mechanism. An additional context embedding vector handed over from the previously processed block helps to encode not only local acoustic information but also global linguistic, channel, and speaker attributes. We introduce a novel mask technique to implement the context inheritance to train the model efficiently. Evaluations of the Wall Street Journal (WSJ), Librispeech, VoxForge Italian, and AISHELL-1 Mandarin speech recognition datasets show that our proposed contextual block processing method outperforms naive block processing consistently. Furthermore, the attention weight tendency of each layer is analyzed to clarify how the added contextual inheritance mechanism models the global information.
Accepted for ASRU 2019
References in corpus (9)
- Deep Speech 2: End-to-End Speech Recognition in English and Mandarin
- Conditional Image Generation with PixelCNN Decoders
- Attention-Based Models for Speech Recognition
- Sequence Transduction with Recurrent Neural Networks
- Generating Long Sequences with Sparse Transformers
- Improved training of end-to-end attention models for speech recognition
- Self-Attention Networks for Connectionist Temporal Classification in Speech Recognition
- A Neural Transducer
- Monotonic Chunkwise Attention