Latent Words Recurrent Neural Network Language Models for Automatic Speech Recognition

Ryo MASUMURA  Taichi ASAMI  Takanobu OBA  Sumitaka SAKAUCHI  Akinori ITO  

Publication
IEICE TRANSACTIONS on Information and Systems   Vol.E102-D   No.12   pp.2557-2567
Publication Date: 2019/12/01
Publicized: 2019/09/25
Online ISSN: 1745-1361
DOI: 10.1587/transinf.2018EDP7242
Type of Manuscript: PAPER
Category: Speech and Hearing
Keyword: 
latent words recurrent neural network language models,  n-gram approximation,  Viterbi approximation,  automatic speech recognition,  

Full Text: PDF>>
Buy this Article




Summary: 
This paper demonstrates latent word recurrent neural network language models (LW-RNN-LMs) for enhancing automatic speech recognition (ASR). LW-RNN-LMs are constructed so as to pick up advantages in both recurrent neural network language models (RNN-LMs) and latent word language models (LW-LMs). The RNN-LMs can capture long-range context information and offer strong performance, and the LW-LMs are robust for out-of-domain tasks based on the latent word space modeling. However, the RNN-LMs cannot explicitly capture hidden relationships behind observed words since a concept of a latent variable space is not present. In addition, the LW-LMs cannot take into account long-range relationships between latent words. Our idea is to combine RNN-LM and LW-LM so as to compensate individual disadvantages. The LW-RNN-LMs can support both a latent variable space modeling as well as LW-LMs and a long-range relationship modeling as well as RNN-LMs at the same time. From the viewpoint of RNN-LMs, LW-RNN-LM can be considered as a soft class RNN-LM with a vast latent variable space. In contrast, from the viewpoint of LW-LMs, LW-RNN-LM can be considered as an LW-LM that uses the RNN structure for latent variable modeling instead of an n-gram structure. This paper also details a parameter inference method and two kinds of implementation methods, an n-gram approximation and a Viterbi approximation, for introducing the LW-LM to ASR. Our experiments show effectiveness of LW-RNN-LMs on a perplexity evaluation for the Penn Treebank corpus and an ASR evaluation for Japanese spontaneous speech tasks.