Semi-Supervised Discriminative Language Modeling with Out-of-Domain Text Data

One way to improve the accuracy of automatic speech recognition (ASR) is to use discriminative language modeling (DLM), which enhances discrimination by learning where the ASR hypotheses deviate from the uttered sentences. However, DLM requires large amounts of ASR output to train. Instead, we can simulate the output of an ASR system, in which case the training becomes semisupervised. The advantage of using simulated hypotheses is that we can generate as many hypotheses as we want provided that we have enough text material. In typical scenarios, transcribed in-domain data is limited but large amounts of out-of-domain (OOD) data is available. In this study, we investigate how semi-supervised training performs with OOD data. We find out that OOD data can yield improvements comparable to in-domain data.