Long-span language modeling for speech recognition
This work addresses improving speech recognition accuracy by leveraging long-span context, though it is incremental in combining existing methods.
The paper tackled the problem of language modeling for speech recognition across multiple sentences, introducing a new LSTM-attention hybrid architecture. Experiments on LibriSpeech showed significant perplexity reductions with paragraph-level training compared to sentence-level training.
We explore neural language modeling for speech recognition where the context spans multiple sentences. Rather than encode history beyond the current sentence using a cache of words or document-level features, we focus our study on the ability of LSTM and Transformer language models to implicitly learn to carry over context across sentence boundaries. We introduce a new architecture that incorporates an attention mechanism into LSTM to combine the benefits of recurrent and attention architectures. We conduct language modeling and speech recognition experiments on the publicly available LibriSpeech corpus. We show that conventional training on a paragraph-level corpus results in significant reductions in perplexity compared to training on a sentence-level corpus. We also describe speech recognition experiments using long-span language models in second-pass re-ranking, and provide insights into the ability of such models to take advantage of context beyond the current sentence.