Scalable Multi Corpora Neural Language Models for ASR
This addresses practical deployment issues for ASR systems, though it is incremental as it builds on existing neural language model approaches.
The paper tackled challenges in deploying neural language models for large-scale automatic speech recognition, achieving a 6.2% relative reduction in word error rate with minimal latency increase in a second-pass rescoring framework.
Neural language models (NLM) have been shown to outperform conventional n-gram language models by a substantial margin in Automatic Speech Recognition (ASR) and other tasks. There are, however, a number of challenges that need to be addressed for an NLM to be used in a practical large-scale ASR system. In this paper, we present solutions to some of the challenges, including training NLM from heterogenous corpora, limiting latency impact and handling personalized bias in the second-pass rescorer. Overall, we show that we can achieve a 6.2% relative WER reduction using neural LM in a second-pass n-best rescoring framework with a minimal increase in latency.