As voice assistants become more ubiquitous, they are increasingly expected to support and perform well on a wide variety of use-cases across different domains. We present a domain-aware rescoring framework suitable for achieving domain-adaptation during second-pass rescoring in production settings. In our framework, we fine-tune a domain-general neural language model on several domains, and use an LSTM-based domain classification model to select the appropriate domain-adapted model to use for second-pass rescoring. This domain-aware rescoring improves the word error rate by up to 2.4% and slot word error rate by up to 4.1% on three individual domains -- shopping, navigation, and music -- compared to domain general rescoring. These improvements are obtained while maintaining accuracy for the general use case.
@article{arxiv.2101.03229,
title = {Domain-aware Neural Language Models for Speech Recognition},
author = {Linda Liu and Yile Gu and Aditya Gourav and Ankur Gandhe and Shashank Kalmane and Denis Filimonov and Ariya Rastrow and Ivan Bulyko},
journal= {arXiv preprint arXiv:2101.03229},
year = {2021}
}