Statistical language models (LM) play a key role in Automatic Speech Recognition (ASR) systems used by conversational agents. These ASR systems should provide a high accuracy under a variety of speaking styles, domains, vocabulary and argots. In this paper, we present a DNN-based method to adapt the LM to each user-agent interaction based on generalized contextual information, by predicting an optimal, context-dependent set of LM interpolation weights. We show that this framework for contextual adaptation provides accuracy improvements under different possible mixture LM partitions that are relevant for both (1) Goal-oriented conversational agents where it’s natural to partition the data by the requested application and for (2) Non-goal oriented conversational agents where the data can be partitioned using topic labels that come from predictions of a topic classifier. We obtain a relative WER reduction of 3% with a 1-pass decoding strategy and 6% in a 2-pass decoding framework, over an unadapted model. We also show up to a 15% relative WER reduction in recognizing named entities which is of significant value for conversational ASR systems.
DOI: 10.21437/Interspeech.2018-1122
Cite as: Raju, A., Hedayatnia, B., Liu, L., Gandhe, A., Khatri, C., Metallinou, A., Venkatesh, A., Rastrow, A. (2018) Contextual Language Model Adaptation for Conversational Agents. Proc. Interspeech 2018, 3333-3337, DOI: 10.21437/Interspeech.2018-1122.
@inproceedings{Raju2018,
author={Anirudh Raju and Behnam Hedayatnia and Linda Liu and Ankur Gandhe and Chandra Khatri and Angeliki Metallinou and Anu Venkatesh and Ariya Rastrow},
title={Contextual Language Model Adaptation for Conversational Agents},
year=2018,
booktitle={Proc. Interspeech 2018},
pages={3333--3337},
doi={10.21437/Interspeech.2018-1122},
url={http://dx.doi.org/10.21437/Interspeech.2018-1122}
}