Phrase-Level Class based Language Model for Mandarin Smart Speaker Query Recognition

TL;DR

This paper introduces a novel phrase-level class-based language model with Difference Language Model correction for Mandarin smart speaker query recognition, effectively handling large entity lists and dynamic updates with improved accuracy.

Contribution

It proposes pruned language models combined with DLM for efficient, accurate recognition of large and frequently updated entity lists in speech assistants.

Findings

Significant performance improvement over traditional methods.

Effective handling of large entity lists exceeding 20 million items.

Efficient model updating with minimal recalculations.

Abstract

The success of speech assistants requires precise recognition of a number of entities on particular contexts. A common solution is to train a class-based n-gram language model and then expand the classes into specific words or phrases. However, when the class has a huge list, e.g., more than 20 million songs, a fully expansion will cause memory explosion. Worse still, the list items in the class need to be updated frequently, which requires a dynamic model updating technique. In this work, we propose to train pruned language models for the word classes to replace the slots in the root n-gram. We further propose to use a novel technique, named Difference Language Model (DLM), to correct the bias from the pruned language models. Once the decoding graph is built, we only need to recalculate the DLM when the entities in word classes are updated. Results show that the proposed method consistently and significantly outperforms the conventional approaches on all datasets, esp. for large lists, which the conventional approaches cannot handle.