BERT-LID: Leveraging BERT to Improve Spoken Language Identification

TL;DR

This paper introduces BERT-LID, a BERT-based system that uses phonetic posteriorgrams to significantly improve language identification accuracy, especially for short speech segments, enhancing multilingual speech system interoperability.

Contribution

It extends BERT with phonetic posteriorgrams for better short-utterance language identification, achieving notable accuracy improvements over baseline methods.

Findings

6.5% accuracy improvement on long segments

19.9% accuracy improvement on short segments

Effective enhancement of language ID performance for short speech

Abstract

Language identification is the task of automatically determining the identity of a language conveyed by a spoken segment. It has a profound impact on the multilingual interoperability of an intelligent speech system. Despite language identification attaining high accuracy on medium or long utterances(>3s), the performance on short utterances (<=1s) is still far from satisfactory. We propose a BERT-based language identification system (BERT-LID) to improve language identification performance, especially on short-duration speech segments. We extend the original BERT model by taking the phonetic posteriorgrams (PPG) derived from the front-end phone recognizer as input. Then we deployed the optimal deep classifier followed by it for language identification. Our BERT-LID model can improve the baseline accuracy by about 6.5% on long-segment identification and 19.9% on short-segment identification, demonstrating our BERT-LID's effectiveness to language identification.