Named Entity Recognition for Address Extraction in Speech-to-Text Transcriptions Using Synthetic Data

TL;DR

This paper presents a BERT-based NER model for extracting address components from speech-to-text transcriptions, trained solely on synthetic data generated via GPT, and evaluated on real data.

Contribution

It introduces a synthetic data generation approach for training NER models in speech transcription tasks, addressing data scarcity issues.

Findings

Model trained on synthetic data performs well on real test data

Synthetic data mimics spoken language variability effectively

Addresses address extraction in speech-to-text transcriptions

Abstract

This paper introduces an approach for building a Named Entity Recognition (NER) model built upon a Bidirectional Encoder Representations from Transformers (BERT) architecture, specifically utilizing the SlovakBERT model. This NER model extracts address parts from data acquired from speech-to-text transcriptions. Due to scarcity of real data, a synthetic dataset using GPT API was generated. The importance of mimicking spoken language variability in this artificial data is emphasized. The performance of our NER model, trained solely on synthetic data, is evaluated using small real test dataset.