Spoken Term Detection and Relevance Score Estimation using Dot-Product of Pronunciation Embeddings

TL;DR

This paper introduces a deep LSTM-based method for spoken term detection that uses pronunciation embeddings and dot-product similarity to localize terms and estimate relevance scores directly from phoneme confusion networks.

Contribution

It extends previous Siamese network approaches by directly localizing spoken terms and estimating relevance scores using a novel embedding and dot-product approach.

Findings

Effective in English and Czech datasets

Achieves accurate localization of spoken terms

Outperforms previous STD methods

Abstract

The paper describes a novel approach to Spoken Term Detection (STD) in large spoken archives using deep LSTM networks. The work is based on the previous approach of using Siamese neural networks for STD and naturally extends it to directly localize a spoken term and estimate its relevance score. The phoneme confusion network generated by a phoneme recognizer is processed by the deep LSTM network which projects each segment of the confusion network into an embedding space. The searched term is projected into the same embedding space using another deep LSTM network. The relevance score is then computed using a simple dot-product in the embedding space and calibrated using a sigmoid function to predict the probability of occurrence. The location of the searched term is then estimated from the sequence of output probabilities. The deep LSTM networks are trained in a self-supervised manner from paired recognition hypotheses on word and phoneme levels. The method is experimentally evaluated on MALACH data in English and Czech languages.