A Statistical Model for Word Discovery in Transcribed Speech

TL;DR

This paper introduces a statistical model and an unsupervised learning algorithm for segmenting continuous speech into words, demonstrating competitive performance through empirical testing.

Contribution

It presents a novel statistical model and an incremental unsupervised algorithm for word discovery in transcribed speech.

Findings

Algorithm is competitive with existing models

Effective for segmentation and word discovery

Empirical results validate the approach

Abstract

A statistical model for segmentation and word discovery in continuous speech is presented. An incremental unsupervised learning algorithm to infer word boundaries based on this model is described. Results of empirical tests showing that the algorithm is competitive with other models that have been used for similar tasks are also presented.