---
title: "Alignment Algorithms for Learning to Read Aloud"
authors:
  - "Charles X. Ling"
  - "Handong Wang"
doi: ""
url: "https://ijcai.org/proceedings/1997/126"
pdf: "https://ijcai.org/proceedings/1997/0126.pdf"
venue: "Proceedings of the Fifteenth International Joint Conference on Artificial Intelligence"
year: 1997
pages: "874-878"
track: "Learning"
keywords:
---

# Alignment Algorithms for Learning to Read Aloud

Charles X. Ling, Handong Wang

*Proceedings of the Fifteenth International Joint Conference on Artificial Intelligence*, pages 874-878.

## Abstract

A complete system of learning spelling-tophoneme conversion of English words consists of three major processes: alignment, mapping learning, and grapheme generation. Such a system can be used to construct prototypes of reading machines for English or other languages quickly and automatically. ~his. pap.er focusses on the alignment process, which IS cntical to mapping learning and grapheme generation. We present several novel alignment algorithms which learn alignment without supervision. The basic alignment algorithm is a hillclimbing algorithm. Several improvements of it are studied and tested. In addition, a method that overcomes the pitfall in hill-climbing algorithms is designed. Our best alignment algorithm produces very impressive results: only 0.5% of error rate.