Jointly Learning to Align and Convert Graphemes to Phonemes with Neural Attention Models

TL;DR

This paper introduces an attention-based encoder-decoder model for grapheme-to-phoneme conversion that jointly learns alignment and conversion, achieving state-of-the-art results on multiple datasets.

Contribution

It presents a novel neural attention model that eliminates the need for explicit alignments in grapheme-to-phoneme conversion tasks.

Findings

Achieves state-of-the-art accuracy on CMUDict, Pronlex, and NetTalk datasets.

Demonstrates effectiveness of global and local attention mechanisms.

Outperforms previous joint sequence models.

Abstract

We propose an attention-enabled encoder-decoder model for the problem of grapheme-to-phoneme conversion. Most previous work has tackled the problem via joint sequence models that require explicit alignments for training. In contrast, the attention-enabled encoder-decoder model allows for jointly learning to align and convert characters to phonemes. We explore different types of attention models, including global and local attention, and our best models achieve state-of-the-art results on three standard data sets (CMUDict, Pronlex, and NetTalk).